(Personal note: My Facebook account was hacked. Attacker has not contacted me, and deleted my name from the profile. Since I am already Against Facebook I do not view this as a great loss unless it leads to further trouble, but readers should be aware that for now I have no Facebook account. If I don’t get it back, I may or may not create a new one.)
We are now a second week into The Great Unmasking, with no sign of trouble in the case numbers. While it ain’t over till it’s over and I’m not quite prepared yet to outright declare victory. On reflection my criteria for V-A day is ‘I notice I am acting the exact same way I would if I was unvaccinated, provided everyone else was OK with that’ and we’re definitely not there yet. Still, it seems likely that in America it’s all over but the shouting and I see a lady preparing to sing.
At that point, we’ll potentially need to worry about future seasonal concerns if not enough people get vaccinated and we return to the full old normal, and we’ll need to keep an eye on variants, but that’s about it.
With numbers declining and life starting to return to normal, talk turned to the question of whether Covid-19 leaked from a lab, and potentially originated from gain of function research. Once considered a vile, racist conspiracy theory to be ruthlessly censored and scorned, it is now being officially investigated and is widely considered highly likely. We’ve seen this kind of Official Facts transformation before, and things are following the standard script. There’s a section on it, but in general I’d still hope to keep this mostly out of the column going forward.
Let’s run the numbers.
The Numbers
Predictions
Prediction (now using Johns Hopkins numbers fully going forward): Positivity rate of 2.7% (down 0.3%) and deaths fall by 8%.
Result: Positivity rate of 2.5% (down 0.5%) and deaths fall by 10%.
Prediction: Positivity rate of 2.2% (down 0.3%) and deaths fall by 9%. Continued steady progress, with a little hedging in case we got out in front of things a bit.
Deaths
I deleted 400 deaths, 300 from Oklahoma and 100 from New Mexico, that were clearly reporting previous deaths rather than new data. Oklahoma reports weekly so there’s no way to know the real number, and I left it on the high end of possible. New Mexico had a day with 114 deaths so taking away most of those seems reasonable. That resulted in this:
| Date | WEST | MIDWEST | SOUTH | NORTHEAST | TOTAL |
| Apr 15-Apr 21 | 883 | 987 | 1747 | 1168 | 4785 |
| Apr 22-Apr 28 | 752 | 1173 | 1609 | 1110 | 4644 |
| Apr 29-May 5 | 943 | 1220 | 1440 | 971 | 4574 |
| May 6-May 12 | 826 | 1069 | 1392 | 855 | 4142 |
| May 13-May 19 | 592 | 1194 | 1277 | 811 | 3874 |
| May 20-May 26 | 615 | 948 | 1279 | 631 | 3473 |
The decline in the Northeast is steep, but at least superficially it appears real. It’s likely last week’s number was higher than the true trend line, slash random variance reduced deaths a lot in the Northeast and Midwest, and hid a true decrease in other regions. The overall rate of decline seems most important here.
Cases
| Date | WEST | MIDWEST | SOUTH | NORTHEAST | |
| Apr 8-Apr 14 | 60,693 | 124,161 | 110,995 | 137,213 | 433,062 |
| Apr 15-Apr 21 | 54,778 | 107,700 | 110,160 | 119,542 | 392,180 |
| Apr 22-Apr 28 | 54,887 | 88,973 | 97,482 | 78,442 | 319,784 |
| Apr 29-May 5 | 52,984 | 78,778 | 85,641 | 68,299 | 285,702 |
| May 6-May 12 | 46,045 | 59,945 | 70,740 | 46,782 | 223,512 |
| May 13-May 19 | 39,601 | 45,030 | 63,529 | 34,309 | 182,469 |
| May 20-May 26 | 33,890 | 34,694 | 48,973 | 24,849 | 142,406 |
Those are huge gains across the board. This is what victory looks like. The control system will doubtless try its best, but it seems we made it.
