From mildly interested reader: it sound highly complex and I am not entirely sure why someone would invest and how risk of not getting paid is for people participating in projects.
Are they paid and they work and investors risk money? How investors can earn extra money to offset risk?
I'm a bit baffled by the extended discussion of how an investor could somehow own the credit for the good work – you point this out briefly early on but then in a later section it seems to be taken as possible that a 100% equity holder could be treated as having done whatever the charitable thing itself is (beyond merely having funded it). Surely that kind of social consideration arises from interactions like interviews and accounts of how the work was actually done, such that nobody would seriously conflate the two. At best I can imagine perhaps someone wants to give the charity founder an award and they contractually have to ask that it goes to the funder instead, but I think the audience at that awards ceremony would likely find it strange and awkward.
Sorry for being late to the party, but I fail to understand the central argument: why would anyone bother to buy the credit? Even without the premium this would hardly make any common sense.
In „normal“ markets patents, companies or brands are bought with an expectation to gain more profits with these later. It makes sense as an investment.
Products and services are bought as a means to an end. It makes sense as a purchase (or as an investment, too).
Buying the credit after the fact seems to me to make sense neither as an investment nor as a purchase. Just „bragging rights“?!
Maybe you addressed this point and I simply missed it, but otherwise it seems as if you're glossing over what looks to me like as the most difficult problem with this idea: how does an "oracular funder" evaluates the outcome of a project, and assigns it a specific dollar value?
Without a well-designed, aligned, transparent and fair mechanism to produce such evaluations. everything falls apart.
I don't get the problem with the capitalistic option, if I'm a normal employee at a for profit company or a charity and I get paid the market rate, I still kinda get "the credit" for being a good worker, right? So why can't I get the credit for being a good founder while also getting the market rate money for it?
The incentivizing evil issue kinda broke the beauty of the idea for me... could it be solved somehow with fines? Like say the oracle funder requires that you publicly buy the shares. Then if the project turns out to do harm, they tell you to pay to offset this and you do it because...
1) they'll refuse paying you for other project shares and harm your career as a charity funder?
2) all oracle funders cooperate on 1) and super ruin your career as a charity funder?
3) other entities cooperate to not cooperate with you and be mean to you unless you pay the fine?
4) it turns into a dystopian court system where the oracles can randomly ruin your life?
Super interesting. Re "whoever snaps up the shares first will get most of the surplus", two possible solutions
1) have an IPO window, and if it's oversubscribed, allocate by lottery. This has the advantage of being easy and the disadvantage that you under reward people for spotting the opportunity (though I think that's ok - when it's oversubscribed, there wasn't much value in spotting the opportunity)
2) have an IPO window with a more complex allocation eg if it's oversubscribed by a factor X, everyone gets 1/X of what they ask for. When the UK government sold the Royal Mail to retail investors about 10 years ago, I think everyone who applied got what they wanted unless you applied for over £10k worth in which case you got that amount
Let me know when you've worked it out - happy to invest my money in impactful ways, but have been around funding long enough to be impatient and sceptical of new/innovative impact models & evaluation frameworks.
Regarding the auctioning off of the tokens: I think this could be done with a Liquidity Bootstrapping Pool [0] [1]. You would have to first sell some tokens personally (over the counter) and then you put the proceeds from that and the remaining tokens into the balancer pool, which then uses an automatic market maker mechanism to sell tokens at a fair market price until a previously set percentage of tokens has been sold.
Optimistically, it feels like a lot of these concerns will ultimately be circumvented by the market. In theory (ideal end-state, as you call it), you don't have to worry about who "should" be allowed to reap financial benefits, for example, because funders and project runners will simply decide for themselves which kind of operation they want.
It seems like a pilot could be run on KickStarter right now, to be honest, and would provide more concrete data for future investors/charities to reference.
I originally parsed "We don’t get the “coolness” boost of using crypto" as an obvious joke, but the following paragraphs appeared to suggest it was intended unironically. It's fascinating that bubbles exist even today where "NFT" and "crypto" signal the precise opposite of everything they do in my techy bubble.
Does the final oracular funder announce in advance what outcomes they're willing to fund, eg. we'll definitely pay $2,000 per QALY? Otherwise, the funder has a perverse incentive to only pay out in a fraction of cases (whenever they can find a flimsy justification) as by that point the good thing has already been done, so it's in their interest to keep their money as an incentive to further projects. This seems doubly so if these are being done for profit.
These also have the potential to become Utilitarian Indulgences - companies that want to rely on child labour or pollute a few rivers can buy these as a moral carbon offset.
>This post isn’t about the theory. It’s about the annoying implementation details. It may not be very interesting to people who are neither effective altruists nor institution design wonks, sorry.
I actually enjoyed this, for whatever it's worth, despite having no particular interest in EA (too poor) or institutional design. Any chapter of The Chronicles of Moloch is gonna be interesting reading; coordination problems seem like the fundamental difficulty of human history.
Also, now I wanna see a noir story about retrospective assassination funding...perhaps call it "Killer Prophets".
a) you end up with 5 million spend on a project that costs 1 million. Given that money for charitable work & great EA projects is limited, why is this not a major disadvantage?
All the other questions on how the 'extra' money gets distributed, result from this.
b) publication of project ideas or: how do make sure that ideas are not stolen or (deliberately or unconsciously) copied?
The *founder* has two functions: having a good idea/ designing a project (*invention*) and executing the project (*implementation*). Imagine my collegue has a brilliant, but - in retrospect - very simply idea on how to make projects 'raising the sanity waterline' more efficient. He designs such a project and wants to implement it, but he has only so much experience, and investors are hesitant to invest. CFAR and everybody else interested in 'raising the sanity waterline' read the idea, set up a project based on this, back it up by all their experience and clout in the field, and get the money. Good for the impact, but the inventor will probably know this, and not put out their idea. Again, with *very* specific things, you can have a norm of 'respect that person x wrote that first'. With a lot of great project ideas (which are often some type of improvements of existing things), it will be very difficult or close to impossible to distinguish between what was 'stolen' and what was 'we just thought about that obviously'. (Even more, people will suddently have this great idea when thinking about a project they are planning, and be totally unaware that they actually *read* this in a slightly different context some months ago.)
Solution: you need to make sure, that ideas / project designs are only available to investors but not to other potential founders.
Disadvantage: I have no idea how to make this work without contradicting other intended mechanisms.
Other solution: you need to reward *inventors* for their great project design, even if they're not getting the investors' money for implementation. Don't know yet how to do this in the context of this design.
Question: How does this work in VC? You don't put your business idea out on the www to see if it attracts capital.
c) I understand that you want all the elegance and mechanisms of a real market with lots of options. Looking at the difficulties and disadvantages, there could be intermediate solutions. EG: Founders put out their project to potential final oracular funders (for the moment a limited number of players known beforehand). A funder says: I find the idea to cure malaria in country x with the planned amount of 1 million USD great, if it works, I would be willing to pay 3 million for this. Now, all the projects that have a backing by a funder, go to the potential investors. Those project that find the necessary amount of investment, get funded.
d) The measurement. How do final oracular funders know how well the project worked out? There is *lots* of incentive to distort this in the reports. (When this whole thing grows bigger; hopefully less so in your next rounds of grants.)
Related: Most projects I know of don't fail on the account of the few measurable indicators they commit to. They fail on everything else. You mentioned this, but I think it's underaccounted for in the reflections on the set-up. Maybe EA has long solved this, and I'm unaware.
e) re 2. In the charity/grants/projects worlds I know, funders would take all the credit for *funding* the greatest projects, and organizations/firms/... would take all the credit for *implementing* the greatest project. (In fact for *designing* and *implementing*.) Makes sense to me, because we're talking about two different tasks.
When the EU funds a project, the 'EU funded' is plastered on each infrastructure project built, each tractor bought or each scientific article published. The firm that built the infrastructure will claim 100% of credit for carrying out this magnificent (EU-funded) building or scientists x for doing that magnificient research. For both it's important.
I guess this goes in direction of 2B (without having read Ben H. article). I find the 'sells all the credit' or 'sells part of the credit' is unnessarily complicating things. I think those latter might be unproblematic or even attractive in certain circles (VC/rat.adj./ EA? ...), but I would be worried it might have a chilling effect when selling the method to many other groups of people, including those usually engaged in non-profit projects.
Admitted, I 'cured malaria' sounds better than 'I funded that malaria was cured' – but I think it still sounds pretty good ;). You could partly solve this by selling 'cure malaria (funding)' certificates and when you're talking about this as 'I cured malaria', everybody at the party would know you're the funder, not the one implementing this. Or vice versa.
I am skeptical about this working - but one useful part for prediction markets or similar would be ability to predict that attempt X will fail (or succeed).
This may provide some help against blatant scams and failures.
Though overall, this sounds like an interesting project with many traps, up to SEC appearing and ending fun or maybe even criminal case for someone (is there prediction market for that already? :) ).
On issue Number 9B - my experience of the UK pharma regulatory context is that when there's enough money on the line people can get quite argumentative about exactly how their intervention is assessed. Quite a lot of this is that it is really hard - probably impossible - to say "This intervention saved X lives for sure" because it will always involve more-or-less reasonable assumptions (for example that, counterfactually, everyone wouldn't have spontaneously killed themselves as a protest against non-intervention, for example. Or maybe more reasonably, what the correct statistical approach is to participants in the trial who deblind themselves somehow or cross over trial arms). So you end up saying, "Our best guess is that we saved Y lives" and then defending your decision from criticism from the people with a large financial stake in Y being as high as possible.
The rough cost of the regulators defending one pharmaceutical submission in the UK is £143,000. This is an overestimate of how much the oracle would spend to do exactly the same thing as NICE, because large government agencies are inefficient, but this is also an underestimate of the true cost because the pharma companies do a lot of the work for the regulator; for example the company always funds the literature review and development of an economic model which are close to £100k each. My guess is that maybe the total cost of a full assessment including salaries is probably fairly close to £500,000(ish). Also it is unclear to me who funds the trial to assess effectiveness, but this could be hundreds of millions of pounds if you want to RCT a big intervention (and if you don't want to RCT the intervention you introduce a lot more points to argue about).
Basically, I'm a bit unclear how the 'guard labour' costs of the oracular funder are accounted for. These costs are non-trivial and it probably wouldn't be suitable to use the current EA method of well-meaning people giving it their best shot to assess outcomes even if there are enough well-meaning EAs around to make this possible, because if the project works well there will be a whole bunch of monied interests attacking your decisions.
I don't see any problems with founders getting rich. If that's a persistent problem, more smart people will flock to the charitable sector. Thus driving down excess returns for founders and also causing more charity to happen.
Why try to make philanthropy more efficient? Philanthropy is on net socially detrimental. Donors are at the heart of the culture way (e.g. George Soros). They are not accountable to the people that they are allegedly helping. I will elaborate on my substack in a few days.
I don't really understand why you have a problem with giving founders a piece of the impact shares? If you're building a system where you pay $5 million after the fact to a project that only needed $1 million to succeed, you're already paying $4 million dollars more than usual, solely for the purpose of promoting risk taking and innovation. Whether that $4 million extra goes primarily towards the founder or the funder seems entirely like a question of who needs to be incentivized more. While it does seem to me that funders need to be incentivized more, it also seems like it would be beneficial if charity founders were properly incentivized for success, such that they actually get a significant personal benefit iff the project succeeds.
Also a minor note on crypto, there are many obvious disadvantages to working in it, but one true advantage is that Crypto is great at dealing with international economics, especially when dealing with something as legally complicated as securities, or when dealing with third world countries.
I got lost at this point: "Foundations see that I have done a great thing with $5 million in benefits, so they give me $5 million."
How does that work exactly? Even just firming up the $5m claimed value seems like a process that is much easier said than done. Why does the foundation then just give that money?
Very interesting, but contains too many assumptions, too many details, and is too long for me to do more than scan. That said, here’s one potential problem and solution.
If I’m the final funder (oracular?), I might not put any money into this! I mean, hey, the good results I want already exist! Maybe I’ll throw them a token, but better to save my money for something else.
So I suggest “results puts”. The initial investors come up with metrics (or use metrics from the founders), and sell results puts, such that anyone who buys a put must pay if the metric is met. Then they conditionally buy certificates from the founders, and the sale executes if they are able to sell the put.
1) A project selling off 100% of the credit removes incentives for the project to go well (in some idealized sense, anyway). That’s why VCs don’t buy 100% of a startup’s equity.
2) Auctions seem to solve the “fast people win” problem reasonably well. Everyone puts in orders like “I want to buy X amount and Y price”. Then after some time, you resolve the auction by picking a price, perhaps the highest price so that at least Z dollars of funding will be raised. Everyone who offered to buy at >= that price gets shares at that price. (This is roughly how “real” auctions in financial markets work, except that in that context there are both buyers and sellers and the price is chosen to maximize the number of shares traded).
One thing I am not clear on /don't think I saw addressed: What happens when an additional participant (s) tries to muscle into the same space?
Oracular Funder pledges to buy 5m worth of Malaria being cured.
Alice sells 1m in Malaria certificates and goes off to start curing Malaria. Bob thinks he has a better solution to cure Malaria and tries to raise 1m to go cure Malaria his way.
If Bob also gets funded, he and Alice can both work simultaneously. This doubled the vc funding for malaria, but may half the potential return, even for Alice's investors who are already in. If it doesn't half the potential return, then who gets the credit? Both Alice and Bob were actively working to cure Malaria, and it has been cured-seems very challenging to figure out which of the two get credit.
Beyond that, what if Bob sees this coming, and decides not to bother because Alice has already started a project aimed at Malaria, and there's no more funding to solve Malaria EVEN THOUGH ALICE HAS NOT YET ACTUALLY DONE ANYTHING. He then goes to work at a hedge fund instead.
Isn't there a risk that when oracular funders have presold a token (or any other form of pre-registration) saying "the owner of this token gets credit for Malaria being cured" there is less incentive for later arrivals to bother trying to solve Malaria?
I am one of those random Google engineers you mention. This idea sounded extremely cool when I read the first half of the post on the train on the way to dinner. I even spent some time enthusiastically telling my wife about it while we ate. But then I read the rest of the post on the train back, and in particular got to point 6.
If this project involves crypto in any way, it immediately becomes anathema to me. In fact, I am willing to pledge that I will invest $10K+ of money in this sort of market within six months of its launch, if and only if it does not involve crypto. I know this sum is not a serious consideration when weighed against all the potential money coming from crypto people, but I write it down to emphasize that this is not empty virtue signaling, and is something I believe strongly enough to put real stakes behind.
Crypto is too much the domain of scammers and speculators. It is a solution very much in search of a problem, and sacrificing this interesting idea on the altar of trying to give it another shot at adding value to the world is ... unfortunate. Any association of this project with crypto brings the project into the realm of the sorts I do not want to associate with, or give any financial backing or societal legitimacy to.
I don't think this opinion of crypto is very tribal. It seems shared by the majority of my fellow gray-tribe software engineers, at least the ones that I know in real life. I think you might want to gather some more data before assuming that crypto = cool and therefore that grafting it onto this project will be a net positive. E.g. NFTs and NFT-holders are actively mocked by my peer group, so if this project got a whiff of NFT-ness near it then I would be incentivized to avoid participating in, or even discussing, impact markets just to avoid the social disgrace of being seen as an NFT-head.
(Secondarily, I don't know too much about the environmental concerns, but my suspicion is that you're blowing past them too easily---to my outsider knowledge, the only serious attempt to avoid the environmental impacts of current blockchains was Chia, which immediately had other negative externalities on the hard drive market.)
I will admit that if this project turns out to use crypto and nevertheless be wildly successful at impacting the world positively, then after a few years I will hold my nose and participate eventually. But if you want the participation of people like myself during the early bootstrapping stages, your choice of whether to align yourself with the crypto ecosystem or not will be important.
1. Sometimes it's really hard to measure impact even after the fact. There is still controversy over whether the New Deal shortened or prolonged the great depression. There is still controversy over what caused the rise in US homicides in the 60s/70s and what caused the decline in the 90s. But at least determining impact after the fact is easier than predicting it.
2. Many investors will be scared away by excessive uncertainty around the FOF payout unless you can get the FOF to contractually precommit "I will pay $X to whoever causes outcome Y"
3. There is a risk of people getting paid for placebo interventions when Y was going to happen anyway.
4. The “Maximum Capitalism” Design may not be that bad because competition between founders to do the same thing incentivizes them to split the surplus with others. If Alice was going to sell 50% equity for 1 million, Bob can undercut her by selling 75% equity for 1 million. In a competitive market for this the founder's surplus will tend to converge on their actual value added.
7 seems easy to solve. The founder submits the amount of cash they need. The market organizers create and auction off the shares. The market organizers give the founder the amount of cash they need, keeping the excess. The excess is used to subsidize the final oracular funders for the entire market.
So imagine 4 projects:
Project 1: Costs $1, IPO worth $2, Final value is $2
Project 2: Costs $1, IPO worth $1, Final value is $5
Project 3: Costs $1, IPO worth $3, Final value is $1
The market auctions off all three projects, using $3 (1 + 1 + 1) to fund the projects and gaining $5 (2 + 1 + 3) as excess. Much later, the final oracles would like to use $8 (2 + 5 + 1) compensating all the projects for their charity, but by spending from the excess they only need to spend $3.
This diverts the riches of "cheap projects" from both the founders and the investors towards funding more charity.
AISafety.com is currently investigating an unlikely hypothesis about the ability of language models to reason strategically at higher meta-levels. This research may be well suited for being funded through impact certificates:
1) The plausibility of the hypothesis depends on a substantial amount of illegible reasoning.
2) The hypothesis is unlikely to be true, but if it is true the impact could be great and legible.
3) We are currently exclusively funded by my personal savings, but we could move faster with more funding.
--Søren Elverlin (I have been running the AISafety.com Reading Group since 2016)
On #7, there's a third alternative: bid on the amount of the prize you'll rebate back to the oracular funder. In the common knowledge $1 million / $5 million example, the auction should clear at something a little bit less than an 80% rebate and the investors would get around a risk-free rate of return on their funds. If the auction can't clear at even a 0% rebate, then the assurance contingency isn't met.
When running the auction, investors could submit different bids at different rebate levels. I might bid the full $1 million at a 0% rebate but a smaller amount at a 75% rebate.
You can't give the founders money and call this a non-profit, or without embezzlement issues. A foundation gives your company an award for charity and you put it in your personal account? You sell stakes in the future non-charity-restricted income that your "non-profit" might have, and that's not stock?
Here I have a great idea for a non-profit charitable market, I will say that I have $1b of funding available to fund a next-generation fighter jet and let the free market decide on proposals and implementations of the fighter jet. Then I will pick what I judge to be "the most impactful" jet, and distribute the $1b accordingly to their founders and capital supporters. It's not for profit though because, like, it's a fighter jet hello that's charity?
In addition to VC, I think there will soon be more traditional funding available for this type of thing. ESG (environmental, social, governance) is the hot new thing on Wall Street and it doesn’t look like it’s going away. Investment funds and most companies are playing along. The SEC recently proposed a rule that investment firms and investment managers must provide ESG guidance.
ESG is also 90% fraudulent and the ratings are based on ridiculous metrics, if not purchased outright. Being able to purchase the credit for charitable actions will be like candy for those who need a bump in scores.
This sets up competition between charities in a way we currently don't have and which will need addressing.
Under a prospective model, if two charities independently set out to eliminate malaria in Senegal, and five years later it's eliminated, they rejoice in their shared victory. *Maybe* some of the funders grumble about getting less-than-expected bang-for-buck, though no single funder is certain they did (maybe the funders of the *other* project wasted their money -- we'll never know!).
But with an impact market, a retrospective funder whose promised $5M now has to decide whom to give it to. Each set of investors considers themselves entitled to the whole $5M ("We said we'd eliminate malaria and it's eliminated -- what more could you want?").
For-profit investing also worries about competition, but in that case it tends to be pretty easy to tell who each customer bought from. Capable firms shadow the fields they work in, preventing new growth where it isn't needed, but loud-mouth firms don't shadow fields in which they can't deliver value.
The whole idea of selling "impact" itself seems both complicated and philosophically dubious. It seems to me it would be much better just to set up one system where altruists can promise to pay out prizes to whoever accomplishes some worthwhile goal, and then a market where groups working toward the goal can sell future shares of the prize money. For example, altruists publish that they'll pay a $5 million prize to whoever cures malaria in Senegal. If a group of biologists think they can cure it for $2.5 million, they can sell $2.5 million tokens for $1 each that pay out $2 each after they collect the prize. There's some issue with determining the exact conditions when the prizes pay out, but the prize pools just need to have arbiters who cultivate a reputation for fairness. That's no worse problem than exists for future oracular funders.
For any "startup" like idea, I think you either need to:
- Talk to the potential users ("investors", "founders", ...), do the mom test
- Just try stuff
I really appreciate how in depth your analysis is, but you'll never be able to go through the nitty gritty by thinking about it. Your grants program was a great example of this: many unforeseen things came up.
I'm also very excited to see where this goes. Please keep us posted!
(apologies if a comment like this ahas already been posted, I browsed through a few of the comments but didn't see anything like this)
I am a tax/estate planning attorney. Exciting topic. I wanted to raise two thoughts—enhancing impact certificates with legislative change and addressing some of the issues from your implementations. No legal advice here, just wanted to share some ideas.
*****Impact Certificates with Legislative Change?*****
First, I note your (Scott's) solutions generally assume no legislative change. But imagine a legal system where impact certificates replace our current charitable giving structure, and can lead to broad government decentralization. Here's an example:
- Like the USPTO, tax-exempt entities apply and get approved a “project patent” from a government agency. The design of this agency, how its examiners are paid, legal doctrines around the project patent (and what needs to be shown to get one), methods (and incentives) for others to challenge an issued patent or require its update, ability of other parties to use an already-issued patent, and a designated project patent court are significant in getting this to work.
- On successful hitting of KPIs specified in the patent, the tax-exempts receive a defined amount of government cash and an assignable tax deduction. Payouts are for much less than 100% of benefit; for projects where benefits were tough to quantify, cost savings over government or other approaches for like benefits could be a relevant metric. Project patents could continue indefinitely, if they are so written (subject to the need for updating).
- The tax-exempt projects then repay (via a debt or equity instrument) funders with some combination of the government cash and deductions (which can be assigned to the investors). The deductions received by investors are forced to be carried into future years if the investor does not meet certain requirements. Such requirements would essentially aim to prevent rich people from zeroing out their taxes using these investments, unless their investments really do represent a replacement for the whole of government. This would include some limit to the deduction tied to the government’s need for taxes to service its debt.
There really is a lot of government that could be rerouted to this system, not just charity. Also, I thought the design could be part of a grand bargain to help with raising taxes. (IMO, the best objection to raising taxes for greater government services in the US is our generally poor governmental performance.) Raised taxes, of course, increase the attractiveness of the deductions generated.
*****Legal and Other Design Issues*****
Second, to address some legal and other design issues raised by your post.
- Have you reviewed the literature on (social) impact bonds? They already have designs aiming to address some of the issues you raise, including quantifying the return. I assume so, but thought I would note it.
- I thought it was interesting “fractionalized impact shares with assurance contract” essentially follows the capital contribution/capital call structure already present in investment funds, except it’s the project itself doing the offering.
- There are regulations for crowdfunding which can somewhat sidestep the accredited investor rules. In general, they're outside my knowledge, but I understand they’re perceived as a bit too burdensome on companies.
- People wanting tax deductions instead of investment returns (or who are not accredited investors but just want to help) might give to certain §501(c)(3) entities. The §501(c)(3)s might then invest into the oracular funder's…funds (set up as a partnership in a standard VC-type structure). Via one legal structure or another, the funds receive a small service/guarantor fee from each project it investigates and agrees to underwrite the successes of. One could consider whether the §501(c)(3) investments into the funder's funds are program-related investments (essentially, investments where a significant part of the “return” is achieving charitable goals, which are recognized under the tax law).
- There are indeed private benefit/§501(c)(3) concerns for investing in an oracular funder (and inurement or various excise taxes, if the §501(c)(3) insiders are involved with a funded project). The private benefit angle appears well-explored in regards to social impact bonds in http://fordhamlawreview.org/wp-content/uploads/assets/pdfs/Vol_81/Dagher_May.pdf. My initial view, not having run it to ground at all, is that §501(c)(3)s investing in this area should generally be segregated from other §501(c)(3)s to avoid putting more charitable tax benefits at risk than is necessary. Coming up with a good position here could take some thought.
- I am not sure I understand your concerns with governance issues, at least to the extent you're indicating there are special problems here—if the impact certificates are, e.g., interests in an LLC, it would be easy to build in state governance rights that already exist, and to model it on corporate law. Further, requiring periodic disclosure would be easy to require in the LLC agreement. No new law needed.
- There are legal rules governing payment processors, which is maybe your vision for an impact marketplace in part? Not sure I understand all the functions you see the marketplace as performing. Tax-wise, payment processors are often considered agents, so they only realize their fees as income; some designs could cause the payment to be income to the marketplace, and the payout to be a deduction. There would be tax information reporting issues, though (think 1099s).
Thanks for this incredibly comprehensive overview!
I’d like to use this chance to position our GoodX project “Impact Markets” (https://impactmarkets.io/) in this context. (Thanks for linking to our MVP post!) I’m trying to write this in a way that is also understandable to people who haven’t thought about impact markets much, so sorry in advance if some of it is unnecessarily verbose.
Very long time lurker finally brought out from the woodwork - this is basically social impact bonds with all their current issues, with extra twists!
There is an additional subtle Goodharting issue going on in SIBs, beyond the mechanism design you propose - projects will be shifted to issues that have a clear evaluation strategy. For example, if you fund projects reducing homelessness, how will the funder know it was the impact of the charity, instead of a change in unobserved city policies, that led to the result you see?
Let's assume that the funder requires an RCT or a proper impact evaluation. In first place, this rules out any setting where a clear evaluation is difficult, such as large scale projects or settings with general equilibrium issues. Assuming that this is avoided and projects stick to easily evaluated outcomes with clear identification strategies, now everyone has incentives to pull the crappiest study that supports the result they need to get paid, or take every single potshot at the funder's evaluation. This enhanced data interrogation technique is already bad with charities under normal funding schemes, btw - now bringing in capital is going to raise the stakes immensely. Imagine that the outcome of your token/bond/share depends on whether someone clustered their standard errors correctly - will there be a court ruling determining that randomization inference is better than clustering, and ordering someone to pay up based on that?
Personal anecdote - I was once an RA involved in writing a grant for the evaluation of a social impact bond in Colombia. The program was basically impossible to evaluate properly, for starters, but we came up with a decent instrumental variable that would allow us to at least calculate the effect size bounds. The funder did not like that, because they would look bad if they funded anything that ended up having zero impact, or even looked ambiguous. They eventually went with a before/after design (not even DiD!) that would be laughed out of any intro to econometrics class.
I think this is a very interesting problem, one where you need to adapt old and known mechanisms - markets - to a new purpose. It is important to keep the good elements of the original mechanism and, I think, you are dropping some of them.
2. It is not clear why monetary mechanism design has to correlate with "credit". They guy who invents the cure for malaria gets the glory and respect for the invention; they guy who funded gets his money and, possibly, some glory and respect for funding it. Money and glory are complementary benefits here, there is no reason to try to design a mechanism to ensure greater correlation of money and glory. Since different people put different relative value on money vs glory, it is better for everyone to give more glory to those who value it most and more money to those who value that most.
You note that although the inventor sold 100% of monetary benefits to the funder, he still might derive some monetary benefits (e.g. Harvard tenure) from his glory. This is great, but why does it follow that the inventor should retain more equity and get a higher monetary payoff? I would argue that the opposite should be true: since the inventor gets some monetary benefits as derivatives of his glory benefits, marginal utility of additional monetary benefits to the inventor is lower, so retaining equity makes less sense than in a world where the inventor is unlikely to get Harvard tenure.
3. "I am nervous about this because I want to be able to pitch impact certificates to non-savvy people who may not be able to make good decisions about equity." In financial markets this is called "misselling". It is often a criminal offence. Do not do that. The whole point of these markets is to mobilise insights from people who have the incentive to understand the problem. People who do not understand funding instruments are not going to have very good insights because they do not understand their own incentives.
4 (and also 7) "If we don’t let founders get rich, it shifts the surplus to fast investors who may not have added any value - that is, to the first people who snapped up the obviously-underpriced certificates." This is not necessarily true. The primary placement of certificates may be done the way bonds are placed through an auction: bidders submits bids with a quantity and the smallest acceptable success contingent yield, the final oracular funder (FOF) looks at the threshold yield Y which is the lowest yield at which they can raise the required amount, and then all bids with yields below Y get certificates with the yield Y. So in a reasonably liquid market for an obviously successfull project, Y will be close to the risk free rate for the term and all benefits accrue to the final oracular funder (FOF).
An example: let's say the final oracle funder wants people to give $1m for malaria cure now. They announce the auction, and there are many potential investors, from very skeptical/greedy ones who are ready to fund £100k at yield of 900% over 5 years (x10) to very confident/altruistic ones, ready to fund $200k at yield of 0%. The order book might look like this:
bid amount yield
A $100k 900%
B $700k 400%
C $300k 250%
D $400k 100%
E $600k 20%
F $200k 0%
FOF starts from the bottom of the book. For $1m to raise, the threshold yield Y is 100%. Bids E and F are filled in full, bid D is filled 50%, these three bidders get certificates with success contingent yield of 100%, other bids are unfilled. The funder commits to pay a return of 100% on $1m, i.e. they will pay $2m in case of project success. If the threshold yield is too high for the final oracle, the auction fails.
5B. The same yield auction design resolves the issues with committed pot of money. The primary issuance is done through an auction where the success contingent yields are fixed. An investor has the option to hold the certificate to maturity and receive the fixed yield. If there are other auctions at later dates, the participants there compete for other tokens and do not affect the hold to maturity yield of earlier issues.
6B. The coolness halo of crypto is dissipating quickly. We already see NFT-like projects carefully avoiding the word "NFT" in their materials (reddit avatar.) There may be genuine benefits of using e.g. a public ledger (some glory for the funders?) or other elements of crypto/blockchains.
9. In case of auction design as above, this question has a simple answer. FOF decides how big prize are they ready to give out in case of success, keep this information secret, and run the auction to raise a public required amount ($1m). If the threshold yield is too high and implies a higher prize than the FOF is ready to give, the auction fails as there are not enough optimistic investors. For non-binary outcomes, certificate format can be adjusted to e.g. pay off $1 per life saved capped at $2m and a similar price auction can be used for primary placement.
Appendix II A. This is what insurance is for. You will need to raise $1.2m and pay $200k for an insurance from the bad outcome. The insurance may be a regulatory requirement in any case for many projects. The insurer is likely to impose some controls on the developer.
Are final oracular funders actually required to be final, or are certificates perpetually transferrable like stocks?
Some consequences of allowing resale:
* Investors could buy underpriced certificates for completed interventions. E.g. say the malaria-curing project is complete and its certificates have been sold for $5M based on the lives it saved. But you find evidence it actually had another $5M in indirect economic benefits. You could buy the shares , publish your evidence, and sell for a profit.
* Individual funders who ran short on money—or just have a change of heart and decide to be selfish after all—could sell off their certificate portfolio. This is kind of weird to think about, but it might lead to a lot more investment in impact certificates.
Ok maybe I'm missing something very basic here, but when Scott says some charitable intervention will produce $5 million in benefits what exactly is that supposed to mean? Does it just mean that some rich person/charity will pay $5 million if the charitable result is achieved?
This circumstance is far more common than the EA community realizes, once you take into account the second-order effects even of projects whose outcome is deemed "successful."
I think issue 5B (Committed Pot of Money) is going to be key to making this work. Following your example, suppose an investor invests $1M, and funds a project that cures malaria in Senegal. Why is the result now worth $5M, or any value at all? Who figures out what it's worth? Supposedly a philanthropist will "buy" the solution for $5M. But if I am a philanthropist, why would I invest any money at all in problems that have already been solved? I would get more bang for my buck by funding problems that *haven't* been solved yet, so the "certificate" of "congrats, you helped cure malaria" will be worth exactly $0 to the original (for-profit) investor. For-profit investors know this, and will never invest.
What you need is to have something that resembles an old-school charitable foundation. Each year, it collects contributions from donors, both large and small. There is no limit on the amount of donations that it can collect, so this is scalable. At the end of each the year, the foundation puts all of the donations for the year into a "prize fund", and issues tokens equal to the size of the fund.
Over the course of the year, the foundation also collects various project proposals, each of which has a cost estimate (a grant amount). There is no limit on the number of proposals, or their dollar value, but the foundation will only fund proposals up to size of the "prize fund".
Once the prize fund is in place, for-profit investors can purchase a token (at an offer price of $1/token), and must name the proposal that they wish the money to go to. Once purchased, the token becomes a "named token", and is a bet on that particular proposal. Tokens for proposals that don't get funded, because they don't meet minimum funding requirements, (or for some other reason), will revert to being "blank" again, and can be redeemed for $1, or redirected to another proposal. Some sort of ranked-choice voting could make this process more automatic. The money from token sales is used to fund the proposals.
After a certain "period of performance", perhaps a few years, the funded projects are scored/graded by some impartial observer, and the prize fund is then used to buy back all of the the tokens, at some price proportional to their score. Tokens thus function like a prediction market; they are a bet on how the project will ultimately be judged. It is important that the observer be named in advance, and have no conflicts of interest, otherwise for-profit investors could game the system. The charitable donors themselves could vote on what constitutes success; this allows for-profit investors to game the system, but only by making charitable donations.
There will be some additional complexity if the "period of performance" differs from project to project (as is likely to occur); some kind of token rollover would need to be performed each year.
As a sweetener to the deal, the original "prize fund" should be invested in low-risk instruments like government bonds, so that the expected rate of return for purchasing a token is positive, and equal to the expected return of the underlying instruments.
Now there's a reason for donors/philanthropists to invest in the system: they know that each donation will be used to fund a project, but the foundation uses a prediction market, backed by for-profit investors, to decide which projects should be funded.
Dumb question: If the founder keeps some equity, how does that take away value from the beneficiaries?
Using the malaria example again, suppose curing malaria takes $1 million, and produces $5 million in value. The founder sells 50% of the equity at the beginning, keeping the other 50%, and then gains 2.5 million from a big charity house at the end of the project. However, the $5 million in value produced for beneficiaries of the malaria treatment still remains with them, right? How does that go to the founder?
I think we should be clear about what we're paying for with inefficiencies introduced by this system, and why we're willing to pay for that.
For a normal investment market, you are paying for starting capital because you don't have that capital in pocket.
In this case, what you're paying for is risk-reduction on the bet that a group with a proposal for fixing a particular problem will be successful. So basically, buying insurance. Or, 'outsourcing' the work of deciding which grants to fund.
Also, you are trying to pay potential founders for searching for cost efficient, novel, and likely-to-succeed proposals.
Then this gets entangled with trying to implement those proposals. The founder must chose a team (often starting with choosing someone to help them choose others). Then you make some initial attempts at starting on the proposal.
Each of these steps may uncover important information.
Did the founder manage to assemble a competent team? If so, they're much more likely to succeed at their aims even if their initial plan wasn't great and will need substantial revisions as the project matures.
Did the initial attempts show up unexpected flaws in the initial plan? If so, was the team able to come up with a new and improved plan to remedy this?
The nice thing about having rounds of funding is that you can seperate out some of these issues. Maybe you can give grants like, "We'll give you 1 year of funding to get a team together and get started on this 5 year project. At the end of year 1, we'll decide whether to extend the grant to the full 5 year funding or cut our losses and tell you that we no longer have faith in your ability to implement the project within the proposed budget."
Alternatively, what about having people commit to joining the team if the funding is achieved? And then making a grant proposal where the proposed team is part of the proposal? Maybe the grant only goes through if you really do get at least 80% of the promised team to commit?
One of the grant problems I faced in academia was that to write a likely-to-be-funded grant proposal (in my field anyway), you needed some initial show of backing for your idea. You needed to be an already established researcher with a lab, or you needed someone with an established lab to believe in your project and grant you some resources 'stolen' from other projects to get your initial data for your grant proposal. This creates a huge status quo bias where established labs are much more likely to win grants. Also, the grants tend to be much more focused on boring incremental progress with high likelihood of success, then on trying radical new ideas with low success likelihood but high potential payoff.
The start-up venture capital culture of taking a bet on a variety of different not-previously-funded groups of ambitious newcomers is a refreshing change. I think having some of that style of funding mixed in with more academic grant style funding would be great.
In the case of Open Phil and Future Fund, they've been doing a pretty good job (I think) of keeping the balance between funding established groups vs newcomers.
It makes sense that funders like OpenPhil and Future Fund would want to be able to 'outsource' or 'insure' their granting decisions. I don't necessarily think this is a fatally flawed way to do that, but I think it hides some of the intentions and costs in a problematic way, and sets up some perverse incentives (makes Goodharting or impact-assessment-manipulation extra tempting).
I think you could have a basically equivalent system that was more clear about what it was doing. Seperate out some of these goals into more clearly delineated systems.
What about starting a 'charity insurance' group who would do their own independant grant evaluation and let grant givers buy insurance at a negotiated rate against the failure of the project? That would take the whole 'founder gets rich' question out of the equation. A separation of concerns.
I do think that founders being able to make extra money, above and beyond a normal paycheck, if they have an unusually good idea and assemble a crack team and guide this team to overwhelming success. If unexpectedly great good gets done in the world as a result of this, shouldn't there be some payback for the founder and team? Probably at a less impressive rate than if it were a capitalist venture, and with some risk-mitigation... but still. That spirit of 'we might all get rich in the pursuit of this awesome but hard thing we're trying to do together' is pretty darn motivating for teams.
Or another example of trying to seperate out the concerns, what if you were offering a prediction market on whether a particular potential grant would end up achieving its aims? The market would close before any grants were given, and the probabilities would be used to guide the eventual grant-making decision. Some time later, when the success of the funded projects is retroactively assessed by the funding agency, the prediction market pays out to the ones who bet correctly.
In terms of flaws of the original retroactive funding idea, I agree with Daniel Speyer and Jacob (Writes Streams of Consciousness) who bring up the issue of 'concern squatting' in their comments here. This arises if there's a particular problem that multiple people want to address, but retroactively it would be hard to appropriately determine who gets the credit for fixing the problem if the problem gets fixed. In the retroactive funding model, this disincentivizes multiple groups from trying to fix this issue at the same time. This is a very different dynamic to the idea that two startups might try to race for control of the same market space. Having charities race to fix a problem sounds like it'd be potentially quite good for the beneficiaries. The idealized healthy competition of an open and volatile capital market space benefiting consumers, ported over to the charity world. So yeah, we want something to help us get some of the good aspects of capitalism into the space, but I don't think this proposal for retroactive funding quite does it.
Re: Should Founders Be Getting Rich Off Impact Certificates?
One possible answer here is "pay less money for completed projects." Every month, the final funder announces a funding multiple between 1 and 0. For all projects that get funded that month, when the impact is finally assessed, the impact is multiplied by that funding multiple. Suppose the malaria project was originally funded during a month where the funding multiple was 25%. When impact is assessed at $5 million, the payout to investors is $1.25 million.
If the final funder announces a funding multiple of 25%, then they're essentially saying that investors shouldn't fund anything below a 1 to 4 return in expectation. But you can tune this value: if you have too many good projects and too little funding, you can lower the funding multiple and be more selective. If you have too few projects getting funding, you can raise the funding multiple.
Note that this is different from proposal 11C: that project is equivalent to a funding multiple of 21%. But in 11C, you're allowing the funding multiple to vary between different projects. If you have one project which is efficient, then maybe 20% of its shares get bought for the startup price. If you have a project which is inefficient, then maybe 40% of its shares get bought for the startup price. But both projects ultimately get funded, and both investors make similar amounts of money, despite the fundamentals of one project being worse.
I find this idea super exciting. It's essentially posing a market design problem, which is my broader field of research (I'm a computer scientist / economist).
However, I think some of the basics here are very fuzzy. The post mentions principal-agent problems (between funder+investor as principals, vs founder as agent. This is a pretty standard problem, no different than funding a startup, or hiring someone for a job. We can align incentives by giving the founder a share of the profit.
But what about the alignment between the funder and the investor? I don't see at all how an investor would ever trust the funder over the span of a multi-year project based on nothing but reputation. Charities, just like any organization, are subject to value drift and their own internal principal-agent problems.
What's missing is a way for the funder to make a promise at least as strong as a legally binding contract. As much as I don't like crypto, this is the exact type of problem that smart contracts are supposed to address. But this reduces the possible scope to very legible outcomes only.
Another solution might be a centralized market maker that creates a two sided market between funders and investors. This reduces the trust problem to a single entity. The way I envision it, the funders would advertise their project through the market maker, who holds their money in escrow and acts as an arbiter, making the final, binding call for any impact calculations. The market maker will gain a track record of making fair decisions much more quickly than any individual funder. They would take a small fee off of every transaction, which incentivizes them to build and keep trust over time.
(This kind of mediating entity already exists in other contexts: e.g. banks facilitate international trade by acting as a trusted third party for a fee. If I want to sell a container of malaria nets to Scott, they way this works is that Scott gives his bank money in escrow, I send the container, and the bank gives the money to me as soon as they make sure that I actually sent a container full of malaria nets. Banks have a very strong incentive to keep this scheme going as they're skimming around 3% off the top for doing essentially nothing. Long-term business partners will usually deal directly with each other, but that requires a good deal of trust which is surprisingly hard to obtain over international borders.)
"If no, I can’t think of a good solution to the “fast people get rich” problem."
One solution would be an auction that sells a variable number of shares for a fixed price.
If a project will obviously do $5 million of good, and costs $1 million, then the founders keep 80% of the impact as meaningless tokens they can't sell.
Another solution is that the extra money gets sent straight to a generic funding body. Say funders can choose where the extra money goes, and that must be an actual charity, and the excess money gets split proportionally, and the money is a perfect gift, the funders get no extra impact certificates.
A third solution would be that the money is transformed into a token. The token can only be redemed through the fund charity, receive impact mechanism.
Suppose the market rate for such tokens is 50c/$. Bidders fund the $1 million in, $5 million out charity to the tune of $9 million. $1 goes to the actual charity. $8 buys tokens, which have a market value of $4 million. The tokens are then given to a less effecive charity, and recieve a reward of $4 million. Thus the funders receive exactly $9 million total.
A token, like a book token, is a dollar in a locked box, where that box can only be opened in specific circumstances. (Anyone can convert dollars into tokens, but if they do, that implies the market is broken. Suppose a project needs a million $, so it auctions off a million tokens, of face value $1. Bids can either be >=$1 real money, or a mixture of real money and tokens with a combined face value of $1. Sort all bids, from most real value to least, and pick the one with the 1,000,001 th largest amount of real money. All the first million bidders pay that. (The million bidders might be just 3 people, each bidding a lot of bids. )
This mechanism means that, if the funder will retroactively pay $1/util retroactively. Then when investors see a 10 util/$ opportunity, they bid it up until they end up with loads of extra tokens. They then have to spend those tokens somewhere. Any charity who produces >1 util /$ will get fully funded by people bidding in $. So you can't get rid of tokens there. So investors look for places with 0.9util /$ to get rid of the tokens. And failing that 0.8 util/$.
Assuming an infinite number of omniscient investors, who will take any opportunity to profit, however slim the profit margin, then this mechanism is 100% efficient at distributing the funders money to the most efficient charities. Its also complex enough to deter any non omniscient funders.
Another option for founder funding: fixed founder bonus. Founder announces a bonus amount they get paid if the project gets bought out. Beyond this they are paid market rate salary and have no equity. When the project gets bought out, the founder gets the bonus, with remaining proceeds divided proportionally to the investors. Investors know the guaranteed founder bonus up front and take this into account when deciding whether to fund.
Example: $1 million dollars is needed to fund a project with a $250k founder bonus. Investors fully fund the project for $1 million up front. The project is successful and gets bought out for $5 million. Founder gets $250k (in addition to salary they have been collecting) and investors are paid out $4.75 million.
The founders will never get as rich as private sector founders, but they get a healthy reward. The promise of the bonus is probably enough to push some founders towards charity rather than private sector.
There are some details to be hashed out if the project gets bought for less than the investors funded + founder bonus. I would suggest that funders get initial investment back first, then founder bonus is paid with whatever remains.
Disadvantage: Founder does not have a financial incentive that scales with the size of success. This is counterbalanced by them retaining some personal credit for the success of the venture, which does scale with the size of success.
What about the problem of marginally useless charities. For each million the charity recieves, it can build a bed net factory.
The countries it plans to build factories in are, in order, Rwanda, Somalia, Greenland, Kenya, Madagascar.
Predictably 4 of these factories are useful, and the Greenland bednets are totally useless. So once they have $2 million, no one will fund any further. (Unless a single big funder or group of smaller ones can organize $3 million)
So instead of the charities choosing the funding threashold, let the investors do it. Let the investor who only thinks the project will do any good if they get >$5 million buy a share they only have to actually pay for if funding exceeds $5 million. (With a fancy algorithm to pick the largest stable equilibrium, namely assume everyone pays, then start refunding people who haven't met their funding condition.)
Fascinating and thought-provoking post for someone unfamiliar with this concept; thank you.
Initially, I had some concern about the moral hazard for oracular funders. Why would they retroactively sponsor completed projects when they could apply their funds towards further improving the world? But I'm less troubled by this the more I think about it. These funders would be playing the long game, and recognize this retroactive funding as valuable for keeping the impact market functioning and incentivizing future projects. In a sense, the concern is similar to asking why people engage in philanthropy at all, even when they aren't directly incentivized to. If they do it out of a sense of altruism, the same altruism could support oracular funding.
I think the most important detail to figure out is the balance between "maximum capitalism" and "project funding only". Like you, I'm deeply uncomfortable with either extreme. I think investors and founders shouldn't capture all the surplus, but there has to be some surplus available to them to make the system work. There seems to be an assumption that the oracular funders would pay out the full value they ascribe to a completed project, but wouldn't it make more sense if they just paid a sort of prize that scaled with the project's impact value? As long as it was calibrated to provide a good return on investment, it seems like it would make more effective use of the oracular funder's funds.
To answer the question of how much to pay, why not hold an auction after the completion of a project? The most generous funder would "win". While someone could certainly pay $5M for the certificate of a project with $5M of impact value, I think a better norm would be for the auction to clear at, say $2M, rewarding the founder and investors without giving them the full surplus.
One aspect I didn’t see discussed explicitly, which might be worth exploring, is risk.
If the $1m malaria project has a 20% chance of success, I’m indifferent at an expected Oracular Funder buyout of $5m. But there is still a trade here; as a Funder I might pay a premium to avoid risk.
So in addition to the “VC” being in some sense better at picking winners, there is also a component where they are just more risk-tolerant. They can sell a service where they buy 5 20%-success projects at $1m, with an EV of $5m, and sell them for $6m; the charity might not be prepared to risk all of their capital on high-risk investments that could all easily go to zero, and instead would be willing to pay a premium to shift the risk into another party that is more risk tolerant.
By financializing in this way, you can increase the total number of projects that can get funded, adding a bunch of more-risky stuff that charities aren’t able to fund. (Of course, this risk premium depends on there being a lot of “high risk” stuff that isn’t currently being funded. I don’t have a feel for whether that is the case.)
"The Copenhagen Interpretation of Ethics says that when you observe or interact with a problem in any way, you can be blamed for it. [...] *In particular, if you interact with a problem and benefit from it, you are a complete monster*. " ([0], my emphasis). This should be considered.
* Any charity with for profit investors is sure to annoy people even vaguely involved with non-EA charities to no end: for them, charities are about attending galas, photo ops with smiling orphans and warm happy feelings. I would argue that such people can perhaps stomach paying a living wage (never mind a market wage) to some charity workers, but ideally, everyone in charity should be a hero, and heroes just don't ask for fair compensation. If you thought the pushback against billionaire philanthropy ([1], [2]) was bad, wait until you see the response to what will be seen as evil capitalists enriching themselves through charity. Making money out of bed nets will probably be viewed as slightly more evil than making money out of selling small arms to some warlord.
* In the public eye, such projects would probably be held to a high standard of virtue of a charity while lacking the advance of good will provided to not-for-profit charity projects. The public won't care how many bed nets the project provided when some minor ethical mishap (like paying bribes) comes to light. If the oracular investor is in any way beholden to public opinion, this is a real risk.
* Donor motivation might be a hard problem: outside of EA circles, where the long term impacts are taken into account, the view "why should the donations pay to reward some greedy capitalists beyond their initial investment?" might prevail.
* I think that the public outrage would vary very much with the field. Health interventions in the third world will probably be a PR minefield, while other fields will fare better. AI alignment is safe, in the mind of mainstream journalists, hard AI is generally considered a pipe dream, so if rich people spend their money on it they generally don't care beyond some op-eds about how "the Real Problem is Algorithms". Development of meat replacement products is safe while interacting in any way with actual animals is not.
Implementation idea: normal startup:
* From my understanding, most startups in the tech sector are not aiming to be independently profitable, but to be bought up by tech giants. When youtube was bought in 2005 by Google for 1.65G$, the yearly revenue of 15M$ clearly was in little relation to that price. If some EA charity established their willingness to buy up successful charity-related for-profit startups, this would allow to use all of the funding infrastructure in place for other startups. I think that this might be more appropriate for continuously run projects (e.g. build and run in the third world), where the startup represents the infrastructure build which can then be taken over by the oracular investor, and less for goal oriented projects (like "vaccinate every child in X").
Further random thoughts:
* If multiple projects compete in the same metric (e.g. QALY improvements), this might lead to the marginal price per unit to be lower, I guess?
* If multiple projects compete over the same population instead, competition might be harmful to the overall outcome. Of course, there is probably no general oversupply of charity, so that competition may be limited. Or projects might try to use patents or lobbying to hinder each other, which seems worse.
So, let’s imagine that these oracular charities agree that a life is worth about $5000.
And then I (a random Joe) save some drowning child, and bystanders collect enough video footage to show the kid would’ve died without my intervention. Could I sell the credit for this deed to an oracular?
And if so, could a volunteer lifeguard sell the credit? What about a paid lifeguard? Or a doctor?
I could go on, but my point is that it’s unclear to me where his line should fall, but also funding good sameritanism seems good.
Suggestion for the design to avoid rushing / paying founders (#11).
(Note: I have actually taken courses in auction theory. I have not read all the comments to see if this was already suggested.)
Project costs $1 million. Oracular funders think it's worth $5 million. Rather than bidding on the share of the project that they're buying, funders bid on the amount of a haircut they're willing to accept. Funders willing to accept the highest haircut are first in line to invest. Maybe you run through all the investors willing to take a 70% haircut, and then have some investors willing to take a 60% haircut who get to invest, but everyone who said a lower haircut can't buy shares. So that project was funded with a 60% haircut. Then, if the project succeeds, the oracular funders only have to pay $2 million to investors (40% of what the project was assumed to be worth).
This avoids the founders getting wealthy or having weird incentives. The savings from "it was broadly believed this project would succeed" go to the oracular funders. Since if folks are willing to accept a really high haircut, probably the oracular funders could have figured out to fund the project without this fancy mechanism. (Conversely, high risk projects will probably have very low haircuts.)
>Ofer and Owen Cotton-Barrett have discussed this here. Their conclusion is that final oracular funders should take this into account when deciding who to grant impact certificates to.
Hi, Ofer here. Speaking only for myself, I don't see that as my conclusion. I elaborated in a comment on the EA Forum:
Based on my very cursory knowledge of nonprofit organization law, I would assume that it would be easiest, from a legal perspective, if the founders don't receive any (or at least don't receive much) of the impact certificates, and certainly don't get rich.
I'm curious what the lawyer you consulted with thought about the question of whether founders should receive impact certificates and potentially get rich.
The reason I would think it would be easiest if the founders don't get the certificates is this:
First, nonprofit founders don't own their nonprofits the way that for-profit company founders own their companies. So, if the nonprofit organization is issuing the impact certificates, then I imagine the nonprofit ORGANIZATION can choose to keep some of the certificates (as opposed to distributing them all to VCs). But I would think that the FOUNDER has no right to 'keep' any of the certificates—Since the founder never owned them to begin with.
That said, of course, the nonprofit has the power to pay the founder for the founder's work as a staff member. And perhaps part of the founder's pay could be in impact certificates, the way that companies sometimes pay their regular staff (not founders) in stock options. But, my understanding is that, for the pay to be legal, the pay must be genuinely based on the founder's work as an employee. Otherwise, as another commenter pointed out, I'd think the founder would technically be embezzling money from the nonprofit by taking more value from the nonprofit than what the founder provided as an employee.
Of course, even if I'm right about my above assumptions, this could be solved by a founder choosing to set up a regular, for-profit company to do good work, instead of setting up a nonprofit! There's no reason that a for-profit company can't do charitable work! :-) But a lot of US charities are currently set up as nonprofits. So that's why I think it'd be simplest and most convenient if the founder doesn't get many impact certificates. Because then existing charities could get in on the fun!
Another reason I think it's easier for the founders NOT to get rich from this is that most US nonprofits seem to have 501(c)(3) tax-exempt status. 501(c)(3) is a really cool tax status that gives a nonprofit lots of benefits. The nonprofit doesn't have to pay income taxes, and donors can get tax deductions, among other things! But one of the rules, as far as I'm aware, of 501(c)(3) tax-exempt status is that no part of the nonprofit's income can inure to the benefit of any private individual. My fear is that, if the founder gets paid too much (including by getting too many impact certificates), the IRS might feel that part of the nonprofit's income is inuring to the benefit of the founder. And then the nonprofit might get in trouble.
Again, this 501(c)(3) issue can be addressed just by the founder choosing NOT to apply for 501(c)(3) status. So there's a clear work-around. But as I mentioned, 501(c)(3) status rocks. So, it'd seem easier if founders don't get rich, so that the nonprofits can be 501(c)(3)s.
Lastly, another option for addressing any of the issues I'm bringing up is to push for legislative change. So I'm only talking about the question of what seems easiest given my cursory understanding of the current law.
Very interested in what Scott's lawyer said about this topic!
Should there be a mechanism for distinguishing projects with investors from projects without? If the incentive for the post-hoc funders is that they get to own the 'best' impacts, and they could pick between rewarding a) a project with investors and a medium impact (say getting bednets to X amount of people) or b) a project that was funded through other means with a more impressive-sounding impact (say inventing a new vaccine and delivering it, or something), then they have no reason not to just pick from the whole range of all available finished projects, right? In fact I guess there are lots of funds that work like this already -post-hoc prizes - like the Nobel prize or smaller equivalents, that run on identifying finished projects, giving their people money/glory, and getting credit for doing that giving to the best projects.
The only reason I can imagine them switching to only investored projects would be for the goal of promoting this exact market and incentivising investments like this. While I imagine many of these foundations would see the value in increasing available funding like this, I also think a lot of them are very prestige driven, and risking their reputation would be a high bar to pass. (also, as someone else here mentioned, what if one of these institutions, independent of the market, wants to give a Nobel or other prize to a project inventor/doer in this market who had sold off 75% of their 'shares'? Do they have to split the prize money with their investors? *Can* that glory be split?)
I think especially if the idea is to start off with a smaller/restricted market of ideas/projects, there will be a strong barrier to entry that any finder who joins will need to be really really into the idea of helping create this market and its future growth, because they are giving up a large opportunity space of 'projects they can pick between to reward' by joining this.
Not insurmountable, at all! But while reading, I thought there was lots of good ideas and discussion of incentives for project makers/leaders and investors, and their pitfalls, but I found myself a little lost in the incentives for the oracular funders and how they would interact with the market format.
From mobile, apologies for typos or potential unclarity from that.
Has this model actually been used anywhere yet? I'm not a lawyer by any stretch, but this seems not legal if the final oracular funder is trying to operate legally as a US foundation, nor will what you are calling a "charity" qualify as a 501(c)(3). In particular, see "none of its earnings may inure to any private shareholder or individual." (https://www.irs.gov/charities-non-profits/charitable-organizations/exemption-requirements-501c3-organizations)
Maybe this is the point? Funding things with social impact, but for a profit? I just don't get how this works from the final funder's perspective or the operator of the non-501 "charity." If they're not operating an organization that is actually legally classified a charity, then they're allowed to profit from what they're doing themselves. So why would they sell all profit to their initial funder instead? That isn't how for-profit startups operate today. Nobody gives 100% of the equity in their company to their first investor. The final funder would need to necessarily not actually be a foundation, because foundations are legally required to give their money to actual charities if they want to remain foundations and not pay enormous amounts of tax on their endowments. So it seems like you're instead going to be forced to rely on extremely high net worth individuals who could donate money and receive large tax writeoffs, but instead choose to donate to something that they can't write off instead. I'm sure some non-zero number of high net worth individuals will do that, but it's a hell of a culture change.
This all gets worse if you involve crypto, because at some point, the organization receiving funding has to pay its employees and buy goods and service, and exchanging crypto for real money to do that is itself a taxable event.
Re 4/11: You could pay out the $5 million in the mana that manifold.markets uses - money that can be paid out in donations to charities of your choice. The founder sells half their impact for $2 million in mana, pays out half that to a charity that writes grants to projects that proved their potential by selling impact certificates, gets $1 million in funding, and keeps $1 million in mana, then gets another $2.5 million later. Founders are allowed to get rich, but in mana. Might this also help with the tax issue?
Re II A: Sounds like a job for insurance. The insurer would be obliged to pay the final oracular funder for whatever negative impact they find. The cost of the insurance would be added to the required funding. So the cancerous malaria cure would cost not $1 million to fund, but $26 million.
I haven't read many other comments, so it's probable some of these points have already been made, but I'm confident enough that at least one is novel that I'm willing to gamble your time to save some of mine. Five points.
I. A finer model of the funding ecosystem
You break the ecosystem into founders, VCs, and occular funders, but I think you need to disambiguate between different sorts of people in your VC category. I don't mean angel vs seed vs series A and B funders -- though those may well be relevant here -- I mean direct funders vs aftermarket funders. Let's call people who directly buy equity from funders VCs and people who buy equity from others traders.
You seem to have a lot of distaste for traders, which I empathize with: they make a profit without increasing the pool of money that goes to charitable projects. There's an argument to be made that the existence of a liquid aftermarket derisks the original VC investment, but it's unclear whether that benefit is enough to justify the reduction in money that goes to the project, the founders and the original VCs.
This leads us to...
II. You can send the surplus where you want it to go, by decree
Once you resolve for yourself the question of where you want the surplus to go, you can build the system out to enforce that distribution. The way this would work is that, any time equity is sold for a profit, some percent of that surplus is automatically siphoned to other parties -- probably founders or VCs. You probably need a trick for when the asset value decreases in between transactions, but I'm sure that could be worked out.
This could be done with NFTs, building it into the code of the blockchain, or with a managed online marketplace or some sort of royalty contract. Which option is best probably depends on your preference for working with lawyers vs programmers.
III. Is charity more like art or like startups?
I think the claim here is that equity in successful charities probably acts a lot like fine art as an asset class. Probably this extends to tropes about starving artists and stories about the wealthy using them as tax-advantaged investment vehicles.
IV. Social credit rules aren't necessary and wouldn't work anyway
> If they sell off 100% of the equity, then in some sense that we will have to hammer out socially, [founders] “get no credit” for their project
is the weirdest line in the whole piece. Founders would want to make successful charities because they want to effect change in the world, because they want to be known for effecting those changes in the world, and maybe because they want to get rich. Telling them "you sold all the equity, so you no longer 'have credit' for your work" won't prevent them from having been the team that did the ting, from being known as the team that took the risk and did the thing, or from enjoying consuming the money that came from getting paid to do the thing.
You could write legal provisions to enforce not claiming responsibility for a charitable act if you don't have the equity -- something between a nondisclosure and a nondisparage -- but that seems like a terrible idea.
To me, the focus on social credit seems misplaced. Instutitional funders know that their real power is in ecosystem and common knowledge effects, so committing to fund projects after completion in order to diversify the sorts of projects that get funded seems, knowing that the functioning of this system rests on their credibility, to already be in line with their incentives.
V. Let founders get rich
I also have to say that your aversion to letting founders get rich dosn't mesh with my model of your beliefs. Isn't the great thing about markets the fact that they reward people for doing what others want? Isn't that what makes capitalism so effective at increasing output? Why should it be different when what others want is altruistic?
This whole thing has the perspective "impact certificates would be kind of like shares of stock". But what if impact certificates just *are* shares in companies that are trying to do good? I think this solves a lot of problems. Once a credible retroactive funder exists, people can start regular companies whose business plan is to do things that will get them paid by the funder. Those companies can use the existing legal structures for startups, sell shares in the usual way to regular seed investors and VCs, and later on could even IPO on existing stock exchanges if they need a very large amount of capital.
There is a whole startup ecosystem that is familiar with how to set up new companies, how to sell shares to VCs and other investors, how to navigate securities laws, etc.
Traditional companies also have experience dealing with problems like this:
> suppose that you are an investor who believes that the plan to cure malaria in Senegal will succeed (produce $5 million in value), but the grander plan to cure it elsewhere in Africa will fail (produce $0 in value).
This exact problem happens with regular companies selling stock. You might invest in a company because you believe the projects it will do with that investment will produce a good return. But then the company managers may later undertake bigger, more expensive projects that don't produce a good return, wasting money and destroying your initial investment. In fact, company managers have an *incentive* to do this, because starting big new projects is fun and interesting for them and increases their personal prestige, and the managers may not particularly care if the investors don't get a good return from the big new projects.
Some traditional ways to reduce this problem for regular companies:
- The share owners in some sense *control the company*. Shareholders can vote out the current management and install new management if they think current management is not doing a good job at producing a return on their investment.
- The managers' incentives can be made more aligned with the shareholders. For example, you can pay the managers in shares, or you can tie managers' pay to the share price. That way managers have an incentive to do things that produce good returns for shareholders.
1. I was discussing this with a colleague, and he pointed out this is very similar to how NIH grants are funded. You write the grant, but in order to get all the data that justifies the grant you basically do 75% of the work the grant is for. Once you get the grant, you're practically done with that project, and the grant funds future endeavors. The only difference between NIH grants and an impact market is that the NIH is effectively funding 'people who did good work in the past, and are expected to continue doing good work in the future'. I kind of think of this as the same fundamental idea with impact markets, but that second half of the idea (funding people we expect to continue doing good work in the future) isn't as explicit.
2. From a great interview with the founders of Air BNB from when they were growing, there's an insight I think impact markets could learn from:
"Paul Graham ... asked us a question, 'Where are all your users?' And we said, 'Well, they are kind of spread out everywhere.' And he said, 'Well, where are most of them?' And we said, 'In New York.' And he said, 'Well, why aren't you in New York right now?' And we said, 'Well, because we are here at Y Combinator, like you told us to be.' And he said, 'Doesn't matter. Go to New York. I want you to meet all your users.' And that wasn't intuitive to us at the time. Because we didn't have any money to do that. ... I mean, the whole idea of being on a web business is that it's scalable. You don't need to meet your users. But he said: It's okay to do things that don't scale when you are just starting out."
This is a powerful insight. You're trying to do some proof-of-concept things. You're trying to get a lot of people into a market and make it scalable. That's awesome and potentially world-changing if it creates strong efficacy incentives for charities. But you already recognize there are some things that only work when there's already a trusted market. In other words, there's a lot you can only do at scale. So don't try to do them yet.
Crypto is great, but is it necessary at this scale? Maybe start by trying a few hand-crafted impact credit projects. Likely, you'll find there are some institutional ideas you hadn't thought of that are necessary to make this work. You'll step in and make the project happen anyway through personal intervention - not something you want to do when the project scales, but that's all part of the learning process - and make sure everyone is still happy with the arrangement. Then you'll tweak some factors, do it again, and have to do a little less intervention than last time. Meanwhile, a few more people will see what you're doing and ask to get in on the game. It'll be small, but manageable, and it will grow in proportion to how well it works.
Don't start at scale. That doesn't give you room to learn and grow. You're either 100% right from the start and wildly successful, or you're a high-profile failure that 'demonstrates these new ideas just don't work'. Start small and build a network while building institutional knowledge, then scale up commensurate with need. That should solve most of the problems you outlined above.
Finally finished a long post on a complicated way to avoid charity founders or investors getting rich off the easy opportunities. In principle the money should go towards charities the investors would otherwise consider not quite worth it. Given many self interested omniscient capitalists, the retroactive funders money is perfectly distributed to charities, most effective first, with no inefficiency. https://forum.effectivealtruism.org/posts/CGqbfs28amWjM6PxK/retroactive-funding-with-subsidy-markets
A while back I randomly thought up a sort of ... blind, assurance contract with a scaling refund mechanism? ... Fundraising scheme but never knew what to do with it. Not sure if it'll help resolve any of the issues you shared, but this post seems like an attractor for the kind of people who could get value out of it. One page description: https://docs.google.com/document/d/1dBTPsiM3nIOS-2KlO8nBIJIlW7EIMyfbshA3n-LgcTM/edit?usp=drivesdk
My initial brainstorming that I haven't thought through completely.
"suppose you invest when there are $1 million of tokens sold, thinking that you are a slightly-better-than-average investor and so can probably do better than break even. Then later some other people come in and buy $9 million more worth of tokens. Now there are $10 million in tokens chasing $1 million in prizes, and on average each investor will only get 10% of their money back."
Hmm. Normally this shouldn't be a problem. You paid what you did, other people buying shouldn't change whether that transaction was a good idea. But it is here.
So, to iterate the problem as I see it. There's a set of projects, that will produce X utilons. Your goal is to predict what fraction of those utilons will be produced by a given project (in the judgement of the granter). Once you've done so, you can tell the value of a project in expectation. If the project sells a fixed amount of shares, then you can calculate the expected value of a share as (fraction of utilons)*(total prize money)/(shares).
That's the goal, but this could be confounded if new projects are added or removed, as that would affect the total utilons (and therefore the fraction of utilons). A common way for projects to be removed is if they fail to meet their assurance contracts. So you do need to predict which projects will get funded in order to predict X, in order to predict relative impact. Similarly, new projects can't enter the market without changing the value of everything else.
So the way others buying tokens affects you is it makes more projects funded, increasing X, and therefore lowering your expected payoff.
So what we want is a fixed price per utilon. An unlimited pot of money works well for this, as described, but requires an unlimited pot of money.
In the interest of brainstorming ideas, another way would be for the granter to commit to spending money beforehand. Each proposal says "Here's the impact I expect to have" and the granter says "if you have that, I'll pay $X for it". They commit their pool of money, and then stop accepting proposals. Every one that has that impact gets paid, everyone that doesn't doesn't, and the leftover money goes into the next round.
Issues are that this breaks things into rounds, and doesn't allow new projects to enter as needed (perhaps fine while still in the startup phase, as long as it doesn't fix a precedent for the full idea). There's also no method for handling "mistakes", i.e. what if the project fails to have the impact it expected? In simple cases, they failed, they don't get their money. But if they succeed wildly, and have twice the impact, they can't receive twice the money. Or if their plan has unintended side effects, those also aren't priced in.
(added post fact: the above seems related to various later points)
"7: How Should The Market Price IPOs?"
Can't this neatly tie in with the "How should founders get paid" part? Something like the founder honestly reports that they're going to split it into 10,000 shares, and how much they need in funding. Then they sell shares, and people buy them at whatever price they're willing to pay, based on how much they think the project will be worth. If the shares are being offered overpriced, then the founder needs to lower the price, and not keep as many of their own shares if they want the project to be funded.
If a project is going to produce $5 million, each of the 10,000 shares is worth $500, and so the founder needs to sell 2,000 of them to get funded, leaving 8,000 for themselves. If people actually expect it to only be worth $2 million, then each share is $200, and they'd need to sell 5,000 of them to get funded. This encourages low cost, high impact projects.
Ah, you mention this. I'm not sure if this is just your auction idea.
"plus people feeling more comfortable knowing that charity founders aren’t getting rich and buying yachts with their donations"
I think while this would be a real effect, it's not a "reasonable" one? The reason charity founders buying yachts with charity money is because it means the charity is usually running away with your money instead of doing the job. But that's what the impact market is for. It kinda guarantees the job is done, so rewarding the person for doing the job is less of an issue. It basically bakes in the accountability that is usually lacking in other cases where a charity buys yachts. Maybe I'm wrong about that, and even if I'm not, it won't stop people from feeling this way, but still.
"I don’t like the fact that it’s important they want to keep as much as possible but they have no real incentive to do so - "
I know it's the same in terms of actual results, but maybe if you marketed it as "79% goes back into funding other charities", it works better? It feels like people in charitable work should be at least partially incentivised for charity in general to be done.
I wonder if you could solve the "founders getting rich or not" problem by giving founders some sort of generic market token equal to, say, 50% of the market value. Say, they sell 50% of the shares in the 1 million funding 5 million value malaria project. Investors get 2.5 million in payout, the founder gets 2.5 million worth of "generic impact tokens" which they can then reinvest. Keeps the money in the market and doesn't feel as icky while still leaving room for an eventual liquidity event for smart founders.
What about treating it more like a market for bonds or commodity futures rather than stock shares? Suppose the Senegalese Ministry of Health and Human Services (or whatever) currently has no cash to spare, but figures if they could collect at least $5 million more in tax revenue over the next thirty years if the malaria problem were completely dealt with. They draw up a stack of 360 "payable to the bearer" certificates, or perhaps some more elaborate system broken down by provinces or individual villages. Each month, another certificate matures and becomes redeemable, at the appropriate government office, for an amount of money calculated based on the prevalence of malaria in the specified region during the previous month: $14,000 for a spotless record, declining incrementally to $0 for "indistinguishable from the status quo," and on down into negative payouts if the situation gets worse. Certificates are distributed to alleged malaria-remediation projects upon request, requiring only a simple credit check (does this person or group have a fixed address, professional reputation, something we could go after if they end up owing us), since a null-hypothesis outcome costs the government nothing. Government-backed certificate plus actual plan can then be sold to an investor, or split up further into derivative contracts.
I'm a bit lost in all the subtleties of this, and not an expert in any kind of finance, but it sounds like you're looking for a model with some properties of equity (unlimited upside, notion of "ownership" of the project itself) and some properties of debt (finite-scope though potentially renegotiable funding, unambiguous "cashing-out" event at some future point).
If you haven't already you should send a friendly email to Matt Levine explaining your conundrum and asking for his suggestions-- because not only is he likely to have helpful input, he might be interested in writing about impact markets and raising awareness about the idea.
I'm an ordinary person, and I'm much less interested in being an investor than I am in being an oracular funder. Currently I give my charity to the SENS foundation because I think life extension is important, and I trust them to be better at evaluating promising life-extension research than I am.
...now that I think about it, isn't giving to a large science funding organisation on the strength of the good research they've previously funded basically an impact market? With the funding org acting as the investor, and me the oracular funder?
My tiny mind confessed to be confused: this looks like a weirdly convoluted approach to something that should be simple.
Instead of all this structuring, why not just say “whoever achieves X I will pay $Y”. Then, companies (structures that already exist to take investors money to utilise for projects they predict will create value inclusive and above of the cost of that capital) will just say, great, Scott will pay us $Y, we can do this project for ($Y*(1-r%)), awesome, let’s do it”. Maybe some companies will form _just_ for this purpose, bypassing the painstaking structuring you outline above to just get the thing done.
It seems that all you need to do is set the prizes, and then the methods will work themselves out with existing collective-projects-for-money (debt-and-equity-funded companies). Why all this headache?
You could also do a cost-plus model, “I will pay $Cost*(1+r) to whoever achieves this outcome”.
Why am I being dumb here? Please tell me what I’m missing!
One thing that seemed glaringly unusual was the idea that funders would want to buy these impact certificates in the first place after the “impact” has already been realized. While that might make these certificates work, I’m not sure that this is psychologically what would motivate a funder. I suppose the comparison is that the funder is paying out prize money for a contest they never set up in the first place. Without this assumption the whole thing kind of falls apart so to address this issue of funder incentives I’d like to the propose the following:
The impact certificates value isn’t actually tied to the measured impact of the program, but it’s tied to the right to be able to provide further funding to the program that the original certificates funded.
A funder in a sense would need to buy the rights to be able to provide funding to this organization and the holders of the impact certificates have those rights and are able to sell those rights to another party.
The assumption here is that private parties may be better at finding and investing in promising organizations than funders and bear the risk and search costs. If a specific program or organization becomes a promising organization to invest in because of evidence or belief or what not, there will be demand to be able to realize the impacts of investing in that organization and holders of these certificates have the exclusive rights to be able to do so. The value of these certificates themselves can be negotiated between parties. With a transparent ledger, like blockchain, the value of these certificates can be made public information and viola, do we have a working market for impact?
I think it would be useful to have a very high level summary of what is trying to be achieved by this project:
Charitable givers have more money than they have ideas. Therefore, they would like to spend some of that money on new and better ideas or selection between ideas (hereafter collectively "wisdom") instead of on the projects themselves.
I like having this simple input/output framing, because it immediately makes obvious the tradeoffs: 1) more money to founders will get you more and better ideas. 2) more money to investors will get you better valuation of existing ideas. But increasing either will mean less money will go to the projects themselves.
The other point I would like to harp on is that a stock market can only value things that have units of [money]. Since companies produce revenue, it's easy to make a stock market work: the thing that is being traded is a proportional claim on future revenue. Stock market fundamentals investors only have to solve one problem: predict the future revenue of company X and compare it to the current price. Impact investors have to solve two problems: predict the impact of project Y and the oracular conversion factor between that predicted impact and money and then compare that to the current price. This is not an unsolvable problem, but it's harder, and one should expect to have to pay more for it.
Finally, I would say that the engineering perspective leads me to believe that an impact market should be structured so that it can give different payouts to different project areas. For example, if the problem that funders want to solve doesn't have any remotely feasible solutions, a payout schedule closer to full capitalism might make the most sense. If the problem has several obvious solutions and the only real obstacle is funding, then limited payouts to market players would make sense. I could also see the need to move from one to the other over time in the same problemspace.
I think founders getting rich might actually be a GOOD thing. Employees at a startup are often granted stock or options that they can't sell for a period of time, in order to motivate them to work hard and make the startup succeed (addresses problem I-D).
Also I suspect it would be hard to stop them, because they can use tricks like buying their own shares through a cutout or lying about their costs or paying themselves a bigger salary. Maybe you can prevent all this, but there's a lot of different options that would need preventing, and there would be enforcement costs.
Also note that founders who get rich can re-donate their money if they want, so this isn't allocating surplus to founders who just want to be charitable, only to founders who became founders at least partly because they wanted the money.
I estimate a 90% chance someone already said this in another comment, but what the heck:
Here's a way to auction off a bunch of shares in something, without requiring one person to buy 100% of the shares:
Announce a window of time for the auction (perhaps a month, like a typical Kickstarter), the total number of shares (let's say 1000 as an example), and a minimum price per share (so that the assurance contract fails unless you earned enough to fund the project).
People submit bids of the form "I will buy up to X shares at $Y per share". (This is treated the same as X separate bids for 1 share each.)
At the end of the time window, the top 1000 bids each get to buy a share at the price of the 1001st-highest bid. (You don't want to punish people for bidding higher than necessary.) For example, if A wants to buy 100 shares at $20, and B wants 500 shares at $15, and C wants 800 shares at $10, then A gets 100, B gets 500, C gets 400, and everyone pays $10/share.
If there are less than 1000 bids, the assurance contract fails. If there are *exactly* 1000 bids, then they buy at the minimum-funding price.
Ties go to whoever bid first (because if you break them randomly, someone will submit a trillion bids to try and win ties). This still gives a little value to fast-movers, but capped at 1 bid increment per share.
(I don't know if this is a standard procedure, because I just made it all up, but it seemed to me like an obvious combination of various auction rules I've seen before.)
What's a recent example of a charity that got funding despite being uncertain at first and now definitely obviously has made an impact and you would totally leap at the possibility of getting the credit now if you could be a final oracular funder? (but the thing has now already been done)
How long did it take to become clear it was successful?
What kind of timescale would the initial investors be expected to wait before finding out they're going to get paid or not? Would this be upfront? What if it takes longer than expected?
I feel like i'm missing something - why would we believe either “Credit For Funding The Project” or “Full Credit For The Impact Of The Project” would work socially?
This is not true in any other markets I’m aware of; if you invest in an IPO you don’t get credit/status for funding the company or for its impact. Intuitively, I’m not sure why anyone would accept the idea that you can take credit for something that has already happened without your involvement. The closest example I can think of is NFTs, of which I remain skeptical for all the typical reasons.
Seems to me like the phrase "and everybody knows it" is doing a lot of work here, no? The point is that not everyone knows what's going to work, so there is a significant discount investors have to accept. For example, if a founder claims to need $1m to do a project that aims toward getting paid out by a $5m grant, investors need a 20% chance of success to simply break even in probabilistic terms, not to mention economic terms (cost of capital, liquidity premium, etc.). My guess is, the success rate is somewhere in the 1-5% range.
If you believe that to be true, the problem becomes, "how many actual projects are going to be funded through this mechanism?" i.e. what is the market clearing quantity? At 1-5% success range, we would expect a $1m project to require a $20m-100m grant. The number of matches seems vanishingly small, such that the median annual funded project seems to be 0.
If you believe that to be true, then I might propose that the contracts on these impact markets need to start smaller, higher frequency. For example, could there be an impact market of, say, $3,000 for someone to put on a classical music concert in their town square? This approach, if it works, makes measurement/verification easier and the system gets more reps. Obviously, the problem is that it doesn't immediately solve effective altruism goals. In the long run, though, if this approach is successful, you can imagine it leveling up into these bigger $5m Senegal-malaria outcomes.
Super late to the show here, but the whole discussion of the disadvantage in question 4 seems incorrect. Here’s a key quote: “But if we allowed founders to capture more of the surplus, then maybe the founder would end up with $50 billion, and the charity and its beneficiaries would only get $50 billion, which seems much worse for them.”
Maybe I’m misunderstanding how this whole thing works. I thought that the final oracular funder decides what the value is. That value is not dependent on the cost to produce it. (That’s the labor theory of value, which is discredited.) To make this more obvious, let’s say that the method for selling off impact certificates for completed projects in an auction. The project investors put together their evidence of completion and value, bidders get to review it, and the bugging happens. The supply of that particular impact certificate is fixed. Demand will set the price. If the auction comes out at $100B, so be it. If it comes out at $1, so be it.
More to the point, the beneficiaries already got all of the benefit of the project, whatever it is. The impact certificates are about who gets credit for producing the benefit, not who gets the benefit itself! Then, in the example quoted above, it’s very unclear who this charity is. Is this the final funder? They don’t get money at all. They are the ones who valued the project at $100B and paid that amount to buy the credit. Are they the intermediate investor, who funded the work in progress? If so, they knew that the funder was reserving half the equity. Just like VC firms, these intermediate funder will not want to invest in risky propositions where their upside is too low. That means that the market will place a practical cap on founder-retained shares.
I think the key thing is the mechanism for determining the value of the completed project. There are a lot of auction methods. One of them is probably best suited to that role.
"Otherwise, I’m not sure this is any worse than eg crypto, which already lets small investors lose all their money quickly."
That is horrifyingly low bar.
From mildly interested reader: it sound highly complex and I am not entirely sure why someone would invest and how risk of not getting paid is for people participating in projects.
Are they paid and they work and investors risk money? How investors can earn extra money to offset risk?
I'm a bit baffled by the extended discussion of how an investor could somehow own the credit for the good work – you point this out briefly early on but then in a later section it seems to be taken as possible that a 100% equity holder could be treated as having done whatever the charitable thing itself is (beyond merely having funded it). Surely that kind of social consideration arises from interactions like interviews and accounts of how the work was actually done, such that nobody would seriously conflate the two. At best I can imagine perhaps someone wants to give the charity founder an award and they contractually have to ask that it goes to the funder instead, but I think the audience at that awards ceremony would likely find it strange and awkward.
Sorry for being late to the party, but I fail to understand the central argument: why would anyone bother to buy the credit? Even without the premium this would hardly make any common sense.
In „normal“ markets patents, companies or brands are bought with an expectation to gain more profits with these later. It makes sense as an investment.
Products and services are bought as a means to an end. It makes sense as a purchase (or as an investment, too).
Buying the credit after the fact seems to me to make sense neither as an investment nor as a purchase. Just „bragging rights“?!
What am I missing?
Maybe you addressed this point and I simply missed it, but otherwise it seems as if you're glossing over what looks to me like as the most difficult problem with this idea: how does an "oracular funder" evaluates the outcome of a project, and assigns it a specific dollar value?
Without a well-designed, aligned, transparent and fair mechanism to produce such evaluations. everything falls apart.
I don't get the problem with the capitalistic option, if I'm a normal employee at a for profit company or a charity and I get paid the market rate, I still kinda get "the credit" for being a good worker, right? So why can't I get the credit for being a good founder while also getting the market rate money for it?
The incentivizing evil issue kinda broke the beauty of the idea for me... could it be solved somehow with fines? Like say the oracle funder requires that you publicly buy the shares. Then if the project turns out to do harm, they tell you to pay to offset this and you do it because...
1) they'll refuse paying you for other project shares and harm your career as a charity funder?
2) all oracle funders cooperate on 1) and super ruin your career as a charity funder?
3) other entities cooperate to not cooperate with you and be mean to you unless you pay the fine?
4) it turns into a dystopian court system where the oracles can randomly ruin your life?
idk but something like that
Super interesting. Re "whoever snaps up the shares first will get most of the surplus", two possible solutions
1) have an IPO window, and if it's oversubscribed, allocate by lottery. This has the advantage of being easy and the disadvantage that you under reward people for spotting the opportunity (though I think that's ok - when it's oversubscribed, there wasn't much value in spotting the opportunity)
2) have an IPO window with a more complex allocation eg if it's oversubscribed by a factor X, everyone gets 1/X of what they ask for. When the UK government sold the Royal Mail to retail investors about 10 years ago, I think everyone who applied got what they wanted unless you applied for over £10k worth in which case you got that amount
Happy to expand on either of these if helpful
Let me know when you've worked it out - happy to invest my money in impactful ways, but have been around funding long enough to be impatient and sceptical of new/innovative impact models & evaluation frameworks.
Regarding the auctioning off of the tokens: I think this could be done with a Liquidity Bootstrapping Pool [0] [1]. You would have to first sell some tokens personally (over the counter) and then you put the proceeds from that and the remaining tokens into the balancer pool, which then uses an automatic market maker mechanism to sell tokens at a fair market price until a previously set percentage of tokens has been sold.
[0]: https://docs.balancer.fi/products/balancer-pools/liquidity-bootstrapping-pools-lbps
[1]: https://docs.alchemist.wtf/copper/lbps/what-is-a-liquidity-bootstrapping-pool
Optimistically, it feels like a lot of these concerns will ultimately be circumvented by the market. In theory (ideal end-state, as you call it), you don't have to worry about who "should" be allowed to reap financial benefits, for example, because funders and project runners will simply decide for themselves which kind of operation they want.
It seems like a pilot could be run on KickStarter right now, to be honest, and would provide more concrete data for future investors/charities to reference.
I originally parsed "We don’t get the “coolness” boost of using crypto" as an obvious joke, but the following paragraphs appeared to suggest it was intended unironically. It's fascinating that bubbles exist even today where "NFT" and "crypto" signal the precise opposite of everything they do in my techy bubble.
Does the final oracular funder announce in advance what outcomes they're willing to fund, eg. we'll definitely pay $2,000 per QALY? Otherwise, the funder has a perverse incentive to only pay out in a fraction of cases (whenever they can find a flimsy justification) as by that point the good thing has already been done, so it's in their interest to keep their money as an incentive to further projects. This seems doubly so if these are being done for profit.
These also have the potential to become Utilitarian Indulgences - companies that want to rely on child labour or pollute a few rivers can buy these as a moral carbon offset.
>This post isn’t about the theory. It’s about the annoying implementation details. It may not be very interesting to people who are neither effective altruists nor institution design wonks, sorry.
I actually enjoyed this, for whatever it's worth, despite having no particular interest in EA (too poor) or institutional design. Any chapter of The Chronicles of Moloch is gonna be interesting reading; coordination problems seem like the fundamental difficulty of human history.
Also, now I wanna see a noir story about retrospective assassination funding...perhaps call it "Killer Prophets".
Interesting piece. Quick thoughts on this:
a) you end up with 5 million spend on a project that costs 1 million. Given that money for charitable work & great EA projects is limited, why is this not a major disadvantage?
All the other questions on how the 'extra' money gets distributed, result from this.
b) publication of project ideas or: how do make sure that ideas are not stolen or (deliberately or unconsciously) copied?
The *founder* has two functions: having a good idea/ designing a project (*invention*) and executing the project (*implementation*). Imagine my collegue has a brilliant, but - in retrospect - very simply idea on how to make projects 'raising the sanity waterline' more efficient. He designs such a project and wants to implement it, but he has only so much experience, and investors are hesitant to invest. CFAR and everybody else interested in 'raising the sanity waterline' read the idea, set up a project based on this, back it up by all their experience and clout in the field, and get the money. Good for the impact, but the inventor will probably know this, and not put out their idea. Again, with *very* specific things, you can have a norm of 'respect that person x wrote that first'. With a lot of great project ideas (which are often some type of improvements of existing things), it will be very difficult or close to impossible to distinguish between what was 'stolen' and what was 'we just thought about that obviously'. (Even more, people will suddently have this great idea when thinking about a project they are planning, and be totally unaware that they actually *read* this in a slightly different context some months ago.)
Solution: you need to make sure, that ideas / project designs are only available to investors but not to other potential founders.
Disadvantage: I have no idea how to make this work without contradicting other intended mechanisms.
Other solution: you need to reward *inventors* for their great project design, even if they're not getting the investors' money for implementation. Don't know yet how to do this in the context of this design.
Question: How does this work in VC? You don't put your business idea out on the www to see if it attracts capital.
c) I understand that you want all the elegance and mechanisms of a real market with lots of options. Looking at the difficulties and disadvantages, there could be intermediate solutions. EG: Founders put out their project to potential final oracular funders (for the moment a limited number of players known beforehand). A funder says: I find the idea to cure malaria in country x with the planned amount of 1 million USD great, if it works, I would be willing to pay 3 million for this. Now, all the projects that have a backing by a funder, go to the potential investors. Those project that find the necessary amount of investment, get funded.
d) The measurement. How do final oracular funders know how well the project worked out? There is *lots* of incentive to distort this in the reports. (When this whole thing grows bigger; hopefully less so in your next rounds of grants.)
Related: Most projects I know of don't fail on the account of the few measurable indicators they commit to. They fail on everything else. You mentioned this, but I think it's underaccounted for in the reflections on the set-up. Maybe EA has long solved this, and I'm unaware.
e) re 2. In the charity/grants/projects worlds I know, funders would take all the credit for *funding* the greatest projects, and organizations/firms/... would take all the credit for *implementing* the greatest project. (In fact for *designing* and *implementing*.) Makes sense to me, because we're talking about two different tasks.
When the EU funds a project, the 'EU funded' is plastered on each infrastructure project built, each tractor bought or each scientific article published. The firm that built the infrastructure will claim 100% of credit for carrying out this magnificent (EU-funded) building or scientists x for doing that magnificient research. For both it's important.
I guess this goes in direction of 2B (without having read Ben H. article). I find the 'sells all the credit' or 'sells part of the credit' is unnessarily complicating things. I think those latter might be unproblematic or even attractive in certain circles (VC/rat.adj./ EA? ...), but I would be worried it might have a chilling effect when selling the method to many other groups of people, including those usually engaged in non-profit projects.
Admitted, I 'cured malaria' sounds better than 'I funded that malaria was cured' – but I think it still sounds pretty good ;). You could partly solve this by selling 'cure malaria (funding)' certificates and when you're talking about this as 'I cured malaria', everybody at the party would know you're the funder, not the one implementing this. Or vice versa.
I am skeptical about this working - but one useful part for prediction markets or similar would be ability to predict that attempt X will fail (or succeed).
This may provide some help against blatant scams and failures.
Though overall, this sounds like an interesting project with many traps, up to SEC appearing and ending fun or maybe even criminal case for someone (is there prediction market for that already? :) ).
On issue Number 9B - my experience of the UK pharma regulatory context is that when there's enough money on the line people can get quite argumentative about exactly how their intervention is assessed. Quite a lot of this is that it is really hard - probably impossible - to say "This intervention saved X lives for sure" because it will always involve more-or-less reasonable assumptions (for example that, counterfactually, everyone wouldn't have spontaneously killed themselves as a protest against non-intervention, for example. Or maybe more reasonably, what the correct statistical approach is to participants in the trial who deblind themselves somehow or cross over trial arms). So you end up saying, "Our best guess is that we saved Y lives" and then defending your decision from criticism from the people with a large financial stake in Y being as high as possible.
The rough cost of the regulators defending one pharmaceutical submission in the UK is £143,000. This is an overestimate of how much the oracle would spend to do exactly the same thing as NICE, because large government agencies are inefficient, but this is also an underestimate of the true cost because the pharma companies do a lot of the work for the regulator; for example the company always funds the literature review and development of an economic model which are close to £100k each. My guess is that maybe the total cost of a full assessment including salaries is probably fairly close to £500,000(ish). Also it is unclear to me who funds the trial to assess effectiveness, but this could be hundreds of millions of pounds if you want to RCT a big intervention (and if you don't want to RCT the intervention you introduce a lot more points to argue about).
Basically, I'm a bit unclear how the 'guard labour' costs of the oracular funder are accounted for. These costs are non-trivial and it probably wouldn't be suitable to use the current EA method of well-meaning people giving it their best shot to assess outcomes even if there are enough well-meaning EAs around to make this possible, because if the project works well there will be a whole bunch of monied interests attacking your decisions.
I don't see any problems with founders getting rich. If that's a persistent problem, more smart people will flock to the charitable sector. Thus driving down excess returns for founders and also causing more charity to happen.
About investors losing all their money:
I would imagine people mind want to buy index fund equivalents instead of putting all their eggs in one basket?
Why try to make philanthropy more efficient? Philanthropy is on net socially detrimental. Donors are at the heart of the culture way (e.g. George Soros). They are not accountable to the people that they are allegedly helping. I will elaborate on my substack in a few days.
About the auctions; it's relatively easy to auction off a limited supply of identical goods
See eg how Singapore's Certificate of Entitlement works (that you need to operate a car in the city).
Basically, everyone submits at many sealed bids as they like. The top n bids win and pay an epsilon more than the n+1th bid.
I don't really understand why you have a problem with giving founders a piece of the impact shares? If you're building a system where you pay $5 million after the fact to a project that only needed $1 million to succeed, you're already paying $4 million dollars more than usual, solely for the purpose of promoting risk taking and innovation. Whether that $4 million extra goes primarily towards the founder or the funder seems entirely like a question of who needs to be incentivized more. While it does seem to me that funders need to be incentivized more, it also seems like it would be beneficial if charity founders were properly incentivized for success, such that they actually get a significant personal benefit iff the project succeeds.
Also a minor note on crypto, there are many obvious disadvantages to working in it, but one true advantage is that Crypto is great at dealing with international economics, especially when dealing with something as legally complicated as securities, or when dealing with third world countries.
I got lost at this point: "Foundations see that I have done a great thing with $5 million in benefits, so they give me $5 million."
How does that work exactly? Even just firming up the $5m claimed value seems like a process that is much easier said than done. Why does the foundation then just give that money?
Very interesting, but contains too many assumptions, too many details, and is too long for me to do more than scan. That said, here’s one potential problem and solution.
If I’m the final funder (oracular?), I might not put any money into this! I mean, hey, the good results I want already exist! Maybe I’ll throw them a token, but better to save my money for something else.
So I suggest “results puts”. The initial investors come up with metrics (or use metrics from the founders), and sell results puts, such that anyone who buys a put must pay if the metric is met. Then they conditionally buy certificates from the founders, and the sale executes if they are able to sell the put.
Two points:
1) A project selling off 100% of the credit removes incentives for the project to go well (in some idealized sense, anyway). That’s why VCs don’t buy 100% of a startup’s equity.
2) Auctions seem to solve the “fast people win” problem reasonably well. Everyone puts in orders like “I want to buy X amount and Y price”. Then after some time, you resolve the auction by picking a price, perhaps the highest price so that at least Z dollars of funding will be raised. Everyone who offered to buy at >= that price gets shares at that price. (This is roughly how “real” auctions in financial markets work, except that in that context there are both buyers and sellers and the price is chosen to maximize the number of shares traded).
One thing I am not clear on /don't think I saw addressed: What happens when an additional participant (s) tries to muscle into the same space?
Oracular Funder pledges to buy 5m worth of Malaria being cured.
Alice sells 1m in Malaria certificates and goes off to start curing Malaria. Bob thinks he has a better solution to cure Malaria and tries to raise 1m to go cure Malaria his way.
If Bob also gets funded, he and Alice can both work simultaneously. This doubled the vc funding for malaria, but may half the potential return, even for Alice's investors who are already in. If it doesn't half the potential return, then who gets the credit? Both Alice and Bob were actively working to cure Malaria, and it has been cured-seems very challenging to figure out which of the two get credit.
Beyond that, what if Bob sees this coming, and decides not to bother because Alice has already started a project aimed at Malaria, and there's no more funding to solve Malaria EVEN THOUGH ALICE HAS NOT YET ACTUALLY DONE ANYTHING. He then goes to work at a hedge fund instead.
Isn't there a risk that when oracular funders have presold a token (or any other form of pre-registration) saying "the owner of this token gets credit for Malaria being cured" there is less incentive for later arrivals to bother trying to solve Malaria?
I am one of those random Google engineers you mention. This idea sounded extremely cool when I read the first half of the post on the train on the way to dinner. I even spent some time enthusiastically telling my wife about it while we ate. But then I read the rest of the post on the train back, and in particular got to point 6.
If this project involves crypto in any way, it immediately becomes anathema to me. In fact, I am willing to pledge that I will invest $10K+ of money in this sort of market within six months of its launch, if and only if it does not involve crypto. I know this sum is not a serious consideration when weighed against all the potential money coming from crypto people, but I write it down to emphasize that this is not empty virtue signaling, and is something I believe strongly enough to put real stakes behind.
Crypto is too much the domain of scammers and speculators. It is a solution very much in search of a problem, and sacrificing this interesting idea on the altar of trying to give it another shot at adding value to the world is ... unfortunate. Any association of this project with crypto brings the project into the realm of the sorts I do not want to associate with, or give any financial backing or societal legitimacy to.
I don't think this opinion of crypto is very tribal. It seems shared by the majority of my fellow gray-tribe software engineers, at least the ones that I know in real life. I think you might want to gather some more data before assuming that crypto = cool and therefore that grafting it onto this project will be a net positive. E.g. NFTs and NFT-holders are actively mocked by my peer group, so if this project got a whiff of NFT-ness near it then I would be incentivized to avoid participating in, or even discussing, impact markets just to avoid the social disgrace of being seen as an NFT-head.
(Secondarily, I don't know too much about the environmental concerns, but my suspicion is that you're blowing past them too easily---to my outsider knowledge, the only serious attempt to avoid the environmental impacts of current blockchains was Chia, which immediately had other negative externalities on the hard drive market.)
I will admit that if this project turns out to use crypto and nevertheless be wildly successful at impacting the world positively, then after a few years I will hold my nose and participate eventually. But if you want the participation of people like myself during the early bootstrapping stages, your choice of whether to align yourself with the crypto ecosystem or not will be important.
I hope this perspective is useful!
1. Sometimes it's really hard to measure impact even after the fact. There is still controversy over whether the New Deal shortened or prolonged the great depression. There is still controversy over what caused the rise in US homicides in the 60s/70s and what caused the decline in the 90s. But at least determining impact after the fact is easier than predicting it.
2. Many investors will be scared away by excessive uncertainty around the FOF payout unless you can get the FOF to contractually precommit "I will pay $X to whoever causes outcome Y"
3. There is a risk of people getting paid for placebo interventions when Y was going to happen anyway.
4. The “Maximum Capitalism” Design may not be that bad because competition between founders to do the same thing incentivizes them to split the surplus with others. If Alice was going to sell 50% equity for 1 million, Bob can undercut her by selling 75% equity for 1 million. In a competitive market for this the founder's surplus will tend to converge on their actual value added.
7 seems easy to solve. The founder submits the amount of cash they need. The market organizers create and auction off the shares. The market organizers give the founder the amount of cash they need, keeping the excess. The excess is used to subsidize the final oracular funders for the entire market.
So imagine 4 projects:
Project 1: Costs $1, IPO worth $2, Final value is $2
Project 2: Costs $1, IPO worth $1, Final value is $5
Project 3: Costs $1, IPO worth $3, Final value is $1
The market auctions off all three projects, using $3 (1 + 1 + 1) to fund the projects and gaining $5 (2 + 1 + 3) as excess. Much later, the final oracles would like to use $8 (2 + 5 + 1) compensating all the projects for their charity, but by spending from the excess they only need to spend $3.
This diverts the riches of "cheap projects" from both the founders and the investors towards funding more charity.
I'd be interested in selling impact certificates.
AISafety.com is currently investigating an unlikely hypothesis about the ability of language models to reason strategically at higher meta-levels. This research may be well suited for being funded through impact certificates:
1) The plausibility of the hypothesis depends on a substantial amount of illegible reasoning.
2) The hypothesis is unlikely to be true, but if it is true the impact could be great and legible.
3) We are currently exclusively funded by my personal savings, but we could move faster with more funding.
--Søren Elverlin (I have been running the AISafety.com Reading Group since 2016)
On #7, there's a third alternative: bid on the amount of the prize you'll rebate back to the oracular funder. In the common knowledge $1 million / $5 million example, the auction should clear at something a little bit less than an 80% rebate and the investors would get around a risk-free rate of return on their funds. If the auction can't clear at even a 0% rebate, then the assurance contingency isn't met.
When running the auction, investors could submit different bids at different rebate levels. I might bid the full $1 million at a 0% rebate but a smaller amount at a 75% rebate.
You can't give the founders money and call this a non-profit, or without embezzlement issues. A foundation gives your company an award for charity and you put it in your personal account? You sell stakes in the future non-charity-restricted income that your "non-profit" might have, and that's not stock?
Here I have a great idea for a non-profit charitable market, I will say that I have $1b of funding available to fund a next-generation fighter jet and let the free market decide on proposals and implementations of the fighter jet. Then I will pick what I judge to be "the most impactful" jet, and distribute the $1b accordingly to their founders and capital supporters. It's not for profit though because, like, it's a fighter jet hello that's charity?
In addition to VC, I think there will soon be more traditional funding available for this type of thing. ESG (environmental, social, governance) is the hot new thing on Wall Street and it doesn’t look like it’s going away. Investment funds and most companies are playing along. The SEC recently proposed a rule that investment firms and investment managers must provide ESG guidance.
ESG is also 90% fraudulent and the ratings are based on ridiculous metrics, if not purchased outright. Being able to purchase the credit for charitable actions will be like candy for those who need a bump in scores.
This sets up competition between charities in a way we currently don't have and which will need addressing.
Under a prospective model, if two charities independently set out to eliminate malaria in Senegal, and five years later it's eliminated, they rejoice in their shared victory. *Maybe* some of the funders grumble about getting less-than-expected bang-for-buck, though no single funder is certain they did (maybe the funders of the *other* project wasted their money -- we'll never know!).
But with an impact market, a retrospective funder whose promised $5M now has to decide whom to give it to. Each set of investors considers themselves entitled to the whole $5M ("We said we'd eliminate malaria and it's eliminated -- what more could you want?").
For-profit investing also worries about competition, but in that case it tends to be pretty easy to tell who each customer bought from. Capable firms shadow the fields they work in, preventing new growth where it isn't needed, but loud-mouth firms don't shadow fields in which they can't deliver value.
The whole idea of selling "impact" itself seems both complicated and philosophically dubious. It seems to me it would be much better just to set up one system where altruists can promise to pay out prizes to whoever accomplishes some worthwhile goal, and then a market where groups working toward the goal can sell future shares of the prize money. For example, altruists publish that they'll pay a $5 million prize to whoever cures malaria in Senegal. If a group of biologists think they can cure it for $2.5 million, they can sell $2.5 million tokens for $1 each that pay out $2 each after they collect the prize. There's some issue with determining the exact conditions when the prizes pay out, but the prize pools just need to have arbiters who cultivate a reputation for fairness. That's no worse problem than exists for future oracular funders.
For any "startup" like idea, I think you either need to:
- Talk to the potential users ("investors", "founders", ...), do the mom test
- Just try stuff
I really appreciate how in depth your analysis is, but you'll never be able to go through the nitty gritty by thinking about it. Your grants program was a great example of this: many unforeseen things came up.
I'm also very excited to see where this goes. Please keep us posted!
(apologies if a comment like this ahas already been posted, I browsed through a few of the comments but didn't see anything like this)
I am a tax/estate planning attorney. Exciting topic. I wanted to raise two thoughts—enhancing impact certificates with legislative change and addressing some of the issues from your implementations. No legal advice here, just wanted to share some ideas.
*****Impact Certificates with Legislative Change?*****
First, I note your (Scott's) solutions generally assume no legislative change. But imagine a legal system where impact certificates replace our current charitable giving structure, and can lead to broad government decentralization. Here's an example:
- Like the USPTO, tax-exempt entities apply and get approved a “project patent” from a government agency. The design of this agency, how its examiners are paid, legal doctrines around the project patent (and what needs to be shown to get one), methods (and incentives) for others to challenge an issued patent or require its update, ability of other parties to use an already-issued patent, and a designated project patent court are significant in getting this to work.
- On successful hitting of KPIs specified in the patent, the tax-exempts receive a defined amount of government cash and an assignable tax deduction. Payouts are for much less than 100% of benefit; for projects where benefits were tough to quantify, cost savings over government or other approaches for like benefits could be a relevant metric. Project patents could continue indefinitely, if they are so written (subject to the need for updating).
- The tax-exempt projects then repay (via a debt or equity instrument) funders with some combination of the government cash and deductions (which can be assigned to the investors). The deductions received by investors are forced to be carried into future years if the investor does not meet certain requirements. Such requirements would essentially aim to prevent rich people from zeroing out their taxes using these investments, unless their investments really do represent a replacement for the whole of government. This would include some limit to the deduction tied to the government’s need for taxes to service its debt.
There really is a lot of government that could be rerouted to this system, not just charity. Also, I thought the design could be part of a grand bargain to help with raising taxes. (IMO, the best objection to raising taxes for greater government services in the US is our generally poor governmental performance.) Raised taxes, of course, increase the attractiveness of the deductions generated.
*****Legal and Other Design Issues*****
Second, to address some legal and other design issues raised by your post.
- Have you reviewed the literature on (social) impact bonds? They already have designs aiming to address some of the issues you raise, including quantifying the return. I assume so, but thought I would note it.
- I thought it was interesting “fractionalized impact shares with assurance contract” essentially follows the capital contribution/capital call structure already present in investment funds, except it’s the project itself doing the offering.
- There are regulations for crowdfunding which can somewhat sidestep the accredited investor rules. In general, they're outside my knowledge, but I understand they’re perceived as a bit too burdensome on companies.
- People wanting tax deductions instead of investment returns (or who are not accredited investors but just want to help) might give to certain §501(c)(3) entities. The §501(c)(3)s might then invest into the oracular funder's…funds (set up as a partnership in a standard VC-type structure). Via one legal structure or another, the funds receive a small service/guarantor fee from each project it investigates and agrees to underwrite the successes of. One could consider whether the §501(c)(3) investments into the funder's funds are program-related investments (essentially, investments where a significant part of the “return” is achieving charitable goals, which are recognized under the tax law).
- There are indeed private benefit/§501(c)(3) concerns for investing in an oracular funder (and inurement or various excise taxes, if the §501(c)(3) insiders are involved with a funded project). The private benefit angle appears well-explored in regards to social impact bonds in http://fordhamlawreview.org/wp-content/uploads/assets/pdfs/Vol_81/Dagher_May.pdf. My initial view, not having run it to ground at all, is that §501(c)(3)s investing in this area should generally be segregated from other §501(c)(3)s to avoid putting more charitable tax benefits at risk than is necessary. Coming up with a good position here could take some thought.
- I am not sure I understand your concerns with governance issues, at least to the extent you're indicating there are special problems here—if the impact certificates are, e.g., interests in an LLC, it would be easy to build in state governance rights that already exist, and to model it on corporate law. Further, requiring periodic disclosure would be easy to require in the LLC agreement. No new law needed.
- There are legal rules governing payment processors, which is maybe your vision for an impact marketplace in part? Not sure I understand all the functions you see the marketplace as performing. Tax-wise, payment processors are often considered agents, so they only realize their fees as income; some designs could cause the payment to be income to the marketplace, and the payout to be a deduction. There would be tax information reporting issues, though (think 1099s).
Thanks for this incredibly comprehensive overview!
I’d like to use this chance to position our GoodX project “Impact Markets” (https://impactmarkets.io/) in this context. (Thanks for linking to our MVP post!) I’m trying to write this in a way that is also understandable to people who haven’t thought about impact markets much, so sorry in advance if some of it is unnecessarily verbose.
I don’t know what the etiquette is around cross-posting comments, so rather than copy-pasting the comment from the EA Forum, here is a link: https://forum.effectivealtruism.org/posts/6LppWMdN2NLHceGTr/impact-markets-the-annoying-details?commentId=gzXwHDXHedREWTjau.
Very long time lurker finally brought out from the woodwork - this is basically social impact bonds with all their current issues, with extra twists!
There is an additional subtle Goodharting issue going on in SIBs, beyond the mechanism design you propose - projects will be shifted to issues that have a clear evaluation strategy. For example, if you fund projects reducing homelessness, how will the funder know it was the impact of the charity, instead of a change in unobserved city policies, that led to the result you see?
Let's assume that the funder requires an RCT or a proper impact evaluation. In first place, this rules out any setting where a clear evaluation is difficult, such as large scale projects or settings with general equilibrium issues. Assuming that this is avoided and projects stick to easily evaluated outcomes with clear identification strategies, now everyone has incentives to pull the crappiest study that supports the result they need to get paid, or take every single potshot at the funder's evaluation. This enhanced data interrogation technique is already bad with charities under normal funding schemes, btw - now bringing in capital is going to raise the stakes immensely. Imagine that the outcome of your token/bond/share depends on whether someone clustered their standard errors correctly - will there be a court ruling determining that randomization inference is better than clustering, and ordering someone to pay up based on that?
Personal anecdote - I was once an RA involved in writing a grant for the evaluation of a social impact bond in Colombia. The program was basically impossible to evaluate properly, for starters, but we came up with a decent instrumental variable that would allow us to at least calculate the effect size bounds. The funder did not like that, because they would look bad if they funded anything that ended up having zero impact, or even looked ambiguous. They eventually went with a before/after design (not even DiD!) that would be laughed out of any intro to econometrics class.
I think this is a very interesting problem, one where you need to adapt old and known mechanisms - markets - to a new purpose. It is important to keep the good elements of the original mechanism and, I think, you are dropping some of them.
2. It is not clear why monetary mechanism design has to correlate with "credit". They guy who invents the cure for malaria gets the glory and respect for the invention; they guy who funded gets his money and, possibly, some glory and respect for funding it. Money and glory are complementary benefits here, there is no reason to try to design a mechanism to ensure greater correlation of money and glory. Since different people put different relative value on money vs glory, it is better for everyone to give more glory to those who value it most and more money to those who value that most.
You note that although the inventor sold 100% of monetary benefits to the funder, he still might derive some monetary benefits (e.g. Harvard tenure) from his glory. This is great, but why does it follow that the inventor should retain more equity and get a higher monetary payoff? I would argue that the opposite should be true: since the inventor gets some monetary benefits as derivatives of his glory benefits, marginal utility of additional monetary benefits to the inventor is lower, so retaining equity makes less sense than in a world where the inventor is unlikely to get Harvard tenure.
3. "I am nervous about this because I want to be able to pitch impact certificates to non-savvy people who may not be able to make good decisions about equity." In financial markets this is called "misselling". It is often a criminal offence. Do not do that. The whole point of these markets is to mobilise insights from people who have the incentive to understand the problem. People who do not understand funding instruments are not going to have very good insights because they do not understand their own incentives.
4 (and also 7) "If we don’t let founders get rich, it shifts the surplus to fast investors who may not have added any value - that is, to the first people who snapped up the obviously-underpriced certificates." This is not necessarily true. The primary placement of certificates may be done the way bonds are placed through an auction: bidders submits bids with a quantity and the smallest acceptable success contingent yield, the final oracular funder (FOF) looks at the threshold yield Y which is the lowest yield at which they can raise the required amount, and then all bids with yields below Y get certificates with the yield Y. So in a reasonably liquid market for an obviously successfull project, Y will be close to the risk free rate for the term and all benefits accrue to the final oracular funder (FOF).
An example: let's say the final oracle funder wants people to give $1m for malaria cure now. They announce the auction, and there are many potential investors, from very skeptical/greedy ones who are ready to fund £100k at yield of 900% over 5 years (x10) to very confident/altruistic ones, ready to fund $200k at yield of 0%. The order book might look like this:
bid amount yield
A $100k 900%
B $700k 400%
C $300k 250%
D $400k 100%
E $600k 20%
F $200k 0%
FOF starts from the bottom of the book. For $1m to raise, the threshold yield Y is 100%. Bids E and F are filled in full, bid D is filled 50%, these three bidders get certificates with success contingent yield of 100%, other bids are unfilled. The funder commits to pay a return of 100% on $1m, i.e. they will pay $2m in case of project success. If the threshold yield is too high for the final oracle, the auction fails.
5B. The same yield auction design resolves the issues with committed pot of money. The primary issuance is done through an auction where the success contingent yields are fixed. An investor has the option to hold the certificate to maturity and receive the fixed yield. If there are other auctions at later dates, the participants there compete for other tokens and do not affect the hold to maturity yield of earlier issues.
6B. The coolness halo of crypto is dissipating quickly. We already see NFT-like projects carefully avoiding the word "NFT" in their materials (reddit avatar.) There may be genuine benefits of using e.g. a public ledger (some glory for the funders?) or other elements of crypto/blockchains.
9. In case of auction design as above, this question has a simple answer. FOF decides how big prize are they ready to give out in case of success, keep this information secret, and run the auction to raise a public required amount ($1m). If the threshold yield is too high and implies a higher prize than the FOF is ready to give, the auction fails as there are not enough optimistic investors. For non-binary outcomes, certificate format can be adjusted to e.g. pay off $1 per life saved capped at $2m and a similar price auction can be used for primary placement.
Appendix II A. This is what insurance is for. You will need to raise $1.2m and pay $200k for an insurance from the bad outcome. The insurance may be a regulatory requirement in any case for many projects. The insurer is likely to impose some controls on the developer.
Another example of A. Simple Retroactive Funding...
Bug bounty programs for vulnerable websites. There's money offered but you don't know how much until after the fact
Minor technical note on 7B: auctions for fungible shares are called Dutch auctions; they are a solved problem and work very well: https://corporatefinanceinstitute.com/resources/knowledge/finance/dutch-auction/
Are final oracular funders actually required to be final, or are certificates perpetually transferrable like stocks?
Some consequences of allowing resale:
* Investors could buy underpriced certificates for completed interventions. E.g. say the malaria-curing project is complete and its certificates have been sold for $5M based on the lives it saved. But you find evidence it actually had another $5M in indirect economic benefits. You could buy the shares , publish your evidence, and sell for a profit.
* Individual funders who ran short on money—or just have a change of heart and decide to be selfish after all—could sell off their certificate portfolio. This is kind of weird to think about, but it might lead to a lot more investment in impact certificates.
Ok maybe I'm missing something very basic here, but when Scott says some charitable intervention will produce $5 million in benefits what exactly is that supposed to mean? Does it just mean that some rich person/charity will pay $5 million if the charitable result is achieved?
> Accidentally Encouraging High-Risk Negative-In-Expectation Projects
This circumstance is far more common than the EA community realizes, once you take into account the second-order effects even of projects whose outcome is deemed "successful."
I think issue 5B (Committed Pot of Money) is going to be key to making this work. Following your example, suppose an investor invests $1M, and funds a project that cures malaria in Senegal. Why is the result now worth $5M, or any value at all? Who figures out what it's worth? Supposedly a philanthropist will "buy" the solution for $5M. But if I am a philanthropist, why would I invest any money at all in problems that have already been solved? I would get more bang for my buck by funding problems that *haven't* been solved yet, so the "certificate" of "congrats, you helped cure malaria" will be worth exactly $0 to the original (for-profit) investor. For-profit investors know this, and will never invest.
What you need is to have something that resembles an old-school charitable foundation. Each year, it collects contributions from donors, both large and small. There is no limit on the amount of donations that it can collect, so this is scalable. At the end of each the year, the foundation puts all of the donations for the year into a "prize fund", and issues tokens equal to the size of the fund.
Over the course of the year, the foundation also collects various project proposals, each of which has a cost estimate (a grant amount). There is no limit on the number of proposals, or their dollar value, but the foundation will only fund proposals up to size of the "prize fund".
Once the prize fund is in place, for-profit investors can purchase a token (at an offer price of $1/token), and must name the proposal that they wish the money to go to. Once purchased, the token becomes a "named token", and is a bet on that particular proposal. Tokens for proposals that don't get funded, because they don't meet minimum funding requirements, (or for some other reason), will revert to being "blank" again, and can be redeemed for $1, or redirected to another proposal. Some sort of ranked-choice voting could make this process more automatic. The money from token sales is used to fund the proposals.
After a certain "period of performance", perhaps a few years, the funded projects are scored/graded by some impartial observer, and the prize fund is then used to buy back all of the the tokens, at some price proportional to their score. Tokens thus function like a prediction market; they are a bet on how the project will ultimately be judged. It is important that the observer be named in advance, and have no conflicts of interest, otherwise for-profit investors could game the system. The charitable donors themselves could vote on what constitutes success; this allows for-profit investors to game the system, but only by making charitable donations.
There will be some additional complexity if the "period of performance" differs from project to project (as is likely to occur); some kind of token rollover would need to be performed each year.
As a sweetener to the deal, the original "prize fund" should be invested in low-risk instruments like government bonds, so that the expected rate of return for purchasing a token is positive, and equal to the expected return of the underlying instruments.
Now there's a reason for donors/philanthropists to invest in the system: they know that each donation will be used to fund a project, but the foundation uses a prediction market, backed by for-profit investors, to decide which projects should be funded.
Dumb question: If the founder keeps some equity, how does that take away value from the beneficiaries?
Using the malaria example again, suppose curing malaria takes $1 million, and produces $5 million in value. The founder sells 50% of the equity at the beginning, keeping the other 50%, and then gains 2.5 million from a big charity house at the end of the project. However, the $5 million in value produced for beneficiaries of the malaria treatment still remains with them, right? How does that go to the founder?
I think we should be clear about what we're paying for with inefficiencies introduced by this system, and why we're willing to pay for that.
For a normal investment market, you are paying for starting capital because you don't have that capital in pocket.
In this case, what you're paying for is risk-reduction on the bet that a group with a proposal for fixing a particular problem will be successful. So basically, buying insurance. Or, 'outsourcing' the work of deciding which grants to fund.
Also, you are trying to pay potential founders for searching for cost efficient, novel, and likely-to-succeed proposals.
Then this gets entangled with trying to implement those proposals. The founder must chose a team (often starting with choosing someone to help them choose others). Then you make some initial attempts at starting on the proposal.
Each of these steps may uncover important information.
Did the founder manage to assemble a competent team? If so, they're much more likely to succeed at their aims even if their initial plan wasn't great and will need substantial revisions as the project matures.
Did the initial attempts show up unexpected flaws in the initial plan? If so, was the team able to come up with a new and improved plan to remedy this?
The nice thing about having rounds of funding is that you can seperate out some of these issues. Maybe you can give grants like, "We'll give you 1 year of funding to get a team together and get started on this 5 year project. At the end of year 1, we'll decide whether to extend the grant to the full 5 year funding or cut our losses and tell you that we no longer have faith in your ability to implement the project within the proposed budget."
Alternatively, what about having people commit to joining the team if the funding is achieved? And then making a grant proposal where the proposed team is part of the proposal? Maybe the grant only goes through if you really do get at least 80% of the promised team to commit?
One of the grant problems I faced in academia was that to write a likely-to-be-funded grant proposal (in my field anyway), you needed some initial show of backing for your idea. You needed to be an already established researcher with a lab, or you needed someone with an established lab to believe in your project and grant you some resources 'stolen' from other projects to get your initial data for your grant proposal. This creates a huge status quo bias where established labs are much more likely to win grants. Also, the grants tend to be much more focused on boring incremental progress with high likelihood of success, then on trying radical new ideas with low success likelihood but high potential payoff.
The start-up venture capital culture of taking a bet on a variety of different not-previously-funded groups of ambitious newcomers is a refreshing change. I think having some of that style of funding mixed in with more academic grant style funding would be great.
In the case of Open Phil and Future Fund, they've been doing a pretty good job (I think) of keeping the balance between funding established groups vs newcomers.
It makes sense that funders like OpenPhil and Future Fund would want to be able to 'outsource' or 'insure' their granting decisions. I don't necessarily think this is a fatally flawed way to do that, but I think it hides some of the intentions and costs in a problematic way, and sets up some perverse incentives (makes Goodharting or impact-assessment-manipulation extra tempting).
I think you could have a basically equivalent system that was more clear about what it was doing. Seperate out some of these goals into more clearly delineated systems.
What about starting a 'charity insurance' group who would do their own independant grant evaluation and let grant givers buy insurance at a negotiated rate against the failure of the project? That would take the whole 'founder gets rich' question out of the equation. A separation of concerns.
I do think that founders being able to make extra money, above and beyond a normal paycheck, if they have an unusually good idea and assemble a crack team and guide this team to overwhelming success. If unexpectedly great good gets done in the world as a result of this, shouldn't there be some payback for the founder and team? Probably at a less impressive rate than if it were a capitalist venture, and with some risk-mitigation... but still. That spirit of 'we might all get rich in the pursuit of this awesome but hard thing we're trying to do together' is pretty darn motivating for teams.
Or another example of trying to seperate out the concerns, what if you were offering a prediction market on whether a particular potential grant would end up achieving its aims? The market would close before any grants were given, and the probabilities would be used to guide the eventual grant-making decision. Some time later, when the success of the funded projects is retroactively assessed by the funding agency, the prediction market pays out to the ones who bet correctly.
In terms of flaws of the original retroactive funding idea, I agree with Daniel Speyer and Jacob (Writes Streams of Consciousness) who bring up the issue of 'concern squatting' in their comments here. This arises if there's a particular problem that multiple people want to address, but retroactively it would be hard to appropriately determine who gets the credit for fixing the problem if the problem gets fixed. In the retroactive funding model, this disincentivizes multiple groups from trying to fix this issue at the same time. This is a very different dynamic to the idea that two startups might try to race for control of the same market space. Having charities race to fix a problem sounds like it'd be potentially quite good for the beneficiaries. The idealized healthy competition of an open and volatile capital market space benefiting consumers, ported over to the charity world. So yeah, we want something to help us get some of the good aspects of capitalism into the space, but I don't think this proposal for retroactive funding quite does it.
Re: Should Founders Be Getting Rich Off Impact Certificates?
One possible answer here is "pay less money for completed projects." Every month, the final funder announces a funding multiple between 1 and 0. For all projects that get funded that month, when the impact is finally assessed, the impact is multiplied by that funding multiple. Suppose the malaria project was originally funded during a month where the funding multiple was 25%. When impact is assessed at $5 million, the payout to investors is $1.25 million.
If the final funder announces a funding multiple of 25%, then they're essentially saying that investors shouldn't fund anything below a 1 to 4 return in expectation. But you can tune this value: if you have too many good projects and too little funding, you can lower the funding multiple and be more selective. If you have too few projects getting funding, you can raise the funding multiple.
Note that this is different from proposal 11C: that project is equivalent to a funding multiple of 21%. But in 11C, you're allowing the funding multiple to vary between different projects. If you have one project which is efficient, then maybe 20% of its shares get bought for the startup price. If you have a project which is inefficient, then maybe 40% of its shares get bought for the startup price. But both projects ultimately get funded, and both investors make similar amounts of money, despite the fundamentals of one project being worse.
I find this idea super exciting. It's essentially posing a market design problem, which is my broader field of research (I'm a computer scientist / economist).
However, I think some of the basics here are very fuzzy. The post mentions principal-agent problems (between funder+investor as principals, vs founder as agent. This is a pretty standard problem, no different than funding a startup, or hiring someone for a job. We can align incentives by giving the founder a share of the profit.
But what about the alignment between the funder and the investor? I don't see at all how an investor would ever trust the funder over the span of a multi-year project based on nothing but reputation. Charities, just like any organization, are subject to value drift and their own internal principal-agent problems.
What's missing is a way for the funder to make a promise at least as strong as a legally binding contract. As much as I don't like crypto, this is the exact type of problem that smart contracts are supposed to address. But this reduces the possible scope to very legible outcomes only.
Another solution might be a centralized market maker that creates a two sided market between funders and investors. This reduces the trust problem to a single entity. The way I envision it, the funders would advertise their project through the market maker, who holds their money in escrow and acts as an arbiter, making the final, binding call for any impact calculations. The market maker will gain a track record of making fair decisions much more quickly than any individual funder. They would take a small fee off of every transaction, which incentivizes them to build and keep trust over time.
(This kind of mediating entity already exists in other contexts: e.g. banks facilitate international trade by acting as a trusted third party for a fee. If I want to sell a container of malaria nets to Scott, they way this works is that Scott gives his bank money in escrow, I send the container, and the bank gives the money to me as soon as they make sure that I actually sent a container full of malaria nets. Banks have a very strong incentive to keep this scheme going as they're skimming around 3% off the top for doing essentially nothing. Long-term business partners will usually deal directly with each other, but that requires a good deal of trust which is surprisingly hard to obtain over international borders.)
"If no, I can’t think of a good solution to the “fast people get rich” problem."
One solution would be an auction that sells a variable number of shares for a fixed price.
If a project will obviously do $5 million of good, and costs $1 million, then the founders keep 80% of the impact as meaningless tokens they can't sell.
Another solution is that the extra money gets sent straight to a generic funding body. Say funders can choose where the extra money goes, and that must be an actual charity, and the excess money gets split proportionally, and the money is a perfect gift, the funders get no extra impact certificates.
A third solution would be that the money is transformed into a token. The token can only be redemed through the fund charity, receive impact mechanism.
Suppose the market rate for such tokens is 50c/$. Bidders fund the $1 million in, $5 million out charity to the tune of $9 million. $1 goes to the actual charity. $8 buys tokens, which have a market value of $4 million. The tokens are then given to a less effecive charity, and recieve a reward of $4 million. Thus the funders receive exactly $9 million total.
A token, like a book token, is a dollar in a locked box, where that box can only be opened in specific circumstances. (Anyone can convert dollars into tokens, but if they do, that implies the market is broken. Suppose a project needs a million $, so it auctions off a million tokens, of face value $1. Bids can either be >=$1 real money, or a mixture of real money and tokens with a combined face value of $1. Sort all bids, from most real value to least, and pick the one with the 1,000,001 th largest amount of real money. All the first million bidders pay that. (The million bidders might be just 3 people, each bidding a lot of bids. )
This mechanism means that, if the funder will retroactively pay $1/util retroactively. Then when investors see a 10 util/$ opportunity, they bid it up until they end up with loads of extra tokens. They then have to spend those tokens somewhere. Any charity who produces >1 util /$ will get fully funded by people bidding in $. So you can't get rid of tokens there. So investors look for places with 0.9util /$ to get rid of the tokens. And failing that 0.8 util/$.
Assuming an infinite number of omniscient investors, who will take any opportunity to profit, however slim the profit margin, then this mechanism is 100% efficient at distributing the funders money to the most efficient charities. Its also complex enough to deter any non omniscient funders.
Another option for founder funding: fixed founder bonus. Founder announces a bonus amount they get paid if the project gets bought out. Beyond this they are paid market rate salary and have no equity. When the project gets bought out, the founder gets the bonus, with remaining proceeds divided proportionally to the investors. Investors know the guaranteed founder bonus up front and take this into account when deciding whether to fund.
Example: $1 million dollars is needed to fund a project with a $250k founder bonus. Investors fully fund the project for $1 million up front. The project is successful and gets bought out for $5 million. Founder gets $250k (in addition to salary they have been collecting) and investors are paid out $4.75 million.
The founders will never get as rich as private sector founders, but they get a healthy reward. The promise of the bonus is probably enough to push some founders towards charity rather than private sector.
There are some details to be hashed out if the project gets bought for less than the investors funded + founder bonus. I would suggest that funders get initial investment back first, then founder bonus is paid with whatever remains.
Disadvantage: Founder does not have a financial incentive that scales with the size of success. This is counterbalanced by them retaining some personal credit for the success of the venture, which does scale with the size of success.
What about the problem of marginally useless charities. For each million the charity recieves, it can build a bed net factory.
The countries it plans to build factories in are, in order, Rwanda, Somalia, Greenland, Kenya, Madagascar.
Predictably 4 of these factories are useful, and the Greenland bednets are totally useless. So once they have $2 million, no one will fund any further. (Unless a single big funder or group of smaller ones can organize $3 million)
So instead of the charities choosing the funding threashold, let the investors do it. Let the investor who only thinks the project will do any good if they get >$5 million buy a share they only have to actually pay for if funding exceeds $5 million. (With a fancy algorithm to pick the largest stable equilibrium, namely assume everyone pays, then start refunding people who haven't met their funding condition.)
Accredited Investor rule strikes again
Fascinating and thought-provoking post for someone unfamiliar with this concept; thank you.
Initially, I had some concern about the moral hazard for oracular funders. Why would they retroactively sponsor completed projects when they could apply their funds towards further improving the world? But I'm less troubled by this the more I think about it. These funders would be playing the long game, and recognize this retroactive funding as valuable for keeping the impact market functioning and incentivizing future projects. In a sense, the concern is similar to asking why people engage in philanthropy at all, even when they aren't directly incentivized to. If they do it out of a sense of altruism, the same altruism could support oracular funding.
I think the most important detail to figure out is the balance between "maximum capitalism" and "project funding only". Like you, I'm deeply uncomfortable with either extreme. I think investors and founders shouldn't capture all the surplus, but there has to be some surplus available to them to make the system work. There seems to be an assumption that the oracular funders would pay out the full value they ascribe to a completed project, but wouldn't it make more sense if they just paid a sort of prize that scaled with the project's impact value? As long as it was calibrated to provide a good return on investment, it seems like it would make more effective use of the oracular funder's funds.
To answer the question of how much to pay, why not hold an auction after the completion of a project? The most generous funder would "win". While someone could certainly pay $5M for the certificate of a project with $5M of impact value, I think a better norm would be for the auction to clear at, say $2M, rewarding the founder and investors without giving them the full surplus.
"When Tesla needs $1 billion in new funding, it doesn’t just ask billionaires."
I feel like this is a particularly bad example, because when Tesla needs $1 billlion, they can just ask Elon Musk.
One aspect I didn’t see discussed explicitly, which might be worth exploring, is risk.
If the $1m malaria project has a 20% chance of success, I’m indifferent at an expected Oracular Funder buyout of $5m. But there is still a trade here; as a Funder I might pay a premium to avoid risk.
So in addition to the “VC” being in some sense better at picking winners, there is also a component where they are just more risk-tolerant. They can sell a service where they buy 5 20%-success projects at $1m, with an EV of $5m, and sell them for $6m; the charity might not be prepared to risk all of their capital on high-risk investments that could all easily go to zero, and instead would be willing to pay a premium to shift the risk into another party that is more risk tolerant.
By financializing in this way, you can increase the total number of projects that can get funded, adding a bunch of more-risky stuff that charities aren’t able to fund. (Of course, this risk premium depends on there being a lot of “high risk” stuff that isn’t currently being funded. I don’t have a feel for whether that is the case.)
"The Copenhagen Interpretation of Ethics says that when you observe or interact with a problem in any way, you can be blamed for it. [...] *In particular, if you interact with a problem and benefit from it, you are a complete monster*. " ([0], my emphasis). This should be considered.
* Any charity with for profit investors is sure to annoy people even vaguely involved with non-EA charities to no end: for them, charities are about attending galas, photo ops with smiling orphans and warm happy feelings. I would argue that such people can perhaps stomach paying a living wage (never mind a market wage) to some charity workers, but ideally, everyone in charity should be a hero, and heroes just don't ask for fair compensation. If you thought the pushback against billionaire philanthropy ([1], [2]) was bad, wait until you see the response to what will be seen as evil capitalists enriching themselves through charity. Making money out of bed nets will probably be viewed as slightly more evil than making money out of selling small arms to some warlord.
* In the public eye, such projects would probably be held to a high standard of virtue of a charity while lacking the advance of good will provided to not-for-profit charity projects. The public won't care how many bed nets the project provided when some minor ethical mishap (like paying bribes) comes to light. If the oracular investor is in any way beholden to public opinion, this is a real risk.
* Donor motivation might be a hard problem: outside of EA circles, where the long term impacts are taken into account, the view "why should the donations pay to reward some greedy capitalists beyond their initial investment?" might prevail.
* I think that the public outrage would vary very much with the field. Health interventions in the third world will probably be a PR minefield, while other fields will fare better. AI alignment is safe, in the mind of mainstream journalists, hard AI is generally considered a pipe dream, so if rich people spend their money on it they generally don't care beyond some op-eds about how "the Real Problem is Algorithms". Development of meat replacement products is safe while interacting in any way with actual animals is not.
Implementation idea: normal startup:
* From my understanding, most startups in the tech sector are not aiming to be independently profitable, but to be bought up by tech giants. When youtube was bought in 2005 by Google for 1.65G$, the yearly revenue of 15M$ clearly was in little relation to that price. If some EA charity established their willingness to buy up successful charity-related for-profit startups, this would allow to use all of the funding infrastructure in place for other startups. I think that this might be more appropriate for continuously run projects (e.g. build and run in the third world), where the startup represents the infrastructure build which can then be taken over by the oracular investor, and less for goal oriented projects (like "vaccinate every child in X").
Further random thoughts:
* If multiple projects compete in the same metric (e.g. QALY improvements), this might lead to the marginal price per unit to be lower, I guess?
* If multiple projects compete over the same population instead, competition might be harmful to the overall outcome. Of course, there is probably no general oversupply of charity, so that competition may be limited. Or projects might try to use patents or lobbying to hinder each other, which seems worse.
[0] https://blog.jaibot.com/the-copenhagen-interpretation-of-ethics/
[1] https://slatestarcodex.com/2019/07/29/against-against-billionaire-philanthropy/
[2] https://slatestarcodex.com/2020/02/24/book-review-just-giving/
So, let’s imagine that these oracular charities agree that a life is worth about $5000.
And then I (a random Joe) save some drowning child, and bystanders collect enough video footage to show the kid would’ve died without my intervention. Could I sell the credit for this deed to an oracular?
And if so, could a volunteer lifeguard sell the credit? What about a paid lifeguard? Or a doctor?
I could go on, but my point is that it’s unclear to me where his line should fall, but also funding good sameritanism seems good.
Suggestion for the design to avoid rushing / paying founders (#11).
(Note: I have actually taken courses in auction theory. I have not read all the comments to see if this was already suggested.)
Project costs $1 million. Oracular funders think it's worth $5 million. Rather than bidding on the share of the project that they're buying, funders bid on the amount of a haircut they're willing to accept. Funders willing to accept the highest haircut are first in line to invest. Maybe you run through all the investors willing to take a 70% haircut, and then have some investors willing to take a 60% haircut who get to invest, but everyone who said a lower haircut can't buy shares. So that project was funded with a 60% haircut. Then, if the project succeeds, the oracular funders only have to pay $2 million to investors (40% of what the project was assumed to be worth).
This avoids the founders getting wealthy or having weird incentives. The savings from "it was broadly believed this project would succeed" go to the oracular funders. Since if folks are willing to accept a really high haircut, probably the oracular funders could have figured out to fund the project without this fancy mechanism. (Conversely, high risk projects will probably have very low haircuts.)
>Accidentally Encouraging High-Risk Negative-In-Expectation Projects
>[…]
>Ofer and Owen Cotton-Barrett have discussed this here. Their conclusion is that final oracular funders should take this into account when deciding who to grant impact certificates to.
Hi, Ofer here. Speaking only for myself, I don't see that as my conclusion. I elaborated in a comment on the EA Forum:
https://forum.effectivealtruism.org/posts/6LppWMdN2NLHceGTr/impact-markets-the-annoying-details?commentId=t7o5Euxxr3F7sukNp
Based on my very cursory knowledge of nonprofit organization law, I would assume that it would be easiest, from a legal perspective, if the founders don't receive any (or at least don't receive much) of the impact certificates, and certainly don't get rich.
I'm curious what the lawyer you consulted with thought about the question of whether founders should receive impact certificates and potentially get rich.
The reason I would think it would be easiest if the founders don't get the certificates is this:
First, nonprofit founders don't own their nonprofits the way that for-profit company founders own their companies. So, if the nonprofit organization is issuing the impact certificates, then I imagine the nonprofit ORGANIZATION can choose to keep some of the certificates (as opposed to distributing them all to VCs). But I would think that the FOUNDER has no right to 'keep' any of the certificates—Since the founder never owned them to begin with.
That said, of course, the nonprofit has the power to pay the founder for the founder's work as a staff member. And perhaps part of the founder's pay could be in impact certificates, the way that companies sometimes pay their regular staff (not founders) in stock options. But, my understanding is that, for the pay to be legal, the pay must be genuinely based on the founder's work as an employee. Otherwise, as another commenter pointed out, I'd think the founder would technically be embezzling money from the nonprofit by taking more value from the nonprofit than what the founder provided as an employee.
Of course, even if I'm right about my above assumptions, this could be solved by a founder choosing to set up a regular, for-profit company to do good work, instead of setting up a nonprofit! There's no reason that a for-profit company can't do charitable work! :-) But a lot of US charities are currently set up as nonprofits. So that's why I think it'd be simplest and most convenient if the founder doesn't get many impact certificates. Because then existing charities could get in on the fun!
Another reason I think it's easier for the founders NOT to get rich from this is that most US nonprofits seem to have 501(c)(3) tax-exempt status. 501(c)(3) is a really cool tax status that gives a nonprofit lots of benefits. The nonprofit doesn't have to pay income taxes, and donors can get tax deductions, among other things! But one of the rules, as far as I'm aware, of 501(c)(3) tax-exempt status is that no part of the nonprofit's income can inure to the benefit of any private individual. My fear is that, if the founder gets paid too much (including by getting too many impact certificates), the IRS might feel that part of the nonprofit's income is inuring to the benefit of the founder. And then the nonprofit might get in trouble.
Again, this 501(c)(3) issue can be addressed just by the founder choosing NOT to apply for 501(c)(3) status. So there's a clear work-around. But as I mentioned, 501(c)(3) status rocks. So, it'd seem easier if founders don't get rich, so that the nonprofits can be 501(c)(3)s.
Lastly, another option for addressing any of the issues I'm bringing up is to push for legislative change. So I'm only talking about the question of what seems easiest given my cursory understanding of the current law.
Very interested in what Scott's lawyer said about this topic!
Should there be a mechanism for distinguishing projects with investors from projects without? If the incentive for the post-hoc funders is that they get to own the 'best' impacts, and they could pick between rewarding a) a project with investors and a medium impact (say getting bednets to X amount of people) or b) a project that was funded through other means with a more impressive-sounding impact (say inventing a new vaccine and delivering it, or something), then they have no reason not to just pick from the whole range of all available finished projects, right? In fact I guess there are lots of funds that work like this already -post-hoc prizes - like the Nobel prize or smaller equivalents, that run on identifying finished projects, giving their people money/glory, and getting credit for doing that giving to the best projects.
The only reason I can imagine them switching to only investored projects would be for the goal of promoting this exact market and incentivising investments like this. While I imagine many of these foundations would see the value in increasing available funding like this, I also think a lot of them are very prestige driven, and risking their reputation would be a high bar to pass. (also, as someone else here mentioned, what if one of these institutions, independent of the market, wants to give a Nobel or other prize to a project inventor/doer in this market who had sold off 75% of their 'shares'? Do they have to split the prize money with their investors? *Can* that glory be split?)
I think especially if the idea is to start off with a smaller/restricted market of ideas/projects, there will be a strong barrier to entry that any finder who joins will need to be really really into the idea of helping create this market and its future growth, because they are giving up a large opportunity space of 'projects they can pick between to reward' by joining this.
Not insurmountable, at all! But while reading, I thought there was lots of good ideas and discussion of incentives for project makers/leaders and investors, and their pitfalls, but I found myself a little lost in the incentives for the oracular funders and how they would interact with the market format.
From mobile, apologies for typos or potential unclarity from that.
Has this model actually been used anywhere yet? I'm not a lawyer by any stretch, but this seems not legal if the final oracular funder is trying to operate legally as a US foundation, nor will what you are calling a "charity" qualify as a 501(c)(3). In particular, see "none of its earnings may inure to any private shareholder or individual." (https://www.irs.gov/charities-non-profits/charitable-organizations/exemption-requirements-501c3-organizations)
Maybe this is the point? Funding things with social impact, but for a profit? I just don't get how this works from the final funder's perspective or the operator of the non-501 "charity." If they're not operating an organization that is actually legally classified a charity, then they're allowed to profit from what they're doing themselves. So why would they sell all profit to their initial funder instead? That isn't how for-profit startups operate today. Nobody gives 100% of the equity in their company to their first investor. The final funder would need to necessarily not actually be a foundation, because foundations are legally required to give their money to actual charities if they want to remain foundations and not pay enormous amounts of tax on their endowments. So it seems like you're instead going to be forced to rely on extremely high net worth individuals who could donate money and receive large tax writeoffs, but instead choose to donate to something that they can't write off instead. I'm sure some non-zero number of high net worth individuals will do that, but it's a hell of a culture change.
This all gets worse if you involve crypto, because at some point, the organization receiving funding has to pay its employees and buy goods and service, and exchanging crypto for real money to do that is itself a taxable event.
Re 4/11: You could pay out the $5 million in the mana that manifold.markets uses - money that can be paid out in donations to charities of your choice. The founder sells half their impact for $2 million in mana, pays out half that to a charity that writes grants to projects that proved their potential by selling impact certificates, gets $1 million in funding, and keeps $1 million in mana, then gets another $2.5 million later. Founders are allowed to get rich, but in mana. Might this also help with the tax issue?
Re II A: Sounds like a job for insurance. The insurer would be obliged to pay the final oracular funder for whatever negative impact they find. The cost of the insurance would be added to the required funding. So the cancerous malaria cure would cost not $1 million to fund, but $26 million.
I haven't read many other comments, so it's probable some of these points have already been made, but I'm confident enough that at least one is novel that I'm willing to gamble your time to save some of mine. Five points.
I. A finer model of the funding ecosystem
You break the ecosystem into founders, VCs, and occular funders, but I think you need to disambiguate between different sorts of people in your VC category. I don't mean angel vs seed vs series A and B funders -- though those may well be relevant here -- I mean direct funders vs aftermarket funders. Let's call people who directly buy equity from funders VCs and people who buy equity from others traders.
You seem to have a lot of distaste for traders, which I empathize with: they make a profit without increasing the pool of money that goes to charitable projects. There's an argument to be made that the existence of a liquid aftermarket derisks the original VC investment, but it's unclear whether that benefit is enough to justify the reduction in money that goes to the project, the founders and the original VCs.
This leads us to...
II. You can send the surplus where you want it to go, by decree
Once you resolve for yourself the question of where you want the surplus to go, you can build the system out to enforce that distribution. The way this would work is that, any time equity is sold for a profit, some percent of that surplus is automatically siphoned to other parties -- probably founders or VCs. You probably need a trick for when the asset value decreases in between transactions, but I'm sure that could be worked out.
This could be done with NFTs, building it into the code of the blockchain, or with a managed online marketplace or some sort of royalty contract. Which option is best probably depends on your preference for working with lawyers vs programmers.
III. Is charity more like art or like startups?
I think the claim here is that equity in successful charities probably acts a lot like fine art as an asset class. Probably this extends to tropes about starving artists and stories about the wealthy using them as tax-advantaged investment vehicles.
IV. Social credit rules aren't necessary and wouldn't work anyway
> If they sell off 100% of the equity, then in some sense that we will have to hammer out socially, [founders] “get no credit” for their project
is the weirdest line in the whole piece. Founders would want to make successful charities because they want to effect change in the world, because they want to be known for effecting those changes in the world, and maybe because they want to get rich. Telling them "you sold all the equity, so you no longer 'have credit' for your work" won't prevent them from having been the team that did the ting, from being known as the team that took the risk and did the thing, or from enjoying consuming the money that came from getting paid to do the thing.
You could write legal provisions to enforce not claiming responsibility for a charitable act if you don't have the equity -- something between a nondisclosure and a nondisparage -- but that seems like a terrible idea.
To me, the focus on social credit seems misplaced. Instutitional funders know that their real power is in ecosystem and common knowledge effects, so committing to fund projects after completion in order to diversify the sorts of projects that get funded seems, knowing that the functioning of this system rests on their credibility, to already be in line with their incentives.
V. Let founders get rich
I also have to say that your aversion to letting founders get rich dosn't mesh with my model of your beliefs. Isn't the great thing about markets the fact that they reward people for doing what others want? Isn't that what makes capitalism so effective at increasing output? Why should it be different when what others want is altruistic?
"Disadvantage: We don’t get the “coolness” boost of using crypto."
I feel like crypto is kind of low status and uncool now. Anyone agree, or maybe just my bubble?
This whole thing has the perspective "impact certificates would be kind of like shares of stock". But what if impact certificates just *are* shares in companies that are trying to do good? I think this solves a lot of problems. Once a credible retroactive funder exists, people can start regular companies whose business plan is to do things that will get them paid by the funder. Those companies can use the existing legal structures for startups, sell shares in the usual way to regular seed investors and VCs, and later on could even IPO on existing stock exchanges if they need a very large amount of capital.
There is a whole startup ecosystem that is familiar with how to set up new companies, how to sell shares to VCs and other investors, how to navigate securities laws, etc.
Traditional companies also have experience dealing with problems like this:
> suppose that you are an investor who believes that the plan to cure malaria in Senegal will succeed (produce $5 million in value), but the grander plan to cure it elsewhere in Africa will fail (produce $0 in value).
This exact problem happens with regular companies selling stock. You might invest in a company because you believe the projects it will do with that investment will produce a good return. But then the company managers may later undertake bigger, more expensive projects that don't produce a good return, wasting money and destroying your initial investment. In fact, company managers have an *incentive* to do this, because starting big new projects is fun and interesting for them and increases their personal prestige, and the managers may not particularly care if the investors don't get a good return from the big new projects.
Some traditional ways to reduce this problem for regular companies:
- The share owners in some sense *control the company*. Shareholders can vote out the current management and install new management if they think current management is not doing a good job at producing a return on their investment.
- The managers' incentives can be made more aligned with the shareholders. For example, you can pay the managers in shares, or you can tie managers' pay to the share price. That way managers have an incentive to do things that produce good returns for shareholders.
Two thoughts:
1. I was discussing this with a colleague, and he pointed out this is very similar to how NIH grants are funded. You write the grant, but in order to get all the data that justifies the grant you basically do 75% of the work the grant is for. Once you get the grant, you're practically done with that project, and the grant funds future endeavors. The only difference between NIH grants and an impact market is that the NIH is effectively funding 'people who did good work in the past, and are expected to continue doing good work in the future'. I kind of think of this as the same fundamental idea with impact markets, but that second half of the idea (funding people we expect to continue doing good work in the future) isn't as explicit.
2. From a great interview with the founders of Air BNB from when they were growing, there's an insight I think impact markets could learn from:
"Paul Graham ... asked us a question, 'Where are all your users?' And we said, 'Well, they are kind of spread out everywhere.' And he said, 'Well, where are most of them?' And we said, 'In New York.' And he said, 'Well, why aren't you in New York right now?' And we said, 'Well, because we are here at Y Combinator, like you told us to be.' And he said, 'Doesn't matter. Go to New York. I want you to meet all your users.' And that wasn't intuitive to us at the time. Because we didn't have any money to do that. ... I mean, the whole idea of being on a web business is that it's scalable. You don't need to meet your users. But he said: It's okay to do things that don't scale when you are just starting out."
https://www.econtalk.org/nathan-blecharczyk-on-airbnb-and-the-sharing-economy/#audio-highlights
This is a powerful insight. You're trying to do some proof-of-concept things. You're trying to get a lot of people into a market and make it scalable. That's awesome and potentially world-changing if it creates strong efficacy incentives for charities. But you already recognize there are some things that only work when there's already a trusted market. In other words, there's a lot you can only do at scale. So don't try to do them yet.
Crypto is great, but is it necessary at this scale? Maybe start by trying a few hand-crafted impact credit projects. Likely, you'll find there are some institutional ideas you hadn't thought of that are necessary to make this work. You'll step in and make the project happen anyway through personal intervention - not something you want to do when the project scales, but that's all part of the learning process - and make sure everyone is still happy with the arrangement. Then you'll tweak some factors, do it again, and have to do a little less intervention than last time. Meanwhile, a few more people will see what you're doing and ask to get in on the game. It'll be small, but manageable, and it will grow in proportion to how well it works.
Don't start at scale. That doesn't give you room to learn and grow. You're either 100% right from the start and wildly successful, or you're a high-profile failure that 'demonstrates these new ideas just don't work'. Start small and build a network while building institutional knowledge, then scale up commensurate with need. That should solve most of the problems you outlined above.
Finally finished a long post on a complicated way to avoid charity founders or investors getting rich off the easy opportunities. In principle the money should go towards charities the investors would otherwise consider not quite worth it. Given many self interested omniscient capitalists, the retroactive funders money is perfectly distributed to charities, most effective first, with no inefficiency. https://forum.effectivealtruism.org/posts/CGqbfs28amWjM6PxK/retroactive-funding-with-subsidy-markets
A while back I randomly thought up a sort of ... blind, assurance contract with a scaling refund mechanism? ... Fundraising scheme but never knew what to do with it. Not sure if it'll help resolve any of the issues you shared, but this post seems like an attractor for the kind of people who could get value out of it. One page description: https://docs.google.com/document/d/1dBTPsiM3nIOS-2KlO8nBIJIlW7EIMyfbshA3n-LgcTM/edit?usp=drivesdk
My initial brainstorming that I haven't thought through completely.
"suppose you invest when there are $1 million of tokens sold, thinking that you are a slightly-better-than-average investor and so can probably do better than break even. Then later some other people come in and buy $9 million more worth of tokens. Now there are $10 million in tokens chasing $1 million in prizes, and on average each investor will only get 10% of their money back."
Hmm. Normally this shouldn't be a problem. You paid what you did, other people buying shouldn't change whether that transaction was a good idea. But it is here.
So, to iterate the problem as I see it. There's a set of projects, that will produce X utilons. Your goal is to predict what fraction of those utilons will be produced by a given project (in the judgement of the granter). Once you've done so, you can tell the value of a project in expectation. If the project sells a fixed amount of shares, then you can calculate the expected value of a share as (fraction of utilons)*(total prize money)/(shares).
That's the goal, but this could be confounded if new projects are added or removed, as that would affect the total utilons (and therefore the fraction of utilons). A common way for projects to be removed is if they fail to meet their assurance contracts. So you do need to predict which projects will get funded in order to predict X, in order to predict relative impact. Similarly, new projects can't enter the market without changing the value of everything else.
So the way others buying tokens affects you is it makes more projects funded, increasing X, and therefore lowering your expected payoff.
So what we want is a fixed price per utilon. An unlimited pot of money works well for this, as described, but requires an unlimited pot of money.
In the interest of brainstorming ideas, another way would be for the granter to commit to spending money beforehand. Each proposal says "Here's the impact I expect to have" and the granter says "if you have that, I'll pay $X for it". They commit their pool of money, and then stop accepting proposals. Every one that has that impact gets paid, everyone that doesn't doesn't, and the leftover money goes into the next round.
Issues are that this breaks things into rounds, and doesn't allow new projects to enter as needed (perhaps fine while still in the startup phase, as long as it doesn't fix a precedent for the full idea). There's also no method for handling "mistakes", i.e. what if the project fails to have the impact it expected? In simple cases, they failed, they don't get their money. But if they succeed wildly, and have twice the impact, they can't receive twice the money. Or if their plan has unintended side effects, those also aren't priced in.
(added post fact: the above seems related to various later points)
"7: How Should The Market Price IPOs?"
Can't this neatly tie in with the "How should founders get paid" part? Something like the founder honestly reports that they're going to split it into 10,000 shares, and how much they need in funding. Then they sell shares, and people buy them at whatever price they're willing to pay, based on how much they think the project will be worth. If the shares are being offered overpriced, then the founder needs to lower the price, and not keep as many of their own shares if they want the project to be funded.
If a project is going to produce $5 million, each of the 10,000 shares is worth $500, and so the founder needs to sell 2,000 of them to get funded, leaving 8,000 for themselves. If people actually expect it to only be worth $2 million, then each share is $200, and they'd need to sell 5,000 of them to get funded. This encourages low cost, high impact projects.
Ah, you mention this. I'm not sure if this is just your auction idea.
"plus people feeling more comfortable knowing that charity founders aren’t getting rich and buying yachts with their donations"
I think while this would be a real effect, it's not a "reasonable" one? The reason charity founders buying yachts with charity money is because it means the charity is usually running away with your money instead of doing the job. But that's what the impact market is for. It kinda guarantees the job is done, so rewarding the person for doing the job is less of an issue. It basically bakes in the accountability that is usually lacking in other cases where a charity buys yachts. Maybe I'm wrong about that, and even if I'm not, it won't stop people from feeling this way, but still.
"I don’t like the fact that it’s important they want to keep as much as possible but they have no real incentive to do so - "
I know it's the same in terms of actual results, but maybe if you marketed it as "79% goes back into funding other charities", it works better? It feels like people in charitable work should be at least partially incentivised for charity in general to be done.
I wonder if you could solve the "founders getting rich or not" problem by giving founders some sort of generic market token equal to, say, 50% of the market value. Say, they sell 50% of the shares in the 1 million funding 5 million value malaria project. Investors get 2.5 million in payout, the founder gets 2.5 million worth of "generic impact tokens" which they can then reinvest. Keeps the money in the market and doesn't feel as icky while still leaving room for an eventual liquidity event for smart founders.
What about treating it more like a market for bonds or commodity futures rather than stock shares? Suppose the Senegalese Ministry of Health and Human Services (or whatever) currently has no cash to spare, but figures if they could collect at least $5 million more in tax revenue over the next thirty years if the malaria problem were completely dealt with. They draw up a stack of 360 "payable to the bearer" certificates, or perhaps some more elaborate system broken down by provinces or individual villages. Each month, another certificate matures and becomes redeemable, at the appropriate government office, for an amount of money calculated based on the prevalence of malaria in the specified region during the previous month: $14,000 for a spotless record, declining incrementally to $0 for "indistinguishable from the status quo," and on down into negative payouts if the situation gets worse. Certificates are distributed to alleged malaria-remediation projects upon request, requiring only a simple credit check (does this person or group have a fixed address, professional reputation, something we could go after if they end up owing us), since a null-hypothesis outcome costs the government nothing. Government-backed certificate plus actual plan can then be sold to an investor, or split up further into derivative contracts.
I'm a bit lost in all the subtleties of this, and not an expert in any kind of finance, but it sounds like you're looking for a model with some properties of equity (unlimited upside, notion of "ownership" of the project itself) and some properties of debt (finite-scope though potentially renegotiable funding, unambiguous "cashing-out" event at some future point).
If you haven't already you should send a friendly email to Matt Levine explaining your conundrum and asking for his suggestions-- because not only is he likely to have helpful input, he might be interested in writing about impact markets and raising awareness about the idea.
I'm an ordinary person, and I'm much less interested in being an investor than I am in being an oracular funder. Currently I give my charity to the SENS foundation because I think life extension is important, and I trust them to be better at evaluating promising life-extension research than I am.
...now that I think about it, isn't giving to a large science funding organisation on the strength of the good research they've previously funded basically an impact market? With the funding org acting as the investor, and me the oracular funder?
My tiny mind confessed to be confused: this looks like a weirdly convoluted approach to something that should be simple.
Instead of all this structuring, why not just say “whoever achieves X I will pay $Y”. Then, companies (structures that already exist to take investors money to utilise for projects they predict will create value inclusive and above of the cost of that capital) will just say, great, Scott will pay us $Y, we can do this project for ($Y*(1-r%)), awesome, let’s do it”. Maybe some companies will form _just_ for this purpose, bypassing the painstaking structuring you outline above to just get the thing done.
It seems that all you need to do is set the prizes, and then the methods will work themselves out with existing collective-projects-for-money (debt-and-equity-funded companies). Why all this headache?
You could also do a cost-plus model, “I will pay $Cost*(1+r) to whoever achieves this outcome”.
Why am I being dumb here? Please tell me what I’m missing!
One thing that seemed glaringly unusual was the idea that funders would want to buy these impact certificates in the first place after the “impact” has already been realized. While that might make these certificates work, I’m not sure that this is psychologically what would motivate a funder. I suppose the comparison is that the funder is paying out prize money for a contest they never set up in the first place. Without this assumption the whole thing kind of falls apart so to address this issue of funder incentives I’d like to the propose the following:
The impact certificates value isn’t actually tied to the measured impact of the program, but it’s tied to the right to be able to provide further funding to the program that the original certificates funded.
A funder in a sense would need to buy the rights to be able to provide funding to this organization and the holders of the impact certificates have those rights and are able to sell those rights to another party.
The assumption here is that private parties may be better at finding and investing in promising organizations than funders and bear the risk and search costs. If a specific program or organization becomes a promising organization to invest in because of evidence or belief or what not, there will be demand to be able to realize the impacts of investing in that organization and holders of these certificates have the exclusive rights to be able to do so. The value of these certificates themselves can be negotiated between parties. With a transparent ledger, like blockchain, the value of these certificates can be made public information and viola, do we have a working market for impact?
I think it would be useful to have a very high level summary of what is trying to be achieved by this project:
Charitable givers have more money than they have ideas. Therefore, they would like to spend some of that money on new and better ideas or selection between ideas (hereafter collectively "wisdom") instead of on the projects themselves.
I like having this simple input/output framing, because it immediately makes obvious the tradeoffs: 1) more money to founders will get you more and better ideas. 2) more money to investors will get you better valuation of existing ideas. But increasing either will mean less money will go to the projects themselves.
The other point I would like to harp on is that a stock market can only value things that have units of [money]. Since companies produce revenue, it's easy to make a stock market work: the thing that is being traded is a proportional claim on future revenue. Stock market fundamentals investors only have to solve one problem: predict the future revenue of company X and compare it to the current price. Impact investors have to solve two problems: predict the impact of project Y and the oracular conversion factor between that predicted impact and money and then compare that to the current price. This is not an unsolvable problem, but it's harder, and one should expect to have to pay more for it.
Finally, I would say that the engineering perspective leads me to believe that an impact market should be structured so that it can give different payouts to different project areas. For example, if the problem that funders want to solve doesn't have any remotely feasible solutions, a payout schedule closer to full capitalism might make the most sense. If the problem has several obvious solutions and the only real obstacle is funding, then limited payouts to market players would make sense. I could also see the need to move from one to the other over time in the same problemspace.
I think founders getting rich might actually be a GOOD thing. Employees at a startup are often granted stock or options that they can't sell for a period of time, in order to motivate them to work hard and make the startup succeed (addresses problem I-D).
Also I suspect it would be hard to stop them, because they can use tricks like buying their own shares through a cutout or lying about their costs or paying themselves a bigger salary. Maybe you can prevent all this, but there's a lot of different options that would need preventing, and there would be enforcement costs.
Also note that founders who get rich can re-donate their money if they want, so this isn't allocating surplus to founders who just want to be charitable, only to founders who became founders at least partly because they wanted the money.
I estimate a 90% chance someone already said this in another comment, but what the heck:
Here's a way to auction off a bunch of shares in something, without requiring one person to buy 100% of the shares:
Announce a window of time for the auction (perhaps a month, like a typical Kickstarter), the total number of shares (let's say 1000 as an example), and a minimum price per share (so that the assurance contract fails unless you earned enough to fund the project).
People submit bids of the form "I will buy up to X shares at $Y per share". (This is treated the same as X separate bids for 1 share each.)
At the end of the time window, the top 1000 bids each get to buy a share at the price of the 1001st-highest bid. (You don't want to punish people for bidding higher than necessary.) For example, if A wants to buy 100 shares at $20, and B wants 500 shares at $15, and C wants 800 shares at $10, then A gets 100, B gets 500, C gets 400, and everyone pays $10/share.
If there are less than 1000 bids, the assurance contract fails. If there are *exactly* 1000 bids, then they buy at the minimum-funding price.
Ties go to whoever bid first (because if you break them randomly, someone will submit a trillion bids to try and win ties). This still gives a little value to fast-movers, but capped at 1 bid increment per share.
(I don't know if this is a standard procedure, because I just made it all up, but it seemed to me like an obvious combination of various auction rules I've seen before.)
What's a recent example of a charity that got funding despite being uncertain at first and now definitely obviously has made an impact and you would totally leap at the possibility of getting the credit now if you could be a final oracular funder? (but the thing has now already been done)
How long did it take to become clear it was successful?
What kind of timescale would the initial investors be expected to wait before finding out they're going to get paid or not? Would this be upfront? What if it takes longer than expected?
I feel like i'm missing something - why would we believe either “Credit For Funding The Project” or “Full Credit For The Impact Of The Project” would work socially?
This is not true in any other markets I’m aware of; if you invest in an IPO you don’t get credit/status for funding the company or for its impact. Intuitively, I’m not sure why anyone would accept the idea that you can take credit for something that has already happened without your involvement. The closest example I can think of is NFTs, of which I remain skeptical for all the typical reasons.
Seems to me like the phrase "and everybody knows it" is doing a lot of work here, no? The point is that not everyone knows what's going to work, so there is a significant discount investors have to accept. For example, if a founder claims to need $1m to do a project that aims toward getting paid out by a $5m grant, investors need a 20% chance of success to simply break even in probabilistic terms, not to mention economic terms (cost of capital, liquidity premium, etc.). My guess is, the success rate is somewhere in the 1-5% range.
If you believe that to be true, the problem becomes, "how many actual projects are going to be funded through this mechanism?" i.e. what is the market clearing quantity? At 1-5% success range, we would expect a $1m project to require a $20m-100m grant. The number of matches seems vanishingly small, such that the median annual funded project seems to be 0.
If you believe that to be true, then I might propose that the contracts on these impact markets need to start smaller, higher frequency. For example, could there be an impact market of, say, $3,000 for someone to put on a classical music concert in their town square? This approach, if it works, makes measurement/verification easier and the system gets more reps. Obviously, the problem is that it doesn't immediately solve effective altruism goals. In the long run, though, if this approach is successful, you can imagine it leveling up into these bigger $5m Senegal-malaria outcomes.
Super late to the show here, but the whole discussion of the disadvantage in question 4 seems incorrect. Here’s a key quote: “But if we allowed founders to capture more of the surplus, then maybe the founder would end up with $50 billion, and the charity and its beneficiaries would only get $50 billion, which seems much worse for them.”
Maybe I’m misunderstanding how this whole thing works. I thought that the final oracular funder decides what the value is. That value is not dependent on the cost to produce it. (That’s the labor theory of value, which is discredited.) To make this more obvious, let’s say that the method for selling off impact certificates for completed projects in an auction. The project investors put together their evidence of completion and value, bidders get to review it, and the bugging happens. The supply of that particular impact certificate is fixed. Demand will set the price. If the auction comes out at $100B, so be it. If it comes out at $1, so be it.
More to the point, the beneficiaries already got all of the benefit of the project, whatever it is. The impact certificates are about who gets credit for producing the benefit, not who gets the benefit itself! Then, in the example quoted above, it’s very unclear who this charity is. Is this the final funder? They don’t get money at all. They are the ones who valued the project at $100B and paid that amount to buy the credit. Are they the intermediate investor, who funded the work in progress? If so, they knew that the funder was reserving half the equity. Just like VC firms, these intermediate funder will not want to invest in risky propositions where their upside is too low. That means that the market will place a practical cap on founder-retained shares.
I think the key thing is the mechanism for determining the value of the completed project. There are a lot of auction methods. One of them is probably best suited to that role.