Is there a concept of insider trading on prediction markets? Also, I can see this type of bets going very wrong on things like "elections" (everyone with money on A. Hitler winning the election is likely to vote for him no matter what).
I am sure that this is addressed somewhere and that I have missed it though, as mine looks like a "day one" objection to prediction markets and everyone else is very excited about them.
"In a prediction market, once you're wrong a couple of times, traders will stop updating on your reports and you'll lose most of your power to move the market."
I don't know about this. Does this happen to people who try to pump and dump stocks? Why would it be different for prediction markets?
> If we've gotten so polarized that no conceivable fact could change who we vote for, then investigative reporting should have the same status as golf
I think this is underselling the value of investigative reporting. Suppose I'm an American reporter investigating Putin or Xi. There's value in my reporting news about them, but not many immediate results. Perhaps the prediction market could measure "American sentiment toward China" or something like that, but I suspect that on predictions that large any given investigation is too small to make the averages pay out in a reasonable timespan. If my investigation makes the result 0.1% more likely, then my newspaper only gets the payout after many investigations on average, making it probably not worth their while. A national-level newspaper like the NYT or the Washington Post might be okay, but anyone smaller won't bother.
In addition, this means that a newspaper could only have one position on a given issue. Suppose I'm a nonpartisan newspaper who wants to investigate the election. If I accurately report on both a Republican-favoring issue and a Democrat-favoring issue, then the election odds are the same as before, but the voting population is better off. This would encourage partisanship in reporting, which seems like a poor result.
"I'm looking at the front page of today's Washington Post, which has something about how there's a culture of misogyny in some important military school. This could charitably be termed investigative reporting - we can imagine the journalist painstakingly tracked down women at that military school, heard their stories, hunted down documents proving that this school discriminated against women in some way, et cetera."
Why "charitably"? Paywall and all that so I'm at a disadvantage, but this sounds like a central example of investigative reporting, and you seem to be straining to "imagine the journalist painstakingly tracked down...", which seems like exactly how I'd expect that reporting to work.
I like this idea, but running some numbers - it might work out? (My original claim was that it didn't, but on second thought I'm not sure).
Let's go with the military school example. A damning study on inequality there might move betting markets on the firing there by, say, 10%. So if you're willing to pay an investigative reporter $x for that story, you need to invest $10x into the betting market to have enough liquidity for this to be worth the reporter's time.
But it's actually worse than that - most investigative reporters' investigations will (hopefully) not turn up firing-worthy offenses. so if say 1/12 month-long investigations come up with something, x needs to be about a year's salary for a reporter. And this 10x (which is now on the order of a million dollars) isn't buying you improvement, it's just buying you insurance that you'll probably catch anything too egregious going on in that particular guy's watch (since you have to get this insurance for every single employee).
So putting all these numbers together - you'd need about a million dollars' worth of insurance per senior employee, that would cover all-cause scandal, and would pay out if the betting market resolves (I'm assuming the school takes most of the money if they don't fire the guy). That... Actually seems kinda doable, I think, at least for senior employees.
Won't this run into the same issues highlighted in Inadequate Equilibria for multi-stage funding, where you're rewarded not for predicting what will happen, but for what other people will think will happen given some news? Ie. the longer the time horizon in play, the more a market has to worry about whether it's predicting outcomes or "just" future market sentiment.
Life satisfaction markets seem dubious to me. Could it incentivise companies to avoid hiring people with histories of depression? Also, could a bad actor "short" such a market, then attempt to reduce the life quality of the market's subjects by nefarious means?
1. It helps the negative or expository reporting, but less ones which are in depth views of topics which don't necessarily have a fireable culprit.
2. Stock markets work because they normalise equities (same re other mkts), each equity looks/ feels/ smells roughly similar to others. That's the hard bit in moving the concept to more esoteric markets.
3. The infotainment industry prediction would sure make TV a hell of a lot more fun. Or at least CNN ! I'd even settle for realtime betting alongside news broadcasts.
That said, I like the idea of fire the CEO markets, even if illiquid, though I'd also note that Glassdoor exists, often with horrible CEO reviews, and the Board rarely takes that into account barring properly illegal conduct.
All in all, I'm very supportive of prediction markets. However, I do see a key flaw here.
Right now, the media has a vague incentive to be partisan - it helps them get more clicks and slowly builds up an audience with a consistent ideology. I would say that this is Not Great - the utility that most people get out of the media is unbiased information, which partisanship is at odds with.
With prediction market being the key to fund them, they have an explicit, definite incentive to be partisan and write as many hit-pieces as possible, take every quote as out-of-context as possible. What if Cade Metz could make $50K, contingent on Scott's blogger approval rating decreasing by 5 points?
Interestingly, this flips the onus of The One Unbiased Truth from the old school media to prediction markets. Culturally, it might take a bit for people to get used to changing where they get unbiased information from, but I'd argue we're going through that already.
> Very few of the people who cared about Watergate bought subscriptions to the Washington Post
Are you sure - breaking such large controversies, in a pre internet era, could significantly drive purchases and subscriptions of newspapers. It also dramatically raises the status and just direct power of the investigators, which is its own motivation and desire.
Also this seems to be precisely the same as insider trading
> Maybe then we should just accept that the market is correctly telling people not to bother investigating Adams
How on earth is this a good predictor of whether or not someone is corrupt? Even if it’s “the sum of the knowledge of the smartest people everywhere” I don’t see how a bunch of common market participants will have any clue whatsoever how likely it is some particular guy out of a million is corrupt or has dirt hidden, and even then how that helps you investigate (investigation is something you do about a person in particular areas for particular reasons, not just abstractly, you have to know what about them to investigate and why, and also there are politics involved that probably drive this stuff way more than abstract likelihood and price)
Also this is legal? One guy in a few weeks could create a prototype investigation market or ceo firing market on the internet or using one of the many crypto platforms.
I wonder how this would relate to stories with multiple possible impacts. Suppose someone does a story which turns up that Florida condos aren't inspected adequately. Does this win only if there's a deadly collapse? If inspections are improved and a deadly collapse is averted? Or a building is found to be in bad shape, evacuated, and repaired?
"This reporting has been brought to you by conflicts of interest!"
I am deeply skeptical of this proposal. If I learned that some newspaper today was shorting the corporations criticized in its pages, I would be shocked and appalled, and I would deeply discount its reporting. I don't think I'd feel better if everyone was doing it officially.
It's possible that in some future age our present norms against this sort of conflict of interest will seem quaint and absurd, like the historic prohibitions of usury. But I suspect not.
How did prediction markets become such a “thing” in the rationalist community? Is there some notable origin story?
Rationalists need to be able to explain why real markets failed to predict [insert Enron, WorldComm, or GFC example here] but how esoteric, illiquid derivatives will do better. Or why there’s liquidity problems in credit markets (the world’s 2nd largest asset class?) but a prediction market subsidy will work super great.
These ideas sound compelling but market construction is really hard. There’s still not a real exchange for bonds. Bond tickers have wide spreads and are thinly traded (despite Matt Levine’s smartass quips, bond market liquidity is a big issue!). The derivatives this community postulates over have similar complexity bond covenants so one would expect huge challenges in getting these ideas to trade to a point where they’re “predictive”.
I’m not going to address the batshit moral hazard opportunities because I expect others will jump on those.
Back when media was all broadcast over public airwaves, as a condition to use bandwidth media companies had to carry content in the public interest, which mostly meant news. And, this mostly meant the news bureaus were quite independent. So, this journalism was essentially publicly subsided (the FCC could have charged more for bandwidth use otherwise).
Now that media delivery is all private, this doesn't work. Fully supporting news with advertising and treating at a profit center is problematic. So, I say go back to directly subsidizing journalism. Support local newspapers according to the population. Establish a national board of journalistic standards, much like those for medical standards, and yank the subsidies if they violate them. Etc.
As you say, it is a public good. Why not treat it like other public goods?
The problem with this is that it only works if there is a lot of money in the prediction markets. But you can only get there if people already care a lot about prediction markets and put a lot of money in them. Why would any investor want to put money into anticipated life satisfaction for a random demographic slice, when they could put it in index funds, or even more lucrative prediction markets.
Also I'm not sure that the math works out for how much money you need in the system for it to work. How much money would need to be being bet on military school principals firings for a swing in the numbers for a particular one to be big enough that it's enough to fund anything meaningful?
I love the idea. Towards its feasibility in the news space, it might be necessary to invent defences against litigation. Media organisations currently can't expose every piece of material that comes their way due to legal costs. They have to pick their battles, which is part of the reason the best investigative reporters like Sy Hersh didn't stay in one place for long. (It also had something to do with his uncompromising personality)
Perhaps it'd be possible for power structures or corporations to tie up reporters in legal battles while denying them funds from their prediction market win, pending court ruling, which could take years. WikiLeaks is a great example of how far a financial blockade can be taken. It may be necessary for a reporter using prediction markets to rely on donations for legal battles in the short term post-release, or maybe it'd be possible to operate through some third party organisation built specifically to protect them in this way.
Another problem to solve would be getting people to take notice in the first place. There've been revelations about fossil fuel companies and the DOJ recently that haven't made a dent in the public discourse because traditional media won't talk about it. Mainstream signal boosting these days depends on being highly partisan or something both sides agree on. A report that harms a bipartisan narrative (such as the US' involvement in Venezuela being defined as "aid", or most other pro-empire narratives) is fighting a massively uphill battle.
No thanks. We should not be funding the Woke Olympics in this way. In fact, this sounds like it could accelerate the "find people who have heterodox opinions and cancel them" movement. I do miss the investigation of things like mistreatment of animals in factory farms. However, the cynical part of me knows that this would be used to attack people with minority. We shouldn't be encouraging paparazzi to "investigate" Chris Pratt and find that he's a devout Christian in order to maybe attack Disney's next film.
This scheme incentivises investigations, and then to reveal the new juicy info in a way that reliably changes what is knowable by people with money who honestly want to find the truth, so that markets adjust. Note though that this isn't the same as convincing the mainstream or publishing honest articles in NYT. We know already that there can be a big disconnect between what is knowable and what info reaches the mainstream. But one can hope it pushes in that direction at least.
It's an interesting concept for sure. I wonder if it could work for all public office holders. Public corruption is where the stakes are highest for citizens, and it's best investigated by local media.
I'm very glad you're writing about prediction markets, as they sound like a great possible answer to the messed up situation right now with public narrative machines, and the absence of a functioning entity that does what 'the news' is ostensibly doing.
Hanson's fire-the-CEO-market is based on the stock market price of the company, conditional on the CEO being fired or not. It's not a bet that the CEO will in fact be fired, although certain differentials would suggest that's what the board should do. So a fire-the-head-of-military-schools market would need to be based on some measurement, and it's unclear whether a poll of the satisfaction of students is actually at all a desirable measure. Student evaluations of university teachers are notoriously awful and just incentivize easy classes rather than actual learning.
Intriguing, but I would distinguish between incentives to inform, and incentives to convince (or provoke, or startle). My main concern would be that this system may incentivize actors to promote a *perspective or belief*, which doesn’t have the same positive externalities as detached investigative journalism.
Once you’re wrong a couple of times people will stop trusting you and your predictions will stop moving markets. A perfect incentive to go Big, even if there’s a chance you’re wrong, because you can cash out even an eventually wrong position if you can move the “stock” down, but you might turn out to be wrong later and people will stop believing you, so bet big. Incentive for over-confidence.
People not valuing true information is a pretty well established problem, and I can see how prediction markets would incentivise people to value it more. But doesn't it suffer from a problem where the more professional the market becomes, the fewer people can realistically make any money from it? Only people with new unknown information can do that, and only people who expend a lot of resources can find new unknown information. It seems like there wouldn't be much of a reason for ordinary people to participate in or care about these markets.
I think the solution we should be looking for looks more like a culture where most people care about true information, not one where most people don't care but a few people do and make a lot of money out of it. Arguably, the latter is a pretty good description of what we already have.
This sounds like a brittle, complex, fraud-prone way of allowing schools or other institutions to offer financial bounties for verified reports of misconduct. Which they can do already if they want to.
I feel like prediction markets are Scott’s hammer for which everything looks like a nail.
Really great to see this idea raised, and Hanson's name invoked in context. He is likely the most knowledgeable person around on the subject of Prediction Markets. We talked about a very similar idea recently... there are a number of investment firms built around a very similar model as Hindenburg's, but focused on the fact that China's economy is what I like to call 'structurally inefficient'. (In very rough economic terms, efficiency means that the value of any given firm or entity is based on the public having relatively accurate knowledge about its activities and therefore its market value. It has been widely observed that the structure of Chinese bureaucracy means that any given layer of management is incentivized to give inaccurate reports of activity and value to those above it... this view has strong explanatory power in the context of 'ghost cities' and firms that have recently been delisted from western stock exchanges for deeply structural reporting fraud. Thus 'structurally inefficient'.)
Hanson observed that a prediction market that incentivized people lower in the bureaucracy to report true information relevant to market valuation (say, with untraceable cryptocurrency) could very well work, but as with all Prediction Markets, it would need someone to pay for it. Reading between the lines, I took this to mean a 'first mover' to kickstart such a market. Of course, something along these lines could be just as interestingly disruptive to Western companies as to those in China.
Anyone looking to throw money at the problem? Maybe a Kickstarter campaign? Maybe something something NFT market something?
I still don't understand why the most obvious outcome of this isn't people terrorizing the women at a given school in order to make a ton of money predicting they'll be very unhappy? That seems a lot easier and more reliable than doing a ton of investigative reporting and hoping that there's actually something to find.
Or, more generally: if you can make $25million by finding a scandal about a politician, can't you make $25million by inventing a convincing fake scandal? The 'infotainment industry where people try to get clicks by accusing each other of being racist pedophiles' seems like it would be more lucrative than investigative reporting, unless we somehow create a society where people do not respond to those accusations in ways that affect things measured by prediction markets.
I feel like you to some extent invented Bloomberg. Which isn't a bad thing to invent, since Bloomberg is something like a $20 billion company with $10 billion in annual revenues.
On the other hand, the stock market as a whole is valued around $40 trillion. Apparently Bloomberg has a few competitors like Thompson Reuters that are maybe collectively another $10 billion in revenue. So that would suggest that you can get about $1 of analytics value for every $2,000 of trade volume.
I can maybe believe that there's a future with predictive markets on whether the president of (say) Iowa State will be fired in the next year. I really struggle to believe, however, that more than, say, $1,000,000 could be invested in the prediction market (already double the president's salary). So that suggests about $500/year in potential investigative reporter revenues--to generate $100,000 in revenue, you'd have to provide a Bloomberg-esque product for 200 million-dollar questions.
For questions like "is there a culture of misogyny at school X," this seems unlikely. The reason the Washington Post cares about this isn't that their readers had a preexisting interest in school X, it's that they had a preexisting interest in cultures of misogyny, and are interested to hear about examples of those cultures. To the extent that that sort of reporting is useful for making predictions that the readers actually care about, it's in a very general, world-view building way, not in a specific, directly actionable way.
Apart from all the usual issues with prediction markets, the failure case here is that you can't have prediction markets for every possible scandal, and you certainly can't have liquid ones.
I don't think ideology really works like this. I've even shown obvious lies within the Rationalist community and only a handful of people updated their priors.
Short reports don't have to be even remotely true to make a ton of money, and the first amendment in the US makes it next to impossible to successfully sue.
This is such a stunningly, obviously bad take that it's not worth taking the time to detail all the reasons that it is misguided and unworkable (and I assume the earlier comments have done so).
But what I did find interesting and worth commenting on is how this pieces beautifully illustrates the shortcomings of libertarianism. Investigative reporting is a public good. There clearly isn't a market mechanism that will fund it, but there are plenty of non-market alternatives one can imagine from non-profits to rich benefactors to government subsidies. The fact that this essay instead tries to posit this utterly ridiculous predictions market scenario is telling. It's a sign of someone who is unable to confront the shortcomings of their ideology.
Are we measuring instantaneous life satisfaction ratings, or aggregate life satisfaction ratings over the next N years? I can boost the former by abolishing tests/grades/classes and subsidizing parties. The latter will tend to be less liquid because there's a lot more opportunity cost to an N year investment. If N covers the entire lifetime of the students, most of the people with money to spare for prediction markets will be dead before they get paid. There's a tradeoff between the liquidity and the meaningfulness of the market.
Do we really want to live in a world where it's common to have a prediction market on you, specifically, being fired? I realise this already exists for public figures, but extending it to a vastly higher percentage of society seems like a huge social negative. Worse, it would obviously be gamed - as soon as we start to measure something, it becomes useless for its original purpose. The unintended consequences of this idea would be massive.
If investigative journalism is a public good, why not support it with grants?
I'm not sure how you would allocate it specifically, but given the incentives market based journalism already face, adding more market seems like a weird solution.
> For example, our current model incentivizes fake news - if something is exciting and outrageous enough, it'll sell a lot of papers, and even if a few spoilsport researchers come out a few years later with definitive proof it was wrong, by that time nobody will care. In a prediction market, once you're wrong a couple of times, traders will stop updating on your reports and you'll lose most of your power to move the market. You'll have no particular reason to care whether what you write is outrageous or exciting, only that it's true and relevant to something that people care about.
your model incentivizes fake news equally as well. Remember we're not betting on "Candidate X did something bad" we're betting on "candidate X is not going to be reelected". You can prevent reelecetion equally well with sufficiently convincing fake scandals as with real scandals. If a fake story sells the papers, there is no reason to assume it won't move the polls.
I could see how (by keeping the money at stake low enough) you could prevent companies from shorting and then firing their own CEO, but how would you stop self reported levels of happiness from being gamed by the students given that they have the ability to self organise and there's nothing else at stake.
As usual, you did predict my main concern with this, but I feel like maybe you didn't take the negative outcomes far enough. This doesn't seem dangerous while we're talking about agents as large as investigative reporting teams, but this (potentially) degenerates into individuals doxxing each other for the profit they can make off any drop in status. At that point, we're all suddenly living in the hellish world you found yourself in for a few months last year, where some people can make a few dollars publishing things about you which you'd rather they didn't. I want to be very careful with this line of critique though, because at the end of the day, it IS just a slippery slope argument. I just wanted to offer the hopefully balanced view that I don't think we know how slippery it really is, or how unpleasant the bottom of that hill. Who knows, maybe it's still a better world, due to improved investigative reporting being worth more than whatever is lost.
All that said, I do really like the idea of this. It seems to reward positive behaviour, at least right up until it potentially doesn't.
I like the system, and would like that it be tried. But I think that the potential problems are larger than what you make it seem. For instance, prediction markets are markets for entropy & chaos, i.e., you are very incentivized to cause unexpected things to happen.
a) If you subsidize markets (as suggested by comment below), you also subsidize fake news. If you can move "the market" with a bit of fake news, and "the market" is actually one trader + millions in subsidies then that's free money, basically. I mean we do have a market for news, and fake news seem to be doing quite well. Maybe it isn't "about" credibility?
b) Insider trading. This isn't as bad as everyone thinks (it was legal in parts of Europe until quite recently), but probably not great for liquidity. But also, the kind of insider trading where the school principal shorts his stock for millions and then goes on a "I have fuck you money now, fuck you all" spree.
c) Weird incentives to move markets as opposed to delivering the truth. We already kind of have these, but I'm kind of happy with the current situation where a lot of the smart people are happily modelling things in finance instead of being worried about how to manipulate prediction markets on things that actually matter.
"Then everyone could just ignore the first thing and go about their lives" that's _got_ to be a typo, surely you meant that everyone would ignore the _second_ thing, the way they do now. But yes, this totally reminds me of the Court of Values and the Bureau of Boringness and seems equally a good idea despite the tongue-firmly-planted-in-the-cheekness https://meaningness.com/virtue-court
"I want a world where protesters march into colleges and ask them why, if they claim to care about female students, they don't have a subsidized female-student-life-satisfaction prediction market, so that they can credibly set targets for gender equality"
Yeah, that's just what we need, discriminating on the basis of gender. I thought that was illegal. Oh, right: the law may be written that way (the Civil Rights Act of 1964 and numerous other laws) but we all know that it now means, discrimination is illegal if done by the wrong people for the wrong reasons but legal (and wonderful!) if done by the right people for the right reasons.
"Even if their dirt isn't quite that good - even if it only makes Adams seem a little bit worse than people thought - it would probably affect his chances by 1 or 2 percent. And that's still $250K to $500K - more than enough to hire a couple of good investigative reporters."
The market math doesn't work this way. The journalists can't trade $25M without moving the price. Perhaps they could trade $500K on the Adams bid, but I suspect it'd be closer to $50K before they moved the price more than 1%. I think their best case on such a story is ~$10K.
> In the far future, the news might diverge into an "infotainment" industry where people try to get clicks by accusing each other of being racist pedophiles, and a prediction-market-movement industry where people try to generate true information on important issues. Then everyone could just ignore the first thing and go about their lives.
I predict that the strong majority will prefer the first, and the desire (both top-down and consumer-driven) to appropriate some of the latter's prestige by aping the format will blur the lines for all but the most diligent consumers. I predict that the commercial success of the infotainment will be an inevitable draw for the market-makers, and some of the compromise will come from that direction, too - gamification of results, clickbait to promote certain markets, maybe outright corporate consolidation with the editorials.
In short, I predict that even if one could snap their fingers and establish such a division, it would largely collapse within a few decades. But it won't get that far, because those forces would already be at work when trying to set up such a system.
Or by analogy - how much do you care about the extant division between reporting and opinion? You'll get more of that, for better and for worse.
1. This only funds journalists who end up finding controversial stories. Lets say that I'm a reporter and I get a tip that politician X had a sketchy affair; in the current model I can safely investigate it since I know I'm salaried and I'll get paid either way. In your model, I get substantial compensation (likely) if I find out that the affair actually happened, but nothing if I find out that the tip was bogus, or if it wasn't bogus, but I just can't prove it. Beyond the substantial pressure that would put on reporters to just claim the most controversial interpretation of the facts it means that reporters would be incentivized to only follow leads they're pretty sure are already true. It's a good thing when a reporter spends 4-5 months investigating a rumor and finds out it isn't true and your system doesn't reward that.
2. The current system has an maximum compensation ceiling for lying reporters. Sure, if they lie well enough they'll get a promotion or two and maybe an award, but in bounds with their existing compensation levels. In your system, however, the right report at the right time could be worth hundreds of thousands or even millions of dollars depending on the market. If I can do that a few times, hell, if I can do that even once I'm set for a while. Sure, maybe my stuff gets discounted in the future, but who cares? I've already made bank. This incentivizes everyone to be maximally controversial at all times. The normal reaction to that is, "ok, let's just pay less attention to controversial reporting," but now you've set up a system where controversial reporting is dismissed out of hand.
"Why should I believe this report on Eric Adams accepting bribes, the paper that revealed it has a huge financial stake in him losing the general election after all!"
“But it's hard for investigative reporters to capture the value they produce.”
"It's hard for them to profit from the damage they inflict." FIFY
Fusion GPS figured out how to fund investigative journalism. You just work on commission generating dirt on a target that someone with money wants to take down. IIRC, I think they got something like $25 million from Hillary and the DNC to set up Trump with the Steele Dossier (or maybe that was just Steele alone).
The dirt little secret is that 95% of investigative journalism by the Legacy Media always consisted of just getting data dumped in their in-boxes by someone with a vested interest in having them publish it. The main sources are: (a) Plaintiffs' attorneys giving you the litigation files they accumulated through court-ordered discovery in order to help pressure a defendant corporation into settling; (b) official government operatives trying to spin the public narrative; and (b) disgruntled "whistleblowers" who are trying to either CYA themselves or take down their enemies.
Woodward and Bernstein's epic "investigation" consisted of having a guy name Mark Felt contact them and give them the whole story to print, because Felt was pissed that Nixon passed him over for a job. https://en.wikipedia.org/wiki/Mark_Felt
> In a prediction market, once you're wrong a couple of times, traders will stop updating on your reports and you'll lose most of your power to move the market.
Is that true, though ? I understand that, currently, prediction markets work more or less this way -- but the current prediction markets are niche, boutique affairs. What happens when they go mainstream ? There does not appear to be any mechanism that prevents prediction markets from succumbing to the same failure modes as the stock market -- with all the insider trading, CDS tranches, high-frequency algorithmic trading, and a myriad of other financial instruments and tricks essentially designed to decouple the value of the trading token from any real-world considerations.
You could pick a group of institutions and bet against any member of the group being successfully prosecuted by whatever authority for whatever crime you think might be happening.
It seems like this is a good way to fund investigative research, but also to amplify little-known stories and to fund negative advertising that has no basis in fact. Not checking the facts is cheaper.
"Suppose some kind of media property uncovered evidence that Adams took bribes or committed some other career-ending indiscretion. In a world where prediction markets were as liquid as stock markets, they could short Adams, reveal his dirty laundry, crash his election prospects, and collect some portion of $25 million."
What a bad idea, and a surefire way to push journalists into smear campaigns and moral outrage "exposes" even worse than they already do today, which hurting politics and business in the process...
"In the far future, the news might diverge into an "infotainment" industry where people try to get clicks by accusing each other of being racist pedophiles, and a prediction-market-movement industry where people try to generate true information on important issues. Then everyone could just ignore the first thing and go about their lives."
We sort of already have this; the financial press (e.g. FT, WSJ, Bloomberg, Reuters to some extent) is generally much more sober and reality-oriented than most of the rest of the media, although their focus is obviously on existing markets. I suspect the FIRE sector is the last real market for non-sensationalist factual news.
As someone else mentioned here in the comments, though, the financial press is much smaller than the markets themselves. Of course, the public financial press cannot profit in the markets by using their own research, while the banks, hedge funds etc. (that probably do more investigation of markets overall and commission a lot of private research) can, and are much bigger, but I don't think much of what they uncover is classifiable as investigative reporting. That you have a small, in principle unbiased public financial press and lots of large profit-seeking institutions that keep their research private for market advantage is unlikely to be an accident. For instance, I suspect that in most markets it better to go for trades on things that don't need 2+ years of investigative reporting to maybe pay off.
A good example might be the Wirecard scandal. More than one hedge fund voiced their skepticism, published reports outlining problems, and shorted the stock, but it's to find info on any of these original shorters making much money, since there was no consensus on the reports being true. In the end, it took several Financial Times reporters years of digging, travelling to various places in Asia, meeting with established contacts, following up on leads etc. in order for their evidence to be compelling enough for an audit to be commissioned revealing the fraud, pushing Wirecard into bankruptcy.
All during this time Wirecard was making both public and private accusations of the FT working with short sellers, even getting the German financial regulator to open investigations into the main reporters on the story. This was not a very credible accusation, although not totally impossible. But how would the story have been received if the FT had said "Yes sure we're shorting Wirecard in order to found our investigative journalism", even if that hadn't been illegal? Even if that had been fine, for years there was no clear evidence that the irregularities at Wirecard were anywhere large enough to tank the company. Would it have made sense for the FT to dedicate years of work on this if the highly uncertain payoff of shorting Wirecard at the opportune time was the only possible reward?
Investigate reporting depends on sources willing to talk. Why they do this can be complicated - to feel important, to fight their own battles within the organization, etc. - but a big element is that blowing the whistle feels like a noble and righteous action. This turns it into something more like insider trading. Perhaps investigators will still find sources by offering sufficient cuts of the booty. But I don’t think anyone will be compelled by conscience to share information with someone who intends to profit from it. And their peers will _definitely_ see them as selling out. To say nothing of the increased investives to lie.
I wonder if that would create an incentive to cause things to be true? I can try to foment dissatisfaction with Eric Adams or the head of a military school - I don't even necessarily have to lie. That might be cheaper and more reliable than investigating him in case he's dirty.
TL;DR: investigative journalism could be funded through the Universal Electronic Library.
It's an interesting idea, one that clearly deserves to be investigated further. But I believe it can only work for a limited subset of kind of investigative journalism. To find a solution, I believe we need to look at the wider picture, and I think the core issue is this:
The everybody-pays-for-oneself model is bad for information goods.
This model is bad for journalism, it is also bad for science — ask any scientist about SciHub —, it is bad too for culture, including fiction.
When there is a good investigative article, I want to read it, of course, but not only: I also want that other people can read it, and be convinced by it. Therefore, when I pay for an article, I do not want to pay just for me, I want to pay for everybody. And the same applies to the other examples: I can read an awesome novel or watch an awesome series, but if my friends have not read/watched it, I cannot discuss it with them, and the fun is hugely diminished.
This is becoming a problem because of dematerialization. When we needed a hard copy to consume some information goods, I had to pay for mine, I could not pay for yours. But at the same time, once I had a hard copy, I was allowed to lend it to my friends.
The core of the issue, of course, is copyright, because it is the legal tool to make information goods scarce, the precondition for the everyone-pays-for-oneself business model.
Copyright was invented to let authors have bargaining power against the publishers (and theatre troupes), but nowadays it has been completely perverted: the likes of Apple and Disney use it to leverage a few hugely popular assets they have in exclusivity to force customers to subscribe to their own service, even though it would probably be widely redundant with services they already have.
I acknowledge and respect the need and right of creators to get material gratitude for their works, but I do not accept they have the right to limit who can enjoy it. Nobody should be allowed to monopolize culture.
In the short run, we can probably fix that with mandatory sharing of catalogues, possibly after some embargo time. In short, now that it's been six months, I should be able to watch Baby Yoda with my Netflix subscription, with the provision, of course, that if many more people watch Disney shows on Netflix than Netflix shows on Disney, then Netflix has to pay Disney.
But in the long run, we need to acknowledge that the system is broken and fix it.
We have to get rid of copyright and replace it with something better.
I understand that this blog leans a lot towards libertarian ideology, which I find rather enjoy, but also that many people confuse libertarianism with propertarianism, which is a tragedy. The libertarian-propertarian solution for issues like this is to expect people to invest what they no longer pay in taxes into their own vision of the public good. To invest in subsidizing a prediction market for investigative journalism, maybe.
Unfortunately, the urge to invest in the public good, if we ever had it, has been negatively selected by two thousand years of mercantile society. Building that kind of mentality would take generations, and only if there is a massive adhesion to the project.
What we have now to foster the public good is Democracy and the Public Service. If we want to fix culture, we need to work with it.
The way the Public Service handles culture is libraries.
And on this, most of the public who enjoys culture and most of the authors agree: libraries are awesome. Libraries spread culture, libraries increase social fabric.
And libraries, in some jurisdictions, have an interesting feature: Public Lending Rights. This is money that is sent directly to creators when their works is borrowed a lot, even when no new copy needs to be bought because of wear-and-tear. Neil Gaiman, among others, has many good things to say about Public Lending Rights.
We could democratically decide to create the Universal Electronic Library: everything that is offered gratis on the Internet. Blogs. Journal websites, when they are not paywalled. Torrents! That means no prosecution for “piracy”, of course, but more importantly it means adequate funding for Public Lending Rights on what gets “borrowed”, i.e. browsed/downloaded a lot.
There would a few technicals hurdles to solve, but not that much. How do we know what gets perused a lot? I think just asking people would mostly work.
There is the issue of funding, of course. For that, I suggest a modern equivalent of the idea of a Land Value Tax transposed to intellectual property.
This idea of an Universal Electronic Library, made from everything gratis on the web and subject to Public Lending Rights could, I think, solve a lot of the problems we have with the funding of culture and information goods in general. Spread it!
If an organisation is subsidising a prediction market in a specific area that may warrant investigative reporting, then they are subsidising that prediction market in order to encourage investigative reporting (in a world of free online content). If this is the sole goal of subsidising the prediction market, why don't they just use the money to directly fund the investigative unit of media outlets, which can then investigate things they think are worth investigating?
I think that subsidising a prediction market would be more useful than the above only if said prediction market was very, very broad. For example, a prediction market in "Non-voting occurrences at schools". Various 'events' (which are still somewhat broad themselves) can be placed within this market (e.g. "Sexism scandal at school X", and "Shooting at school Y", and "Finance scandal at school Z", etc.). As the prediction market gains liquidity, people will begin to place bets across different events within the category. It would take longer to build liquidity, but is the only way I could see it being more useful than just directly funding media outlets over time. I don't do much betting, but I think this is pretty much how it works in sports betting markets.
Two points: 1) FYI you may be interested to know that the problem raised in your military school example is partially solved by the litigation finance market. Basically, this is the market of cross-collateralized debt financing for law firms, often used for financing portfolio's of contingency cases. It acts to facilitate litigation of uncertain claims like your example. If these suits end up public and are notable then the work of the lawyers will trickle into new, and at the least they provide a stick against bad actors for victims who otherwise might not have a means of legal organization. Suboptimal in many ways, but a partial solution to the problem. 2) as Luke G and a couple others have mentioned, P-markets are only politically stable long-term with strict insider trading regulation similar to that of public financial markets. With no insider-trading regulation you end up with a major alignment problem that makes p-markets a political non-starter, and which also would make many organizations less functional. With proper regs, you provide outsiders with the right incentive to uncover information, without actually corrupting the incentives of the underlying institutions. This also includes dishonest market manipulation i.e. you can publish a short report that you believe to be accurate, but you can't (as an investor) legally publish a short report with information you know to be wrong with the intent of damaging the price of a security. Ideally, these markets enable information discovery and accountability without incentivizes participants to encourage bad outcomes
Is the Eric Adams example correct? If investigative reporting lowered the likelihood of his election by 1-2% (as reflected in a market), I don't think a reporter could realistically monetize the $250-500k swing. Wouldn't that 250-500k be spread across the entire "No he won't be elected" side of the market? I think so. The better example is that a news organizations invests, let's say $100,000 on the NO vote side at a value of $0.25. When that moves to $0.27, they'd make a profit of $2k, right? So, it's not really funding a reporter. Plus they'd need to come up with the $100k and run the risk of being wrong or of no one caring about their reporting.
But the news itself causes the market decline regardless of its truth, for examlle, publish horrible lies about an election candidate and their prespects drop.
I dunno. The underpants gnome business model had a ??? step. This one seems sketchy.
Is there a concept of insider trading on prediction markets? Also, I can see this type of bets going very wrong on things like "elections" (everyone with money on A. Hitler winning the election is likely to vote for him no matter what).
I am sure that this is addressed somewhere and that I have missed it though, as mine looks like a "day one" objection to prediction markets and everyone else is very excited about them.
How much liquidity can/will there be in a life-satisfaction-of-female-students-at-this-particular-school prediction market?
"In a prediction market, once you're wrong a couple of times, traders will stop updating on your reports and you'll lose most of your power to move the market."
I don't know about this. Does this happen to people who try to pump and dump stocks? Why would it be different for prediction markets?
> If we've gotten so polarized that no conceivable fact could change who we vote for, then investigative reporting should have the same status as golf
I think this is underselling the value of investigative reporting. Suppose I'm an American reporter investigating Putin or Xi. There's value in my reporting news about them, but not many immediate results. Perhaps the prediction market could measure "American sentiment toward China" or something like that, but I suspect that on predictions that large any given investigation is too small to make the averages pay out in a reasonable timespan. If my investigation makes the result 0.1% more likely, then my newspaper only gets the payout after many investigations on average, making it probably not worth their while. A national-level newspaper like the NYT or the Washington Post might be okay, but anyone smaller won't bother.
In addition, this means that a newspaper could only have one position on a given issue. Suppose I'm a nonpartisan newspaper who wants to investigate the election. If I accurately report on both a Republican-favoring issue and a Democrat-favoring issue, then the election odds are the same as before, but the voting population is better off. This would encourage partisanship in reporting, which seems like a poor result.
"I'm looking at the front page of today's Washington Post, which has something about how there's a culture of misogyny in some important military school. This could charitably be termed investigative reporting - we can imagine the journalist painstakingly tracked down women at that military school, heard their stories, hunted down documents proving that this school discriminated against women in some way, et cetera."
Why "charitably"? Paywall and all that so I'm at a disadvantage, but this sounds like a central example of investigative reporting, and you seem to be straining to "imagine the journalist painstakingly tracked down...", which seems like exactly how I'd expect that reporting to work.
I like this idea, but running some numbers - it might work out? (My original claim was that it didn't, but on second thought I'm not sure).
Let's go with the military school example. A damning study on inequality there might move betting markets on the firing there by, say, 10%. So if you're willing to pay an investigative reporter $x for that story, you need to invest $10x into the betting market to have enough liquidity for this to be worth the reporter's time.
But it's actually worse than that - most investigative reporters' investigations will (hopefully) not turn up firing-worthy offenses. so if say 1/12 month-long investigations come up with something, x needs to be about a year's salary for a reporter. And this 10x (which is now on the order of a million dollars) isn't buying you improvement, it's just buying you insurance that you'll probably catch anything too egregious going on in that particular guy's watch (since you have to get this insurance for every single employee).
So putting all these numbers together - you'd need about a million dollars' worth of insurance per senior employee, that would cover all-cause scandal, and would pay out if the betting market resolves (I'm assuming the school takes most of the money if they don't fire the guy). That... Actually seems kinda doable, I think, at least for senior employees.
Won't this run into the same issues highlighted in Inadequate Equilibria for multi-stage funding, where you're rewarded not for predicting what will happen, but for what other people will think will happen given some news? Ie. the longer the time horizon in play, the more a market has to worry about whether it's predicting outcomes or "just" future market sentiment.
Life satisfaction markets seem dubious to me. Could it incentivise companies to avoid hiring people with histories of depression? Also, could a bad actor "short" such a market, then attempt to reduce the life quality of the market's subjects by nefarious means?
Hmmm ...
1. It helps the negative or expository reporting, but less ones which are in depth views of topics which don't necessarily have a fireable culprit.
2. Stock markets work because they normalise equities (same re other mkts), each equity looks/ feels/ smells roughly similar to others. That's the hard bit in moving the concept to more esoteric markets.
3. The infotainment industry prediction would sure make TV a hell of a lot more fun. Or at least CNN ! I'd even settle for realtime betting alongside news broadcasts.
That said, I like the idea of fire the CEO markets, even if illiquid, though I'd also note that Glassdoor exists, often with horrible CEO reviews, and the Board rarely takes that into account barring properly illegal conduct.
All in all, I'm very supportive of prediction markets. However, I do see a key flaw here.
Right now, the media has a vague incentive to be partisan - it helps them get more clicks and slowly builds up an audience with a consistent ideology. I would say that this is Not Great - the utility that most people get out of the media is unbiased information, which partisanship is at odds with.
With prediction market being the key to fund them, they have an explicit, definite incentive to be partisan and write as many hit-pieces as possible, take every quote as out-of-context as possible. What if Cade Metz could make $50K, contingent on Scott's blogger approval rating decreasing by 5 points?
Interestingly, this flips the onus of The One Unbiased Truth from the old school media to prediction markets. Culturally, it might take a bit for people to get used to changing where they get unbiased information from, but I'd argue we're going through that already.
> Very few of the people who cared about Watergate bought subscriptions to the Washington Post
Are you sure - breaking such large controversies, in a pre internet era, could significantly drive purchases and subscriptions of newspapers. It also dramatically raises the status and just direct power of the investigators, which is its own motivation and desire.
Also this seems to be precisely the same as insider trading
> Maybe then we should just accept that the market is correctly telling people not to bother investigating Adams
How on earth is this a good predictor of whether or not someone is corrupt? Even if it’s “the sum of the knowledge of the smartest people everywhere” I don’t see how a bunch of common market participants will have any clue whatsoever how likely it is some particular guy out of a million is corrupt or has dirt hidden, and even then how that helps you investigate (investigation is something you do about a person in particular areas for particular reasons, not just abstractly, you have to know what about them to investigate and why, and also there are politics involved that probably drive this stuff way more than abstract likelihood and price)
Also this is legal? One guy in a few weeks could create a prototype investigation market or ceo firing market on the internet or using one of the many crypto platforms.
Is there any study that calculates the Brier score of all of PredictIt?
This incentivizes journalism which is effective at influencing outcomes, which isn't necessarily the same thing as accurate journalism.
I wonder how this would relate to stories with multiple possible impacts. Suppose someone does a story which turns up that Florida condos aren't inspected adequately. Does this win only if there's a deadly collapse? If inspections are improved and a deadly collapse is averted? Or a building is found to be in bad shape, evacuated, and repaired?
"This reporting has been brought to you by conflicts of interest!"
I am deeply skeptical of this proposal. If I learned that some newspaper today was shorting the corporations criticized in its pages, I would be shocked and appalled, and I would deeply discount its reporting. I don't think I'd feel better if everyone was doing it officially.
It's possible that in some future age our present norms against this sort of conflict of interest will seem quaint and absurd, like the historic prohibitions of usury. But I suspect not.
How did prediction markets become such a “thing” in the rationalist community? Is there some notable origin story?
Rationalists need to be able to explain why real markets failed to predict [insert Enron, WorldComm, or GFC example here] but how esoteric, illiquid derivatives will do better. Or why there’s liquidity problems in credit markets (the world’s 2nd largest asset class?) but a prediction market subsidy will work super great.
These ideas sound compelling but market construction is really hard. There’s still not a real exchange for bonds. Bond tickers have wide spreads and are thinly traded (despite Matt Levine’s smartass quips, bond market liquidity is a big issue!). The derivatives this community postulates over have similar complexity bond covenants so one would expect huge challenges in getting these ideas to trade to a point where they’re “predictive”.
I’m not going to address the batshit moral hazard opportunities because I expect others will jump on those.
Back when media was all broadcast over public airwaves, as a condition to use bandwidth media companies had to carry content in the public interest, which mostly meant news. And, this mostly meant the news bureaus were quite independent. So, this journalism was essentially publicly subsided (the FCC could have charged more for bandwidth use otherwise).
Now that media delivery is all private, this doesn't work. Fully supporting news with advertising and treating at a profit center is problematic. So, I say go back to directly subsidizing journalism. Support local newspapers according to the population. Establish a national board of journalistic standards, much like those for medical standards, and yank the subsidies if they violate them. Etc.
As you say, it is a public good. Why not treat it like other public goods?
The problem with this is that it only works if there is a lot of money in the prediction markets. But you can only get there if people already care a lot about prediction markets and put a lot of money in them. Why would any investor want to put money into anticipated life satisfaction for a random demographic slice, when they could put it in index funds, or even more lucrative prediction markets.
Also I'm not sure that the math works out for how much money you need in the system for it to work. How much money would need to be being bet on military school principals firings for a swing in the numbers for a particular one to be big enough that it's enough to fund anything meaningful?
I love the idea. Towards its feasibility in the news space, it might be necessary to invent defences against litigation. Media organisations currently can't expose every piece of material that comes their way due to legal costs. They have to pick their battles, which is part of the reason the best investigative reporters like Sy Hersh didn't stay in one place for long. (It also had something to do with his uncompromising personality)
Perhaps it'd be possible for power structures or corporations to tie up reporters in legal battles while denying them funds from their prediction market win, pending court ruling, which could take years. WikiLeaks is a great example of how far a financial blockade can be taken. It may be necessary for a reporter using prediction markets to rely on donations for legal battles in the short term post-release, or maybe it'd be possible to operate through some third party organisation built specifically to protect them in this way.
Another problem to solve would be getting people to take notice in the first place. There've been revelations about fossil fuel companies and the DOJ recently that haven't made a dent in the public discourse because traditional media won't talk about it. Mainstream signal boosting these days depends on being highly partisan or something both sides agree on. A report that harms a bipartisan narrative (such as the US' involvement in Venezuela being defined as "aid", or most other pro-empire narratives) is fighting a massively uphill battle.
No thanks. We should not be funding the Woke Olympics in this way. In fact, this sounds like it could accelerate the "find people who have heterodox opinions and cancel them" movement. I do miss the investigation of things like mistreatment of animals in factory farms. However, the cynical part of me knows that this would be used to attack people with minority. We shouldn't be encouraging paparazzi to "investigate" Chris Pratt and find that he's a devout Christian in order to maybe attack Disney's next film.
This scheme incentivises investigations, and then to reveal the new juicy info in a way that reliably changes what is knowable by people with money who honestly want to find the truth, so that markets adjust. Note though that this isn't the same as convincing the mainstream or publishing honest articles in NYT. We know already that there can be a big disconnect between what is knowable and what info reaches the mainstream. But one can hope it pushes in that direction at least.
It's an interesting concept for sure. I wonder if it could work for all public office holders. Public corruption is where the stakes are highest for citizens, and it's best investigated by local media.
https://www.denverpost.com/2021/04/02/colorado-judicial-department-sexism-allegations/
I'm very glad you're writing about prediction markets, as they sound like a great possible answer to the messed up situation right now with public narrative machines, and the absence of a functioning entity that does what 'the news' is ostensibly doing.
Hanson's fire-the-CEO-market is based on the stock market price of the company, conditional on the CEO being fired or not. It's not a bet that the CEO will in fact be fired, although certain differentials would suggest that's what the board should do. So a fire-the-head-of-military-schools market would need to be based on some measurement, and it's unclear whether a poll of the satisfaction of students is actually at all a desirable measure. Student evaluations of university teachers are notoriously awful and just incentivize easy classes rather than actual learning.
Intriguing, but I would distinguish between incentives to inform, and incentives to convince (or provoke, or startle). My main concern would be that this system may incentivize actors to promote a *perspective or belief*, which doesn’t have the same positive externalities as detached investigative journalism.
Once you’re wrong a couple of times people will stop trusting you and your predictions will stop moving markets. A perfect incentive to go Big, even if there’s a chance you’re wrong, because you can cash out even an eventually wrong position if you can move the “stock” down, but you might turn out to be wrong later and people will stop believing you, so bet big. Incentive for over-confidence.
What happens when every voter in Tinytown, Randomstate receives $100 worth of shares in "Candidate 2 wins" anonymously in their mailbox?
People not valuing true information is a pretty well established problem, and I can see how prediction markets would incentivise people to value it more. But doesn't it suffer from a problem where the more professional the market becomes, the fewer people can realistically make any money from it? Only people with new unknown information can do that, and only people who expend a lot of resources can find new unknown information. It seems like there wouldn't be much of a reason for ordinary people to participate in or care about these markets.
I think the solution we should be looking for looks more like a culture where most people care about true information, not one where most people don't care but a few people do and make a lot of money out of it. Arguably, the latter is a pretty good description of what we already have.
This sounds like a brittle, complex, fraud-prone way of allowing schools or other institutions to offer financial bounties for verified reports of misconduct. Which they can do already if they want to.
I feel like prediction markets are Scott’s hammer for which everything looks like a nail.
Really great to see this idea raised, and Hanson's name invoked in context. He is likely the most knowledgeable person around on the subject of Prediction Markets. We talked about a very similar idea recently... there are a number of investment firms built around a very similar model as Hindenburg's, but focused on the fact that China's economy is what I like to call 'structurally inefficient'. (In very rough economic terms, efficiency means that the value of any given firm or entity is based on the public having relatively accurate knowledge about its activities and therefore its market value. It has been widely observed that the structure of Chinese bureaucracy means that any given layer of management is incentivized to give inaccurate reports of activity and value to those above it... this view has strong explanatory power in the context of 'ghost cities' and firms that have recently been delisted from western stock exchanges for deeply structural reporting fraud. Thus 'structurally inefficient'.)
Hanson observed that a prediction market that incentivized people lower in the bureaucracy to report true information relevant to market valuation (say, with untraceable cryptocurrency) could very well work, but as with all Prediction Markets, it would need someone to pay for it. Reading between the lines, I took this to mean a 'first mover' to kickstart such a market. Of course, something along these lines could be just as interestingly disruptive to Western companies as to those in China.
Anyone looking to throw money at the problem? Maybe a Kickstarter campaign? Maybe something something NFT market something?
> life-satisfaction-of-female-students-at-this-particular-school prediction markets.
I still don't understand why the most obvious outcome of this isn't people terrorizing the women at a given school in order to make a ton of money predicting they'll be very unhappy? That seems a lot easier and more reliable than doing a ton of investigative reporting and hoping that there's actually something to find.
Or, more generally: if you can make $25million by finding a scandal about a politician, can't you make $25million by inventing a convincing fake scandal? The 'infotainment industry where people try to get clicks by accusing each other of being racist pedophiles' seems like it would be more lucrative than investigative reporting, unless we somehow create a society where people do not respond to those accusations in ways that affect things measured by prediction markets.
I'm not quite prepared to say this is the worst idea of yours I've read, but it certainly must be close.
I feel like you to some extent invented Bloomberg. Which isn't a bad thing to invent, since Bloomberg is something like a $20 billion company with $10 billion in annual revenues.
On the other hand, the stock market as a whole is valued around $40 trillion. Apparently Bloomberg has a few competitors like Thompson Reuters that are maybe collectively another $10 billion in revenue. So that would suggest that you can get about $1 of analytics value for every $2,000 of trade volume.
I can maybe believe that there's a future with predictive markets on whether the president of (say) Iowa State will be fired in the next year. I really struggle to believe, however, that more than, say, $1,000,000 could be invested in the prediction market (already double the president's salary). So that suggests about $500/year in potential investigative reporter revenues--to generate $100,000 in revenue, you'd have to provide a Bloomberg-esque product for 200 million-dollar questions.
For questions like "is there a culture of misogyny at school X," this seems unlikely. The reason the Washington Post cares about this isn't that their readers had a preexisting interest in school X, it's that they had a preexisting interest in cultures of misogyny, and are interested to hear about examples of those cultures. To the extent that that sort of reporting is useful for making predictions that the readers actually care about, it's in a very general, world-view building way, not in a specific, directly actionable way.
Apart from all the usual issues with prediction markets, the failure case here is that you can't have prediction markets for every possible scandal, and you certainly can't have liquid ones.
I don't think ideology really works like this. I've even shown obvious lies within the Rationalist community and only a handful of people updated their priors.
ok dis libertarian scott
Short reports don't have to be even remotely true to make a ton of money, and the first amendment in the US makes it next to impossible to successfully sue.
This is such a stunningly, obviously bad take that it's not worth taking the time to detail all the reasons that it is misguided and unworkable (and I assume the earlier comments have done so).
But what I did find interesting and worth commenting on is how this pieces beautifully illustrates the shortcomings of libertarianism. Investigative reporting is a public good. There clearly isn't a market mechanism that will fund it, but there are plenty of non-market alternatives one can imagine from non-profits to rich benefactors to government subsidies. The fact that this essay instead tries to posit this utterly ridiculous predictions market scenario is telling. It's a sign of someone who is unable to confront the shortcomings of their ideology.
Are we measuring instantaneous life satisfaction ratings, or aggregate life satisfaction ratings over the next N years? I can boost the former by abolishing tests/grades/classes and subsidizing parties. The latter will tend to be less liquid because there's a lot more opportunity cost to an N year investment. If N covers the entire lifetime of the students, most of the people with money to spare for prediction markets will be dead before they get paid. There's a tradeoff between the liquidity and the meaningfulness of the market.
As the saying goes "If you love something, put a financial derivative on it".
Do we really want to live in a world where it's common to have a prediction market on you, specifically, being fired? I realise this already exists for public figures, but extending it to a vastly higher percentage of society seems like a huge social negative. Worse, it would obviously be gamed - as soon as we start to measure something, it becomes useless for its original purpose. The unintended consequences of this idea would be massive.
If investigative journalism is a public good, why not support it with grants?
I'm not sure how you would allocate it specifically, but given the incentives market based journalism already face, adding more market seems like a weird solution.
> For example, our current model incentivizes fake news - if something is exciting and outrageous enough, it'll sell a lot of papers, and even if a few spoilsport researchers come out a few years later with definitive proof it was wrong, by that time nobody will care. In a prediction market, once you're wrong a couple of times, traders will stop updating on your reports and you'll lose most of your power to move the market. You'll have no particular reason to care whether what you write is outrageous or exciting, only that it's true and relevant to something that people care about.
your model incentivizes fake news equally as well. Remember we're not betting on "Candidate X did something bad" we're betting on "candidate X is not going to be reelected". You can prevent reelecetion equally well with sufficiently convincing fake scandals as with real scandals. If a fake story sells the papers, there is no reason to assume it won't move the polls.
I could see how (by keeping the money at stake low enough) you could prevent companies from shorting and then firing their own CEO, but how would you stop self reported levels of happiness from being gamed by the students given that they have the ability to self organise and there's nothing else at stake.
As usual, you did predict my main concern with this, but I feel like maybe you didn't take the negative outcomes far enough. This doesn't seem dangerous while we're talking about agents as large as investigative reporting teams, but this (potentially) degenerates into individuals doxxing each other for the profit they can make off any drop in status. At that point, we're all suddenly living in the hellish world you found yourself in for a few months last year, where some people can make a few dollars publishing things about you which you'd rather they didn't. I want to be very careful with this line of critique though, because at the end of the day, it IS just a slippery slope argument. I just wanted to offer the hopefully balanced view that I don't think we know how slippery it really is, or how unpleasant the bottom of that hill. Who knows, maybe it's still a better world, due to improved investigative reporting being worth more than whatever is lost.
All that said, I do really like the idea of this. It seems to reward positive behaviour, at least right up until it potentially doesn't.
I like the system, and would like that it be tried. But I think that the potential problems are larger than what you make it seem. For instance, prediction markets are markets for entropy & chaos, i.e., you are very incentivized to cause unexpected things to happen.
A nice idea, but there are some serious issues:
a) If you subsidize markets (as suggested by comment below), you also subsidize fake news. If you can move "the market" with a bit of fake news, and "the market" is actually one trader + millions in subsidies then that's free money, basically. I mean we do have a market for news, and fake news seem to be doing quite well. Maybe it isn't "about" credibility?
b) Insider trading. This isn't as bad as everyone thinks (it was legal in parts of Europe until quite recently), but probably not great for liquidity. But also, the kind of insider trading where the school principal shorts his stock for millions and then goes on a "I have fuck you money now, fuck you all" spree.
c) Weird incentives to move markets as opposed to delivering the truth. We already kind of have these, but I'm kind of happy with the current situation where a lot of the smart people are happily modelling things in finance instead of being worried about how to manipulate prediction markets on things that actually matter.
"Then everyone could just ignore the first thing and go about their lives" that's _got_ to be a typo, surely you meant that everyone would ignore the _second_ thing, the way they do now. But yes, this totally reminds me of the Court of Values and the Bureau of Boringness and seems equally a good idea despite the tongue-firmly-planted-in-the-cheekness https://meaningness.com/virtue-court
"I want a world where protesters march into colleges and ask them why, if they claim to care about female students, they don't have a subsidized female-student-life-satisfaction prediction market, so that they can credibly set targets for gender equality"
Yeah, that's just what we need, discriminating on the basis of gender. I thought that was illegal. Oh, right: the law may be written that way (the Civil Rights Act of 1964 and numerous other laws) but we all know that it now means, discrimination is illegal if done by the wrong people for the wrong reasons but legal (and wonderful!) if done by the right people for the right reasons.
What a big, stinking pile of pigs**t.
"Even if their dirt isn't quite that good - even if it only makes Adams seem a little bit worse than people thought - it would probably affect his chances by 1 or 2 percent. And that's still $250K to $500K - more than enough to hire a couple of good investigative reporters."
The market math doesn't work this way. The journalists can't trade $25M without moving the price. Perhaps they could trade $500K on the Adams bid, but I suspect it'd be closer to $50K before they moved the price more than 1%. I think their best case on such a story is ~$10K.
> In the far future, the news might diverge into an "infotainment" industry where people try to get clicks by accusing each other of being racist pedophiles, and a prediction-market-movement industry where people try to generate true information on important issues. Then everyone could just ignore the first thing and go about their lives.
I predict that the strong majority will prefer the first, and the desire (both top-down and consumer-driven) to appropriate some of the latter's prestige by aping the format will blur the lines for all but the most diligent consumers. I predict that the commercial success of the infotainment will be an inevitable draw for the market-makers, and some of the compromise will come from that direction, too - gamification of results, clickbait to promote certain markets, maybe outright corporate consolidation with the editorials.
In short, I predict that even if one could snap their fingers and establish such a division, it would largely collapse within a few decades. But it won't get that far, because those forces would already be at work when trying to set up such a system.
Or by analogy - how much do you care about the extant division between reporting and opinion? You'll get more of that, for better and for worse.
Problems I see with this:
1. This only funds journalists who end up finding controversial stories. Lets say that I'm a reporter and I get a tip that politician X had a sketchy affair; in the current model I can safely investigate it since I know I'm salaried and I'll get paid either way. In your model, I get substantial compensation (likely) if I find out that the affair actually happened, but nothing if I find out that the tip was bogus, or if it wasn't bogus, but I just can't prove it. Beyond the substantial pressure that would put on reporters to just claim the most controversial interpretation of the facts it means that reporters would be incentivized to only follow leads they're pretty sure are already true. It's a good thing when a reporter spends 4-5 months investigating a rumor and finds out it isn't true and your system doesn't reward that.
2. The current system has an maximum compensation ceiling for lying reporters. Sure, if they lie well enough they'll get a promotion or two and maybe an award, but in bounds with their existing compensation levels. In your system, however, the right report at the right time could be worth hundreds of thousands or even millions of dollars depending on the market. If I can do that a few times, hell, if I can do that even once I'm set for a while. Sure, maybe my stuff gets discounted in the future, but who cares? I've already made bank. This incentivizes everyone to be maximally controversial at all times. The normal reaction to that is, "ok, let's just pay less attention to controversial reporting," but now you've set up a system where controversial reporting is dismissed out of hand.
"Why should I believe this report on Eric Adams accepting bribes, the paper that revealed it has a huge financial stake in him losing the general election after all!"
Shit, prediction markets might incentivize forcing some really bad outcomes, right?
Say it's possible to sway a nation's election by burning that nation's crop fields -
Basically, we should never have big markets about outcomes that can be greatly influenced by cheap destruction.
You may be interested in the book "Democracy's Detectives: The Economics of Investigative Reporting," which discusses this issue (https://reason.com/2017/02/22/where-did-all-the-investigativ/)
“But it's hard for investigative reporters to capture the value they produce.”
"It's hard for them to profit from the damage they inflict." FIFY
Fusion GPS figured out how to fund investigative journalism. You just work on commission generating dirt on a target that someone with money wants to take down. IIRC, I think they got something like $25 million from Hillary and the DNC to set up Trump with the Steele Dossier (or maybe that was just Steele alone).
The dirt little secret is that 95% of investigative journalism by the Legacy Media always consisted of just getting data dumped in their in-boxes by someone with a vested interest in having them publish it. The main sources are: (a) Plaintiffs' attorneys giving you the litigation files they accumulated through court-ordered discovery in order to help pressure a defendant corporation into settling; (b) official government operatives trying to spin the public narrative; and (b) disgruntled "whistleblowers" who are trying to either CYA themselves or take down their enemies.
Woodward and Bernstein's epic "investigation" consisted of having a guy name Mark Felt contact them and give them the whole story to print, because Felt was pissed that Nixon passed him over for a job. https://en.wikipedia.org/wiki/Mark_Felt
> In a prediction market, once you're wrong a couple of times, traders will stop updating on your reports and you'll lose most of your power to move the market.
Is that true, though ? I understand that, currently, prediction markets work more or less this way -- but the current prediction markets are niche, boutique affairs. What happens when they go mainstream ? There does not appear to be any mechanism that prevents prediction markets from succumbing to the same failure modes as the stock market -- with all the insider trading, CDS tranches, high-frequency algorithmic trading, and a myriad of other financial instruments and tricks essentially designed to decouple the value of the trading token from any real-world considerations.
Wouldnt a social credit system work better for most of these problems?
Kalshi, the first regulated prediction market is now live in beta. Pretty awesome check it out. Only does yes/no questions for regulatory reasons.
https://kalshi.com/
Questions:
Will the Tokyo Summer Olympics be cancelled? Y: $0.05
Will the target federal funds rate be above 0.25%? Y: $0.08
Will more Americans get vaccinated this week than last? Y: $0.99
Will real GDP increase by more than 6%? Y: $0.87
Here's the link to Scott's post that mentioned previously.
https://astralcodexten.substack.com/p/metaculus-monday-2821
Scott, if you're interested in chatting w/ the founder feel free to ping me at rex dot salisbury at gmail.
You could pick a group of institutions and bet against any member of the group being successfully prosecuted by whatever authority for whatever crime you think might be happening.
That is kind of how mandatory reporting by public companies is supposed to work.
It seems like this is a good way to fund investigative research, but also to amplify little-known stories and to fund negative advertising that has no basis in fact. Not checking the facts is cheaper.
1) Short the relevant positions
2) Publish report that is subtly but fatally flawed
3) Earn money
4) Switch to long positions
5) Wait for the inevitable backlash
6) Earn more money
"Suppose some kind of media property uncovered evidence that Adams took bribes or committed some other career-ending indiscretion. In a world where prediction markets were as liquid as stock markets, they could short Adams, reveal his dirty laundry, crash his election prospects, and collect some portion of $25 million."
What a bad idea, and a surefire way to push journalists into smear campaigns and moral outrage "exposes" even worse than they already do today, which hurting politics and business in the process...
"In the far future, the news might diverge into an "infotainment" industry where people try to get clicks by accusing each other of being racist pedophiles, and a prediction-market-movement industry where people try to generate true information on important issues. Then everyone could just ignore the first thing and go about their lives."
We sort of already have this; the financial press (e.g. FT, WSJ, Bloomberg, Reuters to some extent) is generally much more sober and reality-oriented than most of the rest of the media, although their focus is obviously on existing markets. I suspect the FIRE sector is the last real market for non-sensationalist factual news.
As someone else mentioned here in the comments, though, the financial press is much smaller than the markets themselves. Of course, the public financial press cannot profit in the markets by using their own research, while the banks, hedge funds etc. (that probably do more investigation of markets overall and commission a lot of private research) can, and are much bigger, but I don't think much of what they uncover is classifiable as investigative reporting. That you have a small, in principle unbiased public financial press and lots of large profit-seeking institutions that keep their research private for market advantage is unlikely to be an accident. For instance, I suspect that in most markets it better to go for trades on things that don't need 2+ years of investigative reporting to maybe pay off.
A good example might be the Wirecard scandal. More than one hedge fund voiced their skepticism, published reports outlining problems, and shorted the stock, but it's to find info on any of these original shorters making much money, since there was no consensus on the reports being true. In the end, it took several Financial Times reporters years of digging, travelling to various places in Asia, meeting with established contacts, following up on leads etc. in order for their evidence to be compelling enough for an audit to be commissioned revealing the fraud, pushing Wirecard into bankruptcy.
All during this time Wirecard was making both public and private accusations of the FT working with short sellers, even getting the German financial regulator to open investigations into the main reporters on the story. This was not a very credible accusation, although not totally impossible. But how would the story have been received if the FT had said "Yes sure we're shorting Wirecard in order to found our investigative journalism", even if that hadn't been illegal? Even if that had been fine, for years there was no clear evidence that the irregularities at Wirecard were anywhere large enough to tank the company. Would it have made sense for the FT to dedicate years of work on this if the highly uncertain payoff of shorting Wirecard at the opportune time was the only possible reward?
Investigate reporting depends on sources willing to talk. Why they do this can be complicated - to feel important, to fight their own battles within the organization, etc. - but a big element is that blowing the whistle feels like a noble and righteous action. This turns it into something more like insider trading. Perhaps investigators will still find sources by offering sufficient cuts of the booty. But I don’t think anyone will be compelled by conscience to share information with someone who intends to profit from it. And their peers will _definitely_ see them as selling out. To say nothing of the increased investives to lie.
It’s not clear to me why this would approach Truth rather than “what everyone happens to agree is true” (but isn’t)
I wonder if that would create an incentive to cause things to be true? I can try to foment dissatisfaction with Eric Adams or the head of a military school - I don't even necessarily have to lie. That might be cheaper and more reliable than investigating him in case he's dirty.
TL;DR: investigative journalism could be funded through the Universal Electronic Library.
It's an interesting idea, one that clearly deserves to be investigated further. But I believe it can only work for a limited subset of kind of investigative journalism. To find a solution, I believe we need to look at the wider picture, and I think the core issue is this:
The everybody-pays-for-oneself model is bad for information goods.
This model is bad for journalism, it is also bad for science — ask any scientist about SciHub —, it is bad too for culture, including fiction.
When there is a good investigative article, I want to read it, of course, but not only: I also want that other people can read it, and be convinced by it. Therefore, when I pay for an article, I do not want to pay just for me, I want to pay for everybody. And the same applies to the other examples: I can read an awesome novel or watch an awesome series, but if my friends have not read/watched it, I cannot discuss it with them, and the fun is hugely diminished.
This is becoming a problem because of dematerialization. When we needed a hard copy to consume some information goods, I had to pay for mine, I could not pay for yours. But at the same time, once I had a hard copy, I was allowed to lend it to my friends.
The core of the issue, of course, is copyright, because it is the legal tool to make information goods scarce, the precondition for the everyone-pays-for-oneself business model.
Copyright was invented to let authors have bargaining power against the publishers (and theatre troupes), but nowadays it has been completely perverted: the likes of Apple and Disney use it to leverage a few hugely popular assets they have in exclusivity to force customers to subscribe to their own service, even though it would probably be widely redundant with services they already have.
I acknowledge and respect the need and right of creators to get material gratitude for their works, but I do not accept they have the right to limit who can enjoy it. Nobody should be allowed to monopolize culture.
In the short run, we can probably fix that with mandatory sharing of catalogues, possibly after some embargo time. In short, now that it's been six months, I should be able to watch Baby Yoda with my Netflix subscription, with the provision, of course, that if many more people watch Disney shows on Netflix than Netflix shows on Disney, then Netflix has to pay Disney.
But in the long run, we need to acknowledge that the system is broken and fix it.
We have to get rid of copyright and replace it with something better.
I understand that this blog leans a lot towards libertarian ideology, which I find rather enjoy, but also that many people confuse libertarianism with propertarianism, which is a tragedy. The libertarian-propertarian solution for issues like this is to expect people to invest what they no longer pay in taxes into their own vision of the public good. To invest in subsidizing a prediction market for investigative journalism, maybe.
Unfortunately, the urge to invest in the public good, if we ever had it, has been negatively selected by two thousand years of mercantile society. Building that kind of mentality would take generations, and only if there is a massive adhesion to the project.
What we have now to foster the public good is Democracy and the Public Service. If we want to fix culture, we need to work with it.
The way the Public Service handles culture is libraries.
And on this, most of the public who enjoys culture and most of the authors agree: libraries are awesome. Libraries spread culture, libraries increase social fabric.
And libraries, in some jurisdictions, have an interesting feature: Public Lending Rights. This is money that is sent directly to creators when their works is borrowed a lot, even when no new copy needs to be bought because of wear-and-tear. Neil Gaiman, among others, has many good things to say about Public Lending Rights.
We could democratically decide to create the Universal Electronic Library: everything that is offered gratis on the Internet. Blogs. Journal websites, when they are not paywalled. Torrents! That means no prosecution for “piracy”, of course, but more importantly it means adequate funding for Public Lending Rights on what gets “borrowed”, i.e. browsed/downloaded a lot.
There would a few technicals hurdles to solve, but not that much. How do we know what gets perused a lot? I think just asking people would mostly work.
There is the issue of funding, of course. For that, I suggest a modern equivalent of the idea of a Land Value Tax transposed to intellectual property.
This idea of an Universal Electronic Library, made from everything gratis on the web and subject to Public Lending Rights could, I think, solve a lot of the problems we have with the funding of culture and information goods in general. Spread it!
If an organisation is subsidising a prediction market in a specific area that may warrant investigative reporting, then they are subsidising that prediction market in order to encourage investigative reporting (in a world of free online content). If this is the sole goal of subsidising the prediction market, why don't they just use the money to directly fund the investigative unit of media outlets, which can then investigate things they think are worth investigating?
I think that subsidising a prediction market would be more useful than the above only if said prediction market was very, very broad. For example, a prediction market in "Non-voting occurrences at schools". Various 'events' (which are still somewhat broad themselves) can be placed within this market (e.g. "Sexism scandal at school X", and "Shooting at school Y", and "Finance scandal at school Z", etc.). As the prediction market gains liquidity, people will begin to place bets across different events within the category. It would take longer to build liquidity, but is the only way I could see it being more useful than just directly funding media outlets over time. I don't do much betting, but I think this is pretty much how it works in sports betting markets.
Two points: 1) FYI you may be interested to know that the problem raised in your military school example is partially solved by the litigation finance market. Basically, this is the market of cross-collateralized debt financing for law firms, often used for financing portfolio's of contingency cases. It acts to facilitate litigation of uncertain claims like your example. If these suits end up public and are notable then the work of the lawyers will trickle into new, and at the least they provide a stick against bad actors for victims who otherwise might not have a means of legal organization. Suboptimal in many ways, but a partial solution to the problem. 2) as Luke G and a couple others have mentioned, P-markets are only politically stable long-term with strict insider trading regulation similar to that of public financial markets. With no insider-trading regulation you end up with a major alignment problem that makes p-markets a political non-starter, and which also would make many organizations less functional. With proper regs, you provide outsiders with the right incentive to uncover information, without actually corrupting the incentives of the underlying institutions. This also includes dishonest market manipulation i.e. you can publish a short report that you believe to be accurate, but you can't (as an investor) legally publish a short report with information you know to be wrong with the intent of damaging the price of a security. Ideally, these markets enable information discovery and accountability without incentivizes participants to encourage bad outcomes
Is the Eric Adams example correct? If investigative reporting lowered the likelihood of his election by 1-2% (as reflected in a market), I don't think a reporter could realistically monetize the $250-500k swing. Wouldn't that 250-500k be spread across the entire "No he won't be elected" side of the market? I think so. The better example is that a news organizations invests, let's say $100,000 on the NO vote side at a value of $0.25. When that moves to $0.27, they'd make a profit of $2k, right? So, it's not really funding a reporter. Plus they'd need to come up with the $100k and run the risk of being wrong or of no one caring about their reporting.
There may be some parallels to (and lessons from) litigation finance: https://www.epsilontheory.com/litigation-finance/
But the news itself causes the market decline regardless of its truth, for examlle, publish horrible lies about an election candidate and their prespects drop.