India
Things are steadily on their way back down, so some combination of changing seasonality, ending the election and various festivals, and the control system combined to get things back under control. Things are still rather bad out there, and it’s going to be a while before we get enough vaccine shots to India, but the worst is likely over.
As with every other strain, there’s paranoia around how effective the vaccines are against the India strain. Also as usual, everyone quotes the infection percentages (e.g. there 70-75% for Pfizer, here 88% for Pfizer and 60% for AZ) rather than the severe illness percentages or death percentages, with a special emphasis on numbers that are on the low end, so numbers that sound scary mostly aren’t scary. This data looks scary at first, but when compared to the baseline numbers for similar vaccinations in the past it mostly seems fine even if the effect here is fully real.
Escape From the Lab
Until now, I’ve entirely dodged the question of the origins of Covid-19. The issue didn’t seem to have practical implications and there were plenty of other things to focus on. I was content to let others handle that question. Even here, I don’t really have anything new to offer, it’s more that it’s so on point that it requires mentioning.
Now that the consensus has shifted, and the lab escape hypothesis is no longer grounds for censorship and cancellation and the tone of coverage has dramatically shifted overnight, it joins a long list of other questions in which the media, elites, authorities and censors treated a question as definitively one way, then abruptly shifted when that position stopped being defensible, and quickly started rewriting history to pretend we had always been at war with Eastasia. As usual, the elites and authority figures are going to pretend what happened didn’t happen, that they never denied anything, that they’re the ones who figured this all out, and that the people who claimed this crazy hypothesis back when it was still crazy should continue being dismissed as cranks because in their parlance crank means ‘went against elite consensus’ rather than ‘made claims without good evidence, especially ones that aren’t true.’
Consider the ‘stealth edits’ that Vox did to its old articles once a month had gone by and the ‘debunking’ had become common knowledge:
It’s totally fine to be wrong sometimes. It’s completely unacceptable to rewrite the public record to pretend retroactively that you covered your ass. Especially when the certainty that comes from lack of ass covering was used as a justification for censorship, castigation and cancellation.
Until this week, Facebook was removing posts that claimed Covid-19 leaked from a lab.
After being caught doing this, Vox admitted they did the edits:
If that note had also been published in April 2020, then fine, but it wasn’t. One could argue it’s fine anyway, it’s better to fix things than not fix things, and that it’s unreasonable and misleading to call Vox out now for edits they made over a year ago. There’s some validity to that. A lot of Paul’s readers presumably assumed that the edits were made in the last month, which would be a much worse look. Then again, when are you going to call someone out on such stealth edits? Presumably at exactly the same time one would call someone out for writing the original.
I’d be curious what Vox thinks it means by ‘scientific thinking’ here. Does it mean ‘the scientists started using CYA-words so we should too’? Or does it mean the actually correct ‘scientific thinking means not taking people’s words for things or reporting claims as facts, so instead we’re going to do proper journalism and tell you exactly what evidence is present’?
I remember when the New York Times did this every single time because that’s how one is actually a paper of record – ‘Man arrested during an armed robbery, police say.’ Police say is doing important work there. Whereas now, NYT and Vox are the two mainstream sources I will not link to and that I avoid reading.
Essentially all claiming-to-be authoritative sources treated the lab hypothesis as pure conspiracy theory and utterly impossible up until the last few weeks, the same way they insisted there concerns about the virus in February were alarmist and racist, masks didn’t work, outdoor events were dangerous, vaccines would take a minimum of eighteen months and so on and so forth. I grow weary of typing the list out.
I don’t think this is quite the same as the others, as it both did not do practical harm to people’s decision making and in hindsight still seems like a highly reasonable position to put most of one’s probability into at the time. That doesn’t make it acceptable to censor and castigate, but it’s a big step from being clearly and knowably wrong at the time.
Here’s a thread on historical lab leaks, which doesn’t prove anything but does make it clear that a lab leak was always plausible.
Nate Silver asks people to go on record with their probabilities of a lab leak. The comments are telling. Most replies Twitter shows us give absurdly low numbers, as close to 0% as they dare, and continue to attempt to police the discourse and accuse anyone saying otherwise of being crazy, while also attacking the very idea of Bayesian evidence. Then there are the replies saying ‘there’s a lab, there’s a virus, obviously it’s from the lab.’ None of it is a good look.
Noticing both that there is huge unnecessary risk of pandemics coming from wet markets or otherwise having a natural origin, and that a pandemic could easily come from a lab accident, is what matters, regardless of which origin led to Covid-19 in particular.
Even more than that, I’m interested in the mechanisms behind the suppression of information and debate. My odds are mostly on the basis of the shift in other people’s odds being against strong headwinds trying to prevent it, and the improbability that things would have gotten this far unless the case for escape was strong.
At this point, I think I am somewhat below Nate Silver’s 60% odds that the virus escaped from the lab, and put myself at about 40%, but I haven’t looked carefully and this probability is weakly held. I’m sharing it because it’s important to share probabilities even when they’re weakly held. The question of whether we’ll ever prove what happened, or the official story will conclude a lab leak, is very different from the question of the actual origin, so there’s no pure way to evaluate such predictions, but it seems important to give a number even with my uncertainty. I still find the natural origin story likely (that’s why there was a lab in the first place), and the evidence that the lab is acting suspicious and everyone is covering things up is still consistent with them doing that automatically without any need for there to be anything to cover up, but a lab leak is also plausible and the investigation and legitimacy of the lab leak claim getting this far under these conditions was surprising.
Despite all the other ways in which we were misled, until last week I still reliably put lower probability on the lab leak theory than I should have on reflection. A lot of that was that, as noted, I intentionally didn’t consider the question and wouldn’t have even if I thought the probability was higher, and we still don’t know what happened, but none of that means I didn’t get it wrong or that such errors don’t need to be admitted and corrected.
Should we update to give more credence to other things that are labeled as ‘conspiracy theory’? That’s tricky. I don’t think this was a ‘grand conspiracy’ or anything, nor do I think those suppressing the theory had any knowledge of whether or not the virus leaked from a lab. My model says this is how the system works by default, with all who form the system instinctively moving to implement the suppression of such speculations, without any need to coordinate.
It’s important to note that this is not a conspiracy theory because if the theory is true there need not be a conspiracy. The lab didn’t intend to leak the virus (or if it did, that would be a very different theory). If the leak happened, the lab almost certainly accidentally leaked the virus, the same as there have been historical other leaks, and of course they didn’t come forward and admit that and instead covered it up, and the system did its thing automatically rather than because there was some cabal or set of secret orders. That’s true even if the virus was also created in the lab.
This is a sharp contrast to actual conspiracy theories that involve conspiracies that explicitly coordinate to achieve objectives, including but not limited to suppressing The Truth That Is Out There. Such claims should continue to be viewed with extreme skepticism. There likely is no conspiracy.
In contrast, there is totally a cover up. I’m at 98%+ that there was a de facto cover-up, even if the virus didn’t leak from the lab. There is often a truth that is effectively suppressed, because the Powers That Be intuitively sense that it would be better not to spread it around, and you can’t take the elite consensus or media’s word for anything. There’s no reason for such powers to check the validity of the claim before suppressing it, since they’d want to suppress it if it was true and also suppress it if it was false, and checking validity risks giving the claim credibility. If you think that this originates in conspiracy, your map won’t be accurate, and you’ll see it in all the wrong places.
Calls to ‘hold people accountable’ for the suppression of such information don’t interest me much, any more than I want to ‘hold people accountable’ for the mask debacle. I’m not opposed to them, but I don’t see much point. What is important is to know what happened and how such things happen, so we can avoid them in the future, and update our view of various sources of information accordingly.
The likely consequence to the origin being considered a lab leak is that people might react to that information by attempting to punish, or learning to fear or hate, that which they see is responsible. That presumably means some combination of China, likely the United States because we offered some funding and it is popular to turn everything on us whenever possible, and biologists and those who study infectious disease or scientists and labs in general. A lot of people would hear ‘lab leak’ and assume it was intentional or a weapon, and nothing we say would convince them differently. There would be a general rise in conspiracy theory and paranoia, as discussed above. Any movie with a scientist is more likely to paint them as evil, or irresponsible, and the cause of potential disaster. Geopolitical tensions would presumably rise all around. Worries about ‘hate’ and ethnic violence are used as cudgels these days, but they’re also real concerns.
None of that seems positive. If it had happened last year, with conditions still bad and a different president, it presumably would have been far worse.
The positive response technical we would hope for, that we would learn to use better precautions when researching and storing deadly viruses, which we should clearly do regardless of the true origin here (if you look at the precautions they were taking, they clearly weren’t sufficiently robust, and the history of lab leaks is long), seems likely instead to cause a stupid response, regulatory and otherwise, that hugely raises costs in both dollars and optics, and reduces our ability to do research to prevent future pandemics or otherwise make scientific progress, without much in the way of gains to safety.
This was an additional reason I had no interest in pursuing the story of a potential laboratory leak, and made a decision for myself to leave it alone. I draw a big distinction between choosing not to investigate and discuss something and suppressing it via misleading people and actively suppressing dissent, especially via censorship, but it’s also important to know what motivations are going on in your sources of information, and what sources are going on in one’s own head.
I also think that same instinct of ‘why would you go looking into this, it can only cause bad trouble and not good trouble’ then leads to people blaming those who looked into it when they turn out to be right, and thus they expect vindication from the system and instead mostly get more blame. In elite eyes, being seen as right only makes their actions that much worse. Already a New York Times reporter whose beat is primarily Covid-19 is floating that the claim of a lab leak is still racist. No, Scott Alexander wasn’t an isolated incident, and I’m not going to lift the NYT ban any time soon.
Going forward, I intend to continue doing what I can to not cover the question of Covid’s origins, so my silence should be treated as very weak evidence of anything. That does not mean I will succeed, as circumstances are not making this easy, but I’m not going to automatically post if my posterior on this changes from week to week.
In Other News
Turns out the Pfizer vaccine never needed those special freezers.
DeSantis actively trying to stop cruise ship from requiring its passengers be vaccinated.
Meanwhile, many countries that got vaccine allocations are at risk of having them expire unused.
Here’s some obvious nonsense for the week, I leave the math this is implying as an exercise to the reader, while noting that most future workers won’t have been in school at all:
Restaurant reservations back to baseline levels. This matches my personal behavior, except that I’m still in Warwick instead of New York City for now which makes restaurants less appealing for now.
New York City will not be offering remote learning in the fall.
New York also lifts the mask requirement for 2-to-5 year olds at camps. Yay sanity.
Zeynep thread about school infection rates. Masks and ventilation work but not as much as one would hope, and barriers and desk spacing were the goggles and did nothing.
A political analysis asks why Biden has high ratings on the pandemic. It makes no mention of any of Biden’s policies, decisions, actions or statements, because it turns out none of that matters whatsoever. Presumably there’s a point at which something would matter, but we are still waiting to prove that via example.
Marginal Revolution shares this story accusing Noble Laureate Levitt of being irresponsible and misleading via having a Covid opinion that wasn’t sufficiently concerned and serious, and being unacceptably confident about it, while having authority and respect of some kind. Unacceptable. There are some real accusations here, saying that Levitt tried to privately get a critic’s funding pulled, and that he didn’t allow for proper review of his findings. I don’t think his conclusions made sense given the data available, either. Yet I have little doubt that if Levitt had aligned himself with the Very Serious People, no matter how extreme his conclusions or questionable his methods and results, we wouldn’t be looking at an article like this.
Post looking to explain why suicides didn’t rise during Covid-19. Offers a variety of possible explanations, some bottom lines being that depression due to objectively bad conditions is different than depression for internal reasons, the baseline happiness rate generally having less impact on suicidality than one might think (I have doubts but it’s plausible), people often come together in a crisis, and there may have been issues with lack of available methods of suicide slash many suicides may have shifted into drug or alchohol (so, drug) overdoses which are often at least kind of suicides. It’s not mentioned but I suspect another part is that when things are depressing because we’re worried about infection and death, suicide is a weird response, since one could instead throw caution to the wind and suicide feels even more than usual like a betrayal, or something like that. I dunno. Like everyone else, I was most definitely surprised to see suicide rates not rising.
Not Covid, but in terms of ‘media credibility reaches new all time lows’ the Associated Press has let FanDuel buy the exclusive right to have only its odds quoted on sporting events. That would be bad enough, tying AP’s sports odds to a recreational book whose odds are distorted and a severely compromised source of the odds of various outcomes, but they’ve also agreed to let FanDuel embed widgets of select content into coverage, turning AP content into ads for a recreational sportsbook with a predatory business model. Where that predatory business model is things like ‘buy your way into things like the Associated Press coverage to capture suckers, er I mean customers, who don’t know any better.’ Why should we assume their coverage of other things is better?
Not Covid, but worth noting that in highly developed WEIRD countries we are more likely to have someone we believe we can count on than in other places, not less likely.
EDIT: In a story that I misunderstood and spread in error, and isn’t a huge deal but also definitely isn’t a great look, the FDA is going to hold hearings on some other very dangerous substances (but has not, in fact, lost their f***ing minds any more than they already had before):
The full list is here. EDIT: They are not, as of yet, actively coming for your melatonin, or the other things on this list of suspicious products, which include: Asparagus, molasses, bean (yes, bean), grape seed oil, mustard oil, non-fat dry milk, nutmeg oil, parsley (the bitter herb!), pine tar (your bat needs a prescription now), sesame seeds (so we go with General Tso’s now I suppose), soybean protein, and sure, why not, let’s just hold nothing back and include sugars. And yeast. Can’t have anyone going around making unauthorized breads.
I misunderstood what was going on, which is on me, although one ponders why the misinterpretation was plausible enough to happen in the first place. It turns out that all this is, is that these are components companies want to use in their drugs, all of which require holding hearings (again, no unauthorized breads!) and category 1 here is things they think they’ll probably say yes to once the genuflections are complete, whereas categories 2 and 3, which contain much of the above list, are things that they’ll likely say no to because that’s crazy talk and the public must be protected.
We regret and apologize for the error, and thank those who pointed it out.
Meanwhile, the actual damage is this:
For now it’s still relatively easy to get in practice, but it’s basic get-people-through-the-day stuff that several of my friends rely upon.
While things are not as crazy as all that, I still stand by the overdetermined conclusion: FDA Delenda Est. Seriously. Burn the buildings to the ground. Salt the Earth. Wave, and send a message to the next ten generations.
I am pretty sure the FDA thing is just simply completely factually wrong, actual fake news in the original sense of the phrase.
It is a small change that only applies to bulk drugs in compounding pharmacies and is being widely misinterpreted… near as I can tell they are reviewing four supplements including melatonin for inclusion on the 503A Bulk List, which would make it legal for compounding pharmacies to use them, with that long list shown being things they are continually looking at for ADDING TO AUTHORIZATION FOR COMPOUNDING PHARMACIES. List two (not pictured, later in the document) is for things that are anti-recommended for compounding pharmacy approval, and list 3 is things that were recommended but they don’t think they have enough evidence for approving for compounding pharmacies. No banning of supplements anywhere, nothing going away at all in any way shape or form in that document. In fact, it means the OPPOSITE of what is being implied here!
N-acetyl cysteine does not appear anywhere in that document and appears to be an entirely separate and self-contained matter that refers to that substance and that subestance alone. Seems to have been going on for a year now having to do with them enforcing something that was technically true but unenforced before, that it was marketed as a drug decades ago before people started selling it as a supplement. Stupid and definitely an error but not a new regulation and companies are covering their butts.
Nobody is regulating parsley, sesame seeds, ribose sugar, and yeast as drugs. This list should have been a dead giveaway and I am flabbergasted that anyone can take it seriously as some kind of ban list. The referenced twitter accounts appear to be simply forwarding falsehoods and some seem to be kind of… interesting.
Confirmed that I misunderstood this. I didn’t think they wanted to ban that stuff or anything, but the EU heavily regulates cinnamon, so I honestly don’t know what to expect anymore. Still, this week was hectic and I didn’t give this the time it needed, got it wrong, and screwed up, so sorry about that. It’s been edited to reflect that.
There is absolutely nothing about the biology of the virus that suggests laboratory anything. Anyone who says it does does not know what they are talking about. If anyone says something about furin being suspicious or indels being unlikely, or recombination patterns being odd, they can safely be ignored and their opinions discounted. Yes, this includes David Baltimore.
One cannot exclude the possibility that someone had a stock in a lab being studied somewhere and it got into a lab worker by sheer bad luck. That would look quite similar to the utterly normal zoonotic spillovers that happen all the time, especially for something like this that is a tenth as deadly as SARS classic such that the early thin transmission chains are effectively invisible to public health infrastructure.
SARS classic has infected lab workers before, but as something that caused a previous global issue it was being studied by many people in many places while in the scenario of a novel zoonotic transfer having occurred in a lab, you almost certainly just have one poorly characterized sample of something with no published sequence data someone happens to be growing for a quick one-off experiment spilling over such that the rate would presumably be much much lower.
I see no reason to think that this is particularly likely over and above utterly normal outside-world zoonotic spillover that happen all the freaking time (several percent of Chinese villagers tested for antibodies against SARS-like viruses that live near bat caves have them in previous literature), especially since the invisibility of early thin overdispersed transmission chains mean the first big outbreak could occur somewhere quite different than the location of the initial zoonotic event. Even SARS-classic had its first big internationally-noticeable explosion happen thousands of miles away from the initial site of zoonotic spillover, and if it were not as deadly as it was that initial spillover in southern China could potentially have gone unnoticed outside of local authorities and you could’ve been talking about SARS as originating in Hong Kong.
All this being said given the state of recombinant DNA technology and its power I do see the value in most virology work using reconstituted sequence in pseudovirus systems rather than infectious particles.
There are multiple hypotheses and insufficient evidence to settle on just one.
The ‘gain of function’ experimental design—where a chain of lab animals are used, with a slightly harder to cross barrier between each animal—would cause similar ‘natural mutation’ patterns. The difference is that it makes the actual creation of a novel pandemic causing virus many many times as likely, as this same infection chain has to occur by chance in nature.
What we have now is like looking at the residue of a nuclear meltdown but we can’t examine the actual reactor, and the owners of the territory the meltdown occurred in are actively suppressing evidence. Nature can produce a nuclear reactor and has at least once it just isn’t likely.
That… seems like almost precisely the opposite metaphor one would use to compare to reality? Instead we are looking at something that happens naturally ALL THE TIME and you need special effort to sometimes maybe replicate if you really wanted to?
Do we have a quantitative measurement for “all the time”? We have in living memory the emergence of HIV which presumably also came from an animal host initially. And the previous 2 variants of covid which were not very contagious.
Please note I am not “convinced” either way. I am just noting a gain of function experiment is a specific set of conditions that might take nature decades to centuries to replicate by chance. It is a plausible method for the virus evolving. The other way being that lab field workers are going to collect more exotic specimens than commercial meat sellers, going deeper into caves,etc. All it would have taken is a mistake or counterfeit equipment such as HEPA filters, a problem that appears to be more common with current Chinese industries than in equipment from more mature name brand western companies.
https://en.m.wikipedia.org/wiki/Zoonosis#Lists_of_diseases
From the article:
Quite off-topic: what does it mean from a Bayesian perspective to hold a probability weakly vs. confidently? Likelihