Does anyone have statistics on how many people are unable to get (COVID) vaccinated due to medical reasons? I’ve seen it brought up in arguments for continuing COVID restrictions, but I have yet to see anyone present hard numbers on what percent of the population actually can’t (rather than choose not to).
Regarding Delta, I posted this on my private social media:
"Most people I know have tuned out COVID, understandably. Most of you reading this are fully vaccinated and are living in regions that are reopening. If you are fully vaccinated and young, your odds of dying of COVID are about the same as dying from the flu.
All fully vaccinated people have to worry about today is governments reimposing restrictions due to the rise of the delta variant.
Prediction: If you live in a place where less than 65% of the population is vaccinated, you will face some level of restrictions in the upcoming months. Unless your local / federal government goes full French and decides to make life inconvenient for those who choose to not get the vaccine."
I still have to worry about COVID because I live in a place where less than 30% of pop is not vaxxed, including me (I shall get my shot this upcoming week), I have been fascinated by COVID since March 2020, and I believe some COVID prediction markets are or have been mispriced.
Going back to the comment above, why should most people worry about Delta? Other than long covid and future restrictions?
I wrote the review of Addiction by Design (casino gambling book) in the recent book review contest. Thank you to all the people who read and commented, and to all others who wrote reviews.
The big open question is what are the symptomatic rates (i.e. any symptoms, hospitalization, long covid, and death) for people that have been vaccinated and contract it.
Can anyone recommend resources for self-study of getting into meditative/spiritual yoga for someone who already practices Western-style yoga-as-physical-improvement? if anyone knows of something vaguely MTCB-like that would be idea.
Question about testing for Down syndrome, as done approx. 20 years ago. I'm sure that somebody here who can answer it.
Heard an account about a pregnant woman getting fetal genetic testing for Down syndrome approximately 20 years ago. She was told that test indicated there was a 60% chance the baby would have Down's, and that even if the baby did not he would likely have intellectual and psychiatric difficulties -- I believe ADD and learning disabilities were mentioned as possibilities. She chose to keep the baby, and he was born with no sign whatever of Down's, and never developed any of the usual signs. However, he has had substantial psychiatric difficulties (although not of the kind the doctors warned about: he is quite smart, and did well in school until various kinds of life chaos interfered with his schooling).
So here are my questions: (1) What kind of test result would lead the docs to say there was a 60% chance that the fetus had Down's? I know that Down's is trisomy 21, so I'm guessing that examination of the fetal cells showed some normal cells and some with 3 copies of #21. So how did the doc arrive at the odds at 60%? Would that be in a situation where 60% of the cells showed trisomy 21, 40% did not? (2) Does the second part of what the doc reportedly said make sense -- the prediction that even if the baby was not born with Down's, he was likely to have intellectual and psychiatric impairments? What possible result of the Down's genetic test would lead to such a prediction? I know about "mosaic" Down's, but it sounds as though people with that version of Down's would still be considered to have Down's, just a somewhat milder form, so I don't think that could have been what the doc was talking about. He apparently was talking about a whole separate problem the child was likely to have if he did not have Downs.
Do you like audiobooks? Do you prefer listening to audiobooks over reading a physical book (or Kindle)?
There's certainly a convenience factor; you can listen to an audiobook in a car or with your eyes closed. Beyond that, I find that if I'm trying to learn something audiobooks are slow and the inability to jump back and forth is inconvenient; while if I'm simply reading/listening for pleasure (often to a book I have already read) the slow pace can almost transform the story.
On the Covid March 2020 moment point, one thing I don't really understand is why the virus (appears to have) peaked and then plummeted in India.
Infections skyrocketed from about 11,000 cases per day in February to an eye-watering 391,000 cases per day in May (all stats here from ourworldindata.org). At the same time, the share of positive tests went from 1% to 22%, suggesting the true increase was much steeper than even that suggests. But days after it hit that peak, numbers fell right back down to around 40,000/day and 1% positivity today. The fact that positivity rate and deaths followed the same pattern suggests to me that this isn't an artifact of testing or bad data (though open to hearing otherwise).
Yet with a cumulative 31 million confirmed cases, even with massive corrections for incomplete data it seems a stretch to suggest that basically everyone who could get it got it. And while I'm far from an expert, the shape of the progression also puzzles me-- cases were going steeply up about a week before they started going steeply down, rather than the more gradual leveling off I'd expect if the susceptible population was just gradually petering out (glad to be corrected on this point if mistaken).
It definitely wasn't vaccination-- proportion of fully vaccinated individuals is around 5% even now. The main narrative I read is that while federal restrictions were haphazard, many state-level responses such as Maharashtra's/Mumbai's were quite effective (e.g. https://www.economist.com/asia/2021/05/08/why-is-mumbai-handling-its-second-wave-better-than-delhi). But pretty tough restrictions seemed to be having little effect in e.g. Israel before the vaccination campaign took off and that was before super-infectious Delta.
Incidentally, the current wave in Israel seems to show a similarly trajectory, despite the vaccination situation being a mirror-image of India's. Cases have increased from about a dozen a day a month ago to over 1,100 now. But the curious thing is that if you look at the local level, the spread has been "horizontal", not vertical-- the virus takes off in a certain city, infects a lot of people very quickly, then numbers plummet a few days later and numbers go to about zero. The town that started the current wave has been essentially infection-free for a couple weeks now. The increase in infections is because new cities are experiencing outbreaks while the previous ones recover; only one town in Israel is classified "red" per the "traffic-light" system here, and it entered that category yesterday. A good couple dozen cities and towns have gotten that classification and then had it removed when cases dropped.
This might suggest support for the "it just infects everyone and then moves on" approach, but numbers are still very small in absolute terms-- I reckon about 10,000 total people in this wave, while a little under 10% of Israel's population got Covid at any point in the pandemic. One of the areas that have already seen a decline is Tel-Aviv, Israel's second largest city, and it seems unlikely that all of Tel-Aviv already got sick. One disturbing thought: it hasn't yet "hit" Ultra-Orthodox cities, possibly because they are relatively secluded from the rest of the population (especially schools, which is where the current outbreaks tended to start), and vaccination rates in these areas are very low, so very bad news may be forthcoming.
So what stopped the wave in India-- or is it just a statistical mirage? Very curious to hear your thoughts, I have no good answers.
>before you hit “post”, see if you’ve accidentally proven the stock market can’t exist
I don't know which positions this reffers to, but I don't think this is totally fair... 'the results o na survey given to women at this one college' is a lot easier to manipulate than 'the global cost of oil', or whatever.
More generally, the stock market is supposed to represent the value of real things that have their own intrinsic value, and therefore the stock market is resistant to manipulation because you have to either manipulate or misrepresent the value of the real thing, which the market will resist. But a prediction market just cares about an arbitrary outcome, which may have no value in and of itself to people, and therefore may be much easier to manipulate.
Furthermore, stocks may rise and fall freely forever, and tend to homostatistically return to their 'proper' valuation; prediction markets have a set cut-off time, and if you manipulate it really well 5 minutes before that cut-off, you just win. It's not uncommon for a stock price to be manipulated 5 minutes, but that's usually ok because it returns to normal fast; not so in prediction markets.
There are just lots of ways a prediction market is pragmatically different from a stock market, even if they're metaphorically the same kind of thing in principle. It may be that these differences don't actually cause problems or are easily mitigated somehow, but I haven't seen that argument made yet, and it's hard for me to get too excited about prediction markets in practice without seeing it.
So it looks like the Navy is pulling the plug on its decadelong, $500 million railgun experiment. Despite a lot of testing, apparently they couldn't exceed a 110 mile range with a railgun, which is really not that far and definitely less than a missile. Even worse, it doesn't seem like the gun can hold itself together for more than a few shots, which makes intuitive sense. For every force there's an equal & opposite etc. etc. (Perhaps awaiting a leap forward in material sciences technology). It seems that hypersonic missiles are being pursued instead.
(Unless this is a deliberate hoax to head fake Russia & China, and the military is secretly still going ahead with secret railgun deployment!)
I don't quite understand why the range is relatively limited for a railgun though, anyone have any good insights? Could railguns perhaps be more practical on a smaller scale, as artillery, a truck-mounted gun, or even an infantry weapon? In other Space Age Weapons News, it seems that laser technology is coming along pretty well, though I'm unclear how actually destructive the beams are
Just recently discovered the rabbithole of Predictive Processing and Free Energy (primarily via SSC), and while I have not yet even tried to understand the math, the overall framework itself feels
coherent, flexible and powerful.
'Toward a Predictive Theory of Depression' leverages the speculative status to build layers of theory, but I somehow felt it didn't reach "enough".. For example, the stumbling block that Scott lets it rest at
"how do we explain depressed people’s frequent certainty that they’ll fail? A proper Bayesian, barred from having confident beliefs about anything, will be maximally uncertain about whether she’ll fail or succeed – but some depressed people have really strong opinions on this issue."
This doesn't seem like a significant hurdle to me... If we assume that in addition to having low confidence in their predictions, they are also PREDICTING a low confidence in their own predictions, then it makes sense that Depression behaves like an Attractor state. (their brains are "correctly" predicting that they have low confidence in their predictions, which ends up being true) This could also explain how Talk Therapy/CBT works (by changing 'priors' to get the brain to start predicting higher confidence in its predictions)
Am particularly interested in any similarly-technical attempts to apply Predictive Processing for theories to model ADD/ADHD? Also, any relatively accessible forays into the math itself would be very welcome!
I might have missed where this was discussed, but how will survey data (particularly #0, I guess, since that's Scott's) be handled with regard to privacy? Will it be similar to the measures from 2020 (https://slatestarcodex.com/2020/01/20/ssc-survey-results-2020/)?
What is the most efficient way to get the most letters after your name? By 'letters' I mean post-nominal qualification like 'PhD', 'JD', 'MA' which denote a specific level of learning or achievement, and by 'most' I am literally referring to the number of characters, so that in this particular instance MA < MSc < MPhil even though they denote the same level of achievement.
After giving the matter some thought, I believe the most efficient way to achieve the outcome is to take a one-year distance masters in the department of philosophy at as many universities as possible, such that you can put 'MPhil ([Abbreviated name for your University])' each time you qualify, balanced such that you are taking as many courses in one year as you think you have a reasonable chance of qualifying from (chemistry and pharmacy would also work to give you MChem and MPharm, but they seem harder to get than an MPhil to me!). However although this 'officially' counts it would be highly unusual for people to actually put the name of their university in their post-nominals with the exception of Oxford / Cambridge where the reason for doing so is to identify that the Masters is not 'real' and instead just what Oxford / Cambridge call a Bachelors.
The most successful 'rules as intended' method I have found is to get a medical qualification in a specialty field, then join the associated Royal College. For example, if you did this in surgery you would get MChir and MRCS after your name for a total of nine letters. I think this would take about ten years to achieve (although the first time you did it you would also get MB ChB as your undergrad degree and I suspect converting from one medical specialty to another takes less time than going from no qualifications in a field to consultant level). There are also a couple of high-value professional qualifications you could try and 'snipe' along the way such as 'RAnimTechnol' for Registered Animal Technologist - I don't know how long that takes, but it would have to be longer than a decade for it to crowd out more medical degrees. Overall, I think getting more than one letter a year is quite challenging except through a couple of these high-value 'snipes'.
Has anyone ever studied this problem? Has anyone ever achieved a particularly impressive list of n length in real life?
I think the fears in subreddit comment are a bit overblown (based on my experience in the UK). We aren't at herd immunity levels of vaccination and the fact that our vaccination programme is heavily skewed by age complicates a simple herd immunity level calculation anyway. The UK is about 55% double jabbed, another 15% single jabbed, and then ~10% with some level of immunity from a past infection, which adds to about 80% but given that all three won't be perfect, that takes you comfortably below 80% as a herd immunity threshold (and delta probably needs 85-90%). But we haven't vaccinated anyone under 18 (officially), and although about 85% of the over 50s have received two doses, that's true for only about 35-40% of the under 40s. One dose provides significantly less protection against the delta variant (~30% effective vs ~90%) and that means about 2/3 of the under 40s are partially or unvaccinated (it splits roughly 1/3 double, 1/3 single, 1/3 unvaccinated for the 18-40 age group) - more than enough people to sustain widespread transmission.
(scroll down to "The percentage of people testing positive increased in all age groups in the week ending 10 July 2021 in England").
Cases in the over 40s are up but way below the rates for the under 40s. And even here, many of the people catching it will be the remaining 10% or so who remain unvaccinated.
The main risk of delta is probably that a much bigger percentage of unvaccinated people will now get infected, and the risk of vaccine escape variants seems higher because the virus will repeatedly be given opportunities to infect vaccinated people (although I don't know how big a risk this is because I'm not an expert and I haven't read up on it).
As an aside, a big problem here is that the government is insisting on maintaining an 8 week gap between doses, with the reasoning that it provides more durable immunity - which makes little sense when you've already committed to autumn boosters, and the enormous present risk of having one dose (30% protection) vs two (90%) when cases are so high right now. Many of those single dosed under 40s are not allowed to get a second shot for a few more weeks, even when supply allows.
Engineering consciousness... It is actually being attempted. Some (very few?) may have read Dr Mark Solms' book on consciousness, published earlier this year called The Hidden Spring. He (and Karl Friston) suggest consciousness arises *entirely* from within the brain stem, citing evidence from congenitally decorticate (hydranencephalic) children. He develops this is idea by reference to Friston's free energy principle. Because Solms believes he now has a workable theory of consciousness, he is openly trying to engineer it electronically with a group of volunteers connected (I think) to Friston at University College London and a group in South Africa. Anil Seth has reviews Hidden Spring here (quite negatively): https://neurobanter.com/2021/02/18/mixed-feelings-about-a-hard-problem-review-of-the-hidden-spring/
Ad alleged unstopability of Delta variant: wasn´t it, you know, stopped in India without much help from vaccines and with much, much less than 100 % of Indians getting infected?
I want to thank everybody who participated in the optical illusion study! We are aware that it`s long and exhausting and already tried our best at reducing the amount of trials to the minimum that we can work with. But since there are different conditions within the experiment and variability in reaction times is super high between individuals and different stimuli, this amount is still quite high. However, not compared to most perceptual experiments that you typically run in the lab, which often take 300 trials and more o.O
Anyways, more than 300 people completed the experiment, which is just amazing and we are really thankful for that! If something interesting shows up in the data, I`ll definitely post the results here and try to publish a paper about it (where all of you will be mentioned in the acknowledgments of course).
I think I've had COVID, but the symptoms are weird.
Sometime in March (late February?) my sense of taste became weird. My sense of smell was unaffected, or possibly became a little more sensitive.
A lot of food became less pleasant, and notably, mediocre pork had a really nasty taste. Sweets were unaffected.
My tolerance for hot peppers and for sour flavors went way down.
Some food could taste bad, and then good twelve hours later.
I got vaccinated in late March (second dose early April), and that might have caused an improvement.
My sense of taste has been gradually steadying itself, and I haven't tasted that nasty pork flavor for a while. I should probably get some mediocre pork to make sure.
I've had no other symptoms-- no breathing problems, no exhaustion.
This isn't in the range of standard COVID symptoms. On the other hand, it doesn't resemble anything else I know of and COVID famously affects the sense of taste.
Anyone else have something like this? Or heard of this sort of thing?
Am I the only one who is excited about using the ACX survey data for a community matching algorithm? It could be useful for everything from dating to friend finding to locating compatible families for home schooling. Any chance of such a thing being possible based on the 2021 or future surveys (with opt-in option)?
Is "zero covid" actually possible? I thought the point of the lockdowns/restrictions was to prevent hospitals from being overrun, not to try to eliminate covid. That the point of the vaccine wasn't to eliminate covid, but to -- again -- prevent severe infections and hospitalization.
Zero covid seems like it would take multiple years of living in a lockdown state to achieve. Is that what we're headed for?
I cast zero votes in the Book Review contest, despite reading all the main entries and some of the runners up. The cause was simple: for the early entries, I had no basis for comparison with later ones (which obviously I hadn't read yet). But the series was slow enough that by the time I got to the later ones, I no longer remembered my impressions of the earlier ones. Rating on a fixed scale inevitably has this problem; maintaining calibration for what the top end (or bottom end) of the scale is, is difficult.
I have a proposed mechanism for getting around this, if another contest is held. Each rater anchors their first rating as 100 (or, equally good, 500, 1000, or any other large round number). They give every successive rating a cardinal number, relating it to the most recent few ratings (which they probably remember well). At the end, each rater's series of numbers is normalized as a 0.0 to 5.0 scale, and that's used as input. Linear normalization seems best, but logarithmic might also work.
This has the advantage of being fair with respect to publication order, since it's not going to get early reviews anchored too low or too high and fail to leave room for later entries. It has the disadvantage that it is stateful; a user must maintain a record of all ratings for it to produce meaningful results. Improvements are presumably possible as I have spent only an hour or so thinking it through. However, I think it's a good framework for future contests.
I'll be spending the fall in Washington DC, any chance there's an SSC/rat-adjacent group house looking for a September through December subletter? lateish 20s, male, science PhD student. If so, email me at cubecumbered@gmail.com or DM me on twitter (same handle as here) and I'll share more details about myself.
I still can't figure out whether we (late 30's - early 40's couple) should be flying in early Aug timeframe to see family (3 hours flight-time) between TX and WA. Everybody in our cohort is going to be vaccinated, but I can't figure out how to price-in the risk of long-COVID effectively into my decision process.
My indecision gets worse when faced with the prospect of flying 5-6 hours to the east coast later in the year (around US Thanksgiving in Nov) to see family for the first time in nearly 2 years, but it's offset by the hope that there will be more information available to us all by then (even if it turns out to contraindicate travel, at least I'm hoping that drawing such conclusions would be less ambiguous by then).
Free startup idea: block-level drilling for ground source aka geothermal heat pumps in low-rise urban and suburban areas.
It seems clear that if you are in an area with both intense summer heat and intense winter cold, the most efficient and environmentally friendly HVAC solution is a ground source heat pump. But the up front cost of drilling the hole in the ground is high, even after government incentives, and a lot of smaller lots in low-rise urban and suburban areas may just not have the space.
So, what if you could get a bunch of adjacent homeowners to agree to have a common hole drilled that could cross their property lines and split the cost? Seems like a potentially solvable coordination problem in affluent areas where the neighbors are all likely to care about climate change and energy efficiency. The idea was sparked by conversations with my friends in the Twin Cities in Minnesota, and there are a bunch of cities with similar climate, economic status, and density where you could try and do this.
I do not have the HVAC systems knowledge to know whether this would work technically, nor the urban development knowledge to know if there are regulatory/legal obstacles, but it seemed plausible enough that someone should at least look into these.
I'm recommending _We Sold Our Souls_ by Grady Hendrix because it has a vivid depiction of Moloch, which Elua rather surprisingly showing up as heavy metal.
Fair warning, this is a horror novel. While it doesn't go for the gross-out as much as some, it has its moments. It evokes a remarkably wide range of things to be afraid of. (For horror fans, this is a plus.)
Kris Pulaski, the main character, used to be the lead guitarist in a heavy metal band, but everything fell apart after a night that she can't entirely remember. The band was offered contracts, the lead singer became a megastar, and everyone else in the band got out of music.
Okay, a question for international readers: “Why does Oliver Daemen, the newly minted 18 year old Dutch astronaut speak English with the same generic US accent of a Des Moines, Iowa or San Francisco California television weatherman?
I have seen the claim that 99% of current Covid deaths, presumably in the U.S., are unvaccinated, and I don't believe it. The vaccines are estimate to be about 95% effective, so if that were the only difference it should be 95% of deaths. But a majority of adults in the U.S. are vaccinated, and the vaccinated are on average much more likely to be elderly, and hence at risk of dying from Covid, than the unvaccinated, so the percentage of deaths that are unvaccinated should be substantially under 95%.
Two questions:
1. Is there a mistake in my argument? Is a 99% figure plausible?
2. What is the source of the 99% claim? What data is it based on?
Finally! An immunologist addresses the herd immunity elephant in the room. I always thought herd immunity was a modeling mirage — simply because we've never built up any herd immunity against other CoVs. Here, Dr. John Yewdell from the NIH confirms my bias. ;-) ...
Bottom line implications: The COVID-19 pandemic will continue among unvaccinated populations, and it will very likely infect the majority of humans worldwide. If you're one one of the dreamers who thinks the COVID-19 IFR is only 0.15 percent, well that will lead to modest world wide death toll approaching 11,550,000 people. If you use the official CDC IFR number, we will approach 53,900,000 deaths. But the CDC number seems to assume all variants will have the same IFR. I think that's counterintuitive, because the new variants have a higher viral load that hits earlier after initial infection. To my mind, a higher viral load will lead to higher deaths. But what do I know? Anyway, my back-of-the-cocktail-napkin estimates make it look like B.1.1.7, P1, and B.1.315 all have IFRs of around 1 percent (looking at death tolls and the sketchy data about seropositivity). And it looks like B.1.617.2 may be even worse. But assuming a 1 percent IFR, the death toll will be closer to 77 million deaths from COVID-19 in the coming years. Unless we can roll out the vaccines faster...
(made a numerical typo in my original post. Deleting and correcting.)
From the article (linked below)...
"However, seasonal CoVs, which cause approximately 20% of common colds, remain endemic, even though demonstrating only limited antigenic evolution in epitopes targeted by neutralizing antibodies. Although there are only 4 circulating seasonal CoVs, infections frequently recur, even yearly, likely related to waning antibody levels. Human challenge studies established that seasonal CoV reinfection with the identical strain can occur within a year after initial exposure, though typically with reduced shedding and milder symptoms. Reinfection also appears to occur following mild COVID-19 cases, where the serum neutralizing antibody half-life is only approximately 5 weeks."
For Scott, this paper by a team working with Gelman at Columbia is super-interesting and explains something you have talked about a bit in terms of multiple causes and limited possible room for explanation, but it explains this far more clearly than anything I have seen before.
Abstract: In some scientific fields, it is common to have certain variables of interest that are of particular importance and for which there are many studies indicating a relationship with a different explanatory variable. In such cases, particularly those where no relationships are known among explanatory variables, it is worth asking under what conditions it is possible for all such claimed effects to exist simultaneously. This paper addresses this question by reviewing some theorems from multivariate analysis that show, unless the explanatory variables also have sizable effects on each other, it is impossible to have many such large effects. We also discuss implications for the replication crisis in social science.
Mega thread on Kierkegaard and the origins of modern culture https://mobile.twitter.com/ZoharAtkins/status/1415725919327633417
Does anyone have statistics on how many people are unable to get (COVID) vaccinated due to medical reasons? I’ve seen it brought up in arguments for continuing COVID restrictions, but I have yet to see anyone present hard numbers on what percent of the population actually can’t (rather than choose not to).
Regarding Delta, I posted this on my private social media:
"Most people I know have tuned out COVID, understandably. Most of you reading this are fully vaccinated and are living in regions that are reopening. If you are fully vaccinated and young, your odds of dying of COVID are about the same as dying from the flu.
All fully vaccinated people have to worry about today is governments reimposing restrictions due to the rise of the delta variant.
Prediction: If you live in a place where less than 65% of the population is vaccinated, you will face some level of restrictions in the upcoming months. Unless your local / federal government goes full French and decides to make life inconvenient for those who choose to not get the vaccine."
I still have to worry about COVID because I live in a place where less than 30% of pop is not vaxxed, including me (I shall get my shot this upcoming week), I have been fascinated by COVID since March 2020, and I believe some COVID prediction markets are or have been mispriced.
Going back to the comment above, why should most people worry about Delta? Other than long covid and future restrictions?
I wrote the review of Addiction by Design (casino gambling book) in the recent book review contest. Thank you to all the people who read and commented, and to all others who wrote reviews.
If you liked that review, you might also like another book review I wrote of Jeremy Bentham's Deontology here: https://forum.effectivealtruism.org/posts/zpxvCBwgkm9fSGGp5/book-review-deontology-by-jeremy-bentham
Re: delta
The big open question is what are the symptomatic rates (i.e. any symptoms, hospitalization, long covid, and death) for people that have been vaccinated and contract it.
Can anyone recommend resources for self-study of getting into meditative/spiritual yoga for someone who already practices Western-style yoga-as-physical-improvement? if anyone knows of something vaguely MTCB-like that would be idea.
Question about testing for Down syndrome, as done approx. 20 years ago. I'm sure that somebody here who can answer it.
Heard an account about a pregnant woman getting fetal genetic testing for Down syndrome approximately 20 years ago. She was told that test indicated there was a 60% chance the baby would have Down's, and that even if the baby did not he would likely have intellectual and psychiatric difficulties -- I believe ADD and learning disabilities were mentioned as possibilities. She chose to keep the baby, and he was born with no sign whatever of Down's, and never developed any of the usual signs. However, he has had substantial psychiatric difficulties (although not of the kind the doctors warned about: he is quite smart, and did well in school until various kinds of life chaos interfered with his schooling).
So here are my questions: (1) What kind of test result would lead the docs to say there was a 60% chance that the fetus had Down's? I know that Down's is trisomy 21, so I'm guessing that examination of the fetal cells showed some normal cells and some with 3 copies of #21. So how did the doc arrive at the odds at 60%? Would that be in a situation where 60% of the cells showed trisomy 21, 40% did not? (2) Does the second part of what the doc reportedly said make sense -- the prediction that even if the baby was not born with Down's, he was likely to have intellectual and psychiatric impairments? What possible result of the Down's genetic test would lead to such a prediction? I know about "mosaic" Down's, but it sounds as though people with that version of Down's would still be considered to have Down's, just a somewhat milder form, so I don't think that could have been what the doc was talking about. He apparently was talking about a whole separate problem the child was likely to have if he did not have Downs.
Do you like audiobooks? Do you prefer listening to audiobooks over reading a physical book (or Kindle)?
There's certainly a convenience factor; you can listen to an audiobook in a car or with your eyes closed. Beyond that, I find that if I'm trying to learn something audiobooks are slow and the inability to jump back and forth is inconvenient; while if I'm simply reading/listening for pleasure (often to a book I have already read) the slow pace can almost transform the story.
On the Covid March 2020 moment point, one thing I don't really understand is why the virus (appears to have) peaked and then plummeted in India.
Infections skyrocketed from about 11,000 cases per day in February to an eye-watering 391,000 cases per day in May (all stats here from ourworldindata.org). At the same time, the share of positive tests went from 1% to 22%, suggesting the true increase was much steeper than even that suggests. But days after it hit that peak, numbers fell right back down to around 40,000/day and 1% positivity today. The fact that positivity rate and deaths followed the same pattern suggests to me that this isn't an artifact of testing or bad data (though open to hearing otherwise).
Yet with a cumulative 31 million confirmed cases, even with massive corrections for incomplete data it seems a stretch to suggest that basically everyone who could get it got it. And while I'm far from an expert, the shape of the progression also puzzles me-- cases were going steeply up about a week before they started going steeply down, rather than the more gradual leveling off I'd expect if the susceptible population was just gradually petering out (glad to be corrected on this point if mistaken).
It definitely wasn't vaccination-- proportion of fully vaccinated individuals is around 5% even now. The main narrative I read is that while federal restrictions were haphazard, many state-level responses such as Maharashtra's/Mumbai's were quite effective (e.g. https://www.economist.com/asia/2021/05/08/why-is-mumbai-handling-its-second-wave-better-than-delhi). But pretty tough restrictions seemed to be having little effect in e.g. Israel before the vaccination campaign took off and that was before super-infectious Delta.
Incidentally, the current wave in Israel seems to show a similarly trajectory, despite the vaccination situation being a mirror-image of India's. Cases have increased from about a dozen a day a month ago to over 1,100 now. But the curious thing is that if you look at the local level, the spread has been "horizontal", not vertical-- the virus takes off in a certain city, infects a lot of people very quickly, then numbers plummet a few days later and numbers go to about zero. The town that started the current wave has been essentially infection-free for a couple weeks now. The increase in infections is because new cities are experiencing outbreaks while the previous ones recover; only one town in Israel is classified "red" per the "traffic-light" system here, and it entered that category yesterday. A good couple dozen cities and towns have gotten that classification and then had it removed when cases dropped.
This might suggest support for the "it just infects everyone and then moves on" approach, but numbers are still very small in absolute terms-- I reckon about 10,000 total people in this wave, while a little under 10% of Israel's population got Covid at any point in the pandemic. One of the areas that have already seen a decline is Tel-Aviv, Israel's second largest city, and it seems unlikely that all of Tel-Aviv already got sick. One disturbing thought: it hasn't yet "hit" Ultra-Orthodox cities, possibly because they are relatively secluded from the rest of the population (especially schools, which is where the current outbreaks tended to start), and vaccination rates in these areas are very low, so very bad news may be forthcoming.
So what stopped the wave in India-- or is it just a statistical mirage? Very curious to hear your thoughts, I have no good answers.
>before you hit “post”, see if you’ve accidentally proven the stock market can’t exist
I don't know which positions this reffers to, but I don't think this is totally fair... 'the results o na survey given to women at this one college' is a lot easier to manipulate than 'the global cost of oil', or whatever.
More generally, the stock market is supposed to represent the value of real things that have their own intrinsic value, and therefore the stock market is resistant to manipulation because you have to either manipulate or misrepresent the value of the real thing, which the market will resist. But a prediction market just cares about an arbitrary outcome, which may have no value in and of itself to people, and therefore may be much easier to manipulate.
Furthermore, stocks may rise and fall freely forever, and tend to homostatistically return to their 'proper' valuation; prediction markets have a set cut-off time, and if you manipulate it really well 5 minutes before that cut-off, you just win. It's not uncommon for a stock price to be manipulated 5 minutes, but that's usually ok because it returns to normal fast; not so in prediction markets.
There are just lots of ways a prediction market is pragmatically different from a stock market, even if they're metaphorically the same kind of thing in principle. It may be that these differences don't actually cause problems or are easily mitigated somehow, but I haven't seen that argument made yet, and it's hard for me to get too excited about prediction markets in practice without seeing it.
If you could eliminate one bias entirely from the world, which would you remove?
(Or, feel free to interpret this question as, “which bias do you think is the strongest, if you had to pick one?”)
What we talk about when we talk about religion: https://whatiscalledthinking.substack.com/p/what-is-religion
So it looks like the Navy is pulling the plug on its decadelong, $500 million railgun experiment. Despite a lot of testing, apparently they couldn't exceed a 110 mile range with a railgun, which is really not that far and definitely less than a missile. Even worse, it doesn't seem like the gun can hold itself together for more than a few shots, which makes intuitive sense. For every force there's an equal & opposite etc. etc. (Perhaps awaiting a leap forward in material sciences technology). It seems that hypersonic missiles are being pursued instead.
(Unless this is a deliberate hoax to head fake Russia & China, and the military is secretly still going ahead with secret railgun deployment!)
I don't quite understand why the range is relatively limited for a railgun though, anyone have any good insights? Could railguns perhaps be more practical on a smaller scale, as artillery, a truck-mounted gun, or even an infantry weapon? In other Space Age Weapons News, it seems that laser technology is coming along pretty well, though I'm unclear how actually destructive the beams are
https://thehill.com/policy/defense/561231-navy-pulls-plug-on-500-million-railgun-effort
Just recently discovered the rabbithole of Predictive Processing and Free Energy (primarily via SSC), and while I have not yet even tried to understand the math, the overall framework itself feels
coherent, flexible and powerful.
'Toward a Predictive Theory of Depression' leverages the speculative status to build layers of theory, but I somehow felt it didn't reach "enough".. For example, the stumbling block that Scott lets it rest at
"how do we explain depressed people’s frequent certainty that they’ll fail? A proper Bayesian, barred from having confident beliefs about anything, will be maximally uncertain about whether she’ll fail or succeed – but some depressed people have really strong opinions on this issue."
This doesn't seem like a significant hurdle to me... If we assume that in addition to having low confidence in their predictions, they are also PREDICTING a low confidence in their own predictions, then it makes sense that Depression behaves like an Attractor state. (their brains are "correctly" predicting that they have low confidence in their predictions, which ends up being true) This could also explain how Talk Therapy/CBT works (by changing 'priors' to get the brain to start predicting higher confidence in its predictions)
Am particularly interested in any similarly-technical attempts to apply Predictive Processing for theories to model ADD/ADHD? Also, any relatively accessible forays into the math itself would be very welcome!
I might have missed where this was discussed, but how will survey data (particularly #0, I guess, since that's Scott's) be handled with regard to privacy? Will it be similar to the measures from 2020 (https://slatestarcodex.com/2020/01/20/ssc-survey-results-2020/)?
What is the most efficient way to get the most letters after your name? By 'letters' I mean post-nominal qualification like 'PhD', 'JD', 'MA' which denote a specific level of learning or achievement, and by 'most' I am literally referring to the number of characters, so that in this particular instance MA < MSc < MPhil even though they denote the same level of achievement.
After giving the matter some thought, I believe the most efficient way to achieve the outcome is to take a one-year distance masters in the department of philosophy at as many universities as possible, such that you can put 'MPhil ([Abbreviated name for your University])' each time you qualify, balanced such that you are taking as many courses in one year as you think you have a reasonable chance of qualifying from (chemistry and pharmacy would also work to give you MChem and MPharm, but they seem harder to get than an MPhil to me!). However although this 'officially' counts it would be highly unusual for people to actually put the name of their university in their post-nominals with the exception of Oxford / Cambridge where the reason for doing so is to identify that the Masters is not 'real' and instead just what Oxford / Cambridge call a Bachelors.
The most successful 'rules as intended' method I have found is to get a medical qualification in a specialty field, then join the associated Royal College. For example, if you did this in surgery you would get MChir and MRCS after your name for a total of nine letters. I think this would take about ten years to achieve (although the first time you did it you would also get MB ChB as your undergrad degree and I suspect converting from one medical specialty to another takes less time than going from no qualifications in a field to consultant level). There are also a couple of high-value professional qualifications you could try and 'snipe' along the way such as 'RAnimTechnol' for Registered Animal Technologist - I don't know how long that takes, but it would have to be longer than a decade for it to crowd out more medical degrees. Overall, I think getting more than one letter a year is quite challenging except through a couple of these high-value 'snipes'.
Has anyone ever studied this problem? Has anyone ever achieved a particularly impressive list of n length in real life?
I'm in the UK and would prefer to use UK honourifics if possible - https://en.wikipedia.org/wiki/List_of_post-nominal_letters_(United_Kingdom)
I think the fears in subreddit comment are a bit overblown (based on my experience in the UK). We aren't at herd immunity levels of vaccination and the fact that our vaccination programme is heavily skewed by age complicates a simple herd immunity level calculation anyway. The UK is about 55% double jabbed, another 15% single jabbed, and then ~10% with some level of immunity from a past infection, which adds to about 80% but given that all three won't be perfect, that takes you comfortably below 80% as a herd immunity threshold (and delta probably needs 85-90%). But we haven't vaccinated anyone under 18 (officially), and although about 85% of the over 50s have received two doses, that's true for only about 35-40% of the under 40s. One dose provides significantly less protection against the delta variant (~30% effective vs ~90%) and that means about 2/3 of the under 40s are partially or unvaccinated (it splits roughly 1/3 double, 1/3 single, 1/3 unvaccinated for the 18-40 age group) - more than enough people to sustain widespread transmission.
The case numbers broken down by age reflect this:
https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/articles/coronaviruscovid19/latestinsights
(scroll down to "The percentage of people testing positive increased in all age groups in the week ending 10 July 2021 in England").
Cases in the over 40s are up but way below the rates for the under 40s. And even here, many of the people catching it will be the remaining 10% or so who remain unvaccinated.
The main risk of delta is probably that a much bigger percentage of unvaccinated people will now get infected, and the risk of vaccine escape variants seems higher because the virus will repeatedly be given opportunities to infect vaccinated people (although I don't know how big a risk this is because I'm not an expert and I haven't read up on it).
As an aside, a big problem here is that the government is insisting on maintaining an 8 week gap between doses, with the reasoning that it provides more durable immunity - which makes little sense when you've already committed to autumn boosters, and the enormous present risk of having one dose (30% protection) vs two (90%) when cases are so high right now. Many of those single dosed under 40s are not allowed to get a second shot for a few more weeks, even when supply allows.
Engineering consciousness... It is actually being attempted. Some (very few?) may have read Dr Mark Solms' book on consciousness, published earlier this year called The Hidden Spring. He (and Karl Friston) suggest consciousness arises *entirely* from within the brain stem, citing evidence from congenitally decorticate (hydranencephalic) children. He develops this is idea by reference to Friston's free energy principle. Because Solms believes he now has a workable theory of consciousness, he is openly trying to engineer it electronically with a group of volunteers connected (I think) to Friston at University College London and a group in South Africa. Anil Seth has reviews Hidden Spring here (quite negatively): https://neurobanter.com/2021/02/18/mixed-feelings-about-a-hard-problem-review-of-the-hidden-spring/
Any thoughts?
Last week WHO published guidelines for human genetic modification.
https://www.who.int/publications/i/item/9789240030381
Was there any discussion of it in the media/blogosphere?
Ad alleged unstopability of Delta variant: wasn´t it, you know, stopped in India without much help from vaccines and with much, much less than 100 % of Indians getting infected?
I want to thank everybody who participated in the optical illusion study! We are aware that it`s long and exhausting and already tried our best at reducing the amount of trials to the minimum that we can work with. But since there are different conditions within the experiment and variability in reaction times is super high between individuals and different stimuli, this amount is still quite high. However, not compared to most perceptual experiments that you typically run in the lab, which often take 300 trials and more o.O
Anyways, more than 300 people completed the experiment, which is just amazing and we are really thankful for that! If something interesting shows up in the data, I`ll definitely post the results here and try to publish a paper about it (where all of you will be mentioned in the acknowledgments of course).
I think I've had COVID, but the symptoms are weird.
Sometime in March (late February?) my sense of taste became weird. My sense of smell was unaffected, or possibly became a little more sensitive.
A lot of food became less pleasant, and notably, mediocre pork had a really nasty taste. Sweets were unaffected.
My tolerance for hot peppers and for sour flavors went way down.
Some food could taste bad, and then good twelve hours later.
I got vaccinated in late March (second dose early April), and that might have caused an improvement.
My sense of taste has been gradually steadying itself, and I haven't tasted that nasty pork flavor for a while. I should probably get some mediocre pork to make sure.
I've had no other symptoms-- no breathing problems, no exhaustion.
This isn't in the range of standard COVID symptoms. On the other hand, it doesn't resemble anything else I know of and COVID famously affects the sense of taste.
Anyone else have something like this? Or heard of this sort of thing?
Am I the only one who is excited about using the ACX survey data for a community matching algorithm? It could be useful for everything from dating to friend finding to locating compatible families for home schooling. Any chance of such a thing being possible based on the 2021 or future surveys (with opt-in option)?
Is "zero covid" actually possible? I thought the point of the lockdowns/restrictions was to prevent hospitals from being overrun, not to try to eliminate covid. That the point of the vaccine wasn't to eliminate covid, but to -- again -- prevent severe infections and hospitalization.
Zero covid seems like it would take multiple years of living in a lockdown state to achieve. Is that what we're headed for?
Rating a long intermittent series
I cast zero votes in the Book Review contest, despite reading all the main entries and some of the runners up. The cause was simple: for the early entries, I had no basis for comparison with later ones (which obviously I hadn't read yet). But the series was slow enough that by the time I got to the later ones, I no longer remembered my impressions of the earlier ones. Rating on a fixed scale inevitably has this problem; maintaining calibration for what the top end (or bottom end) of the scale is, is difficult.
I have a proposed mechanism for getting around this, if another contest is held. Each rater anchors their first rating as 100 (or, equally good, 500, 1000, or any other large round number). They give every successive rating a cardinal number, relating it to the most recent few ratings (which they probably remember well). At the end, each rater's series of numbers is normalized as a 0.0 to 5.0 scale, and that's used as input. Linear normalization seems best, but logarithmic might also work.
This has the advantage of being fair with respect to publication order, since it's not going to get early reviews anchored too low or too high and fail to leave room for later entries. It has the disadvantage that it is stateful; a user must maintain a record of all ratings for it to produce meaningful results. Improvements are presumably possible as I have spent only an hour or so thinking it through. However, I think it's a good framework for future contests.
US stock markets may be having a March 2020 moment. Someone is taking the Delta variant seriously.
https://www.google.com/amp/s/amp.cnn.com/cnn/2021/07/19/investing/dow-stock-market-today/index.html
I'll be spending the fall in Washington DC, any chance there's an SSC/rat-adjacent group house looking for a September through December subletter? lateish 20s, male, science PhD student. If so, email me at cubecumbered@gmail.com or DM me on twitter (same handle as here) and I'll share more details about myself.
There is a good discussion of the Delta virus from another blog I like reading at https://yourlocalepidemiologist.substack.com/p/current-state-of-affairs-july-19.
I still can't figure out whether we (late 30's - early 40's couple) should be flying in early Aug timeframe to see family (3 hours flight-time) between TX and WA. Everybody in our cohort is going to be vaccinated, but I can't figure out how to price-in the risk of long-COVID effectively into my decision process.
My indecision gets worse when faced with the prospect of flying 5-6 hours to the east coast later in the year (around US Thanksgiving in Nov) to see family for the first time in nearly 2 years, but it's offset by the hope that there will be more information available to us all by then (even if it turns out to contraindicate travel, at least I'm hoping that drawing such conclusions would be less ambiguous by then).
https://iz2020.substack.com/p/mandatory-union
Comments and critique appreciated
Just to throw that in:
https://blogs.bmj.com/bmj/2021/07/05/time-to-assume-that-health-research-is-fraudulent-until-proved-otherwise/
When do you predict there will be space tourism flights where you get to orbit the Earth at least once?
Free startup idea: block-level drilling for ground source aka geothermal heat pumps in low-rise urban and suburban areas.
It seems clear that if you are in an area with both intense summer heat and intense winter cold, the most efficient and environmentally friendly HVAC solution is a ground source heat pump. But the up front cost of drilling the hole in the ground is high, even after government incentives, and a lot of smaller lots in low-rise urban and suburban areas may just not have the space.
So, what if you could get a bunch of adjacent homeowners to agree to have a common hole drilled that could cross their property lines and split the cost? Seems like a potentially solvable coordination problem in affluent areas where the neighbors are all likely to care about climate change and energy efficiency. The idea was sparked by conversations with my friends in the Twin Cities in Minnesota, and there are a bunch of cities with similar climate, economic status, and density where you could try and do this.
I do not have the HVAC systems knowledge to know whether this would work technically, nor the urban development knowledge to know if there are regulatory/legal obstacles, but it seemed plausible enough that someone should at least look into these.
I'm recommending _We Sold Our Souls_ by Grady Hendrix because it has a vivid depiction of Moloch, which Elua rather surprisingly showing up as heavy metal.
Fair warning, this is a horror novel. While it doesn't go for the gross-out as much as some, it has its moments. It evokes a remarkably wide range of things to be afraid of. (For horror fans, this is a plus.)
Kris Pulaski, the main character, used to be the lead guitarist in a heavy metal band, but everything fell apart after a night that she can't entirely remember. The band was offered contracts, the lead singer became a megastar, and everyone else in the band got out of music.
https://slatestarcodex.com/2014/07/30/meditations-on-moloch/
This is Scott's essay about Moloch-- Moloch is the power of inhuman optimization.
Elua is the opposite-- enough room so that living things can thrive.
Okay, a question for international readers: “Why does Oliver Daemen, the newly minted 18 year old Dutch astronaut speak English with the same generic US accent of a Des Moines, Iowa or San Francisco California television weatherman?
I have seen the claim that 99% of current Covid deaths, presumably in the U.S., are unvaccinated, and I don't believe it. The vaccines are estimate to be about 95% effective, so if that were the only difference it should be 95% of deaths. But a majority of adults in the U.S. are vaccinated, and the vaccinated are on average much more likely to be elderly, and hence at risk of dying from Covid, than the unvaccinated, so the percentage of deaths that are unvaccinated should be substantially under 95%.
Two questions:
1. Is there a mistake in my argument? Is a 99% figure plausible?
2. What is the source of the 99% claim? What data is it based on?
According to one story I found: https://apnews.com/article/coronavirus-pandemic-health-941fcf43d9731c76c16e7354f5d5e187
the basis is an AP calculation based on CDC data. Beyond that there were quote from various people.
Family is discussing whether it is a good idea to vaccinate my 17 year old sister (for COVID).
Has anyone compiled human-readable info about risks/benefits for such age?
Ross Douthat has some interesting thoughts on overcoming vaccine hesitancy in his column today and makes the good point:
"The idea that every prominent conservative entertaining skeptical arguments must be a knowing liar is an important error in its own right."
https://www.nytimes.com/2021/07/20/opinion/covid-vaccine-hesitancy.html?action=click&module=Opinion&pgtype=Homepage
The articles at the ACX page are listed in reverse chronological order now, which is much more convenient. Thank you very much.
Finally! An immunologist addresses the herd immunity elephant in the room. I always thought herd immunity was a modeling mirage — simply because we've never built up any herd immunity against other CoVs. Here, Dr. John Yewdell from the NIH confirms my bias. ;-) ...
Bottom line implications: The COVID-19 pandemic will continue among unvaccinated populations, and it will very likely infect the majority of humans worldwide. If you're one one of the dreamers who thinks the COVID-19 IFR is only 0.15 percent, well that will lead to modest world wide death toll approaching 11,550,000 people. If you use the official CDC IFR number, we will approach 53,900,000 deaths. But the CDC number seems to assume all variants will have the same IFR. I think that's counterintuitive, because the new variants have a higher viral load that hits earlier after initial infection. To my mind, a higher viral load will lead to higher deaths. But what do I know? Anyway, my back-of-the-cocktail-napkin estimates make it look like B.1.1.7, P1, and B.1.315 all have IFRs of around 1 percent (looking at death tolls and the sketchy data about seropositivity). And it looks like B.1.617.2 may be even worse. But assuming a 1 percent IFR, the death toll will be closer to 77 million deaths from COVID-19 in the coming years. Unless we can roll out the vaccines faster...
(made a numerical typo in my original post. Deleting and correcting.)
From the article (linked below)...
"However, seasonal CoVs, which cause approximately 20% of common colds, remain endemic, even though demonstrating only limited antigenic evolution in epitopes targeted by neutralizing antibodies. Although there are only 4 circulating seasonal CoVs, infections frequently recur, even yearly, likely related to waning antibody levels. Human challenge studies established that seasonal CoV reinfection with the identical strain can occur within a year after initial exposure, though typically with reduced shedding and milder symptoms. Reinfection also appears to occur following mild COVID-19 cases, where the serum neutralizing antibody half-life is only approximately 5 weeks."
https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1009509
For Scott, this paper by a team working with Gelman at Columbia is super-interesting and explains something you have talked about a bit in terms of multiple causes and limited possible room for explanation, but it explains this far more clearly than anything I have seen before.
http://stat.columbia.edu/~gelman/research/unpublished/piranhas.pdf
Abstract: In some scientific fields, it is common to have certain variables of interest that are of particular importance and for which there are many studies indicating a relationship with a different explanatory variable. In such cases, particularly those where no relationships are known among explanatory variables, it is worth asking under what conditions it is possible for all such claimed effects to exist simultaneously. This paper addresses this question by reviewing some theorems from multivariate analysis that show, unless the explanatory variables also have sizable effects on each other, it is impossible to have many such large effects. We also discuss implications for the replication crisis in social science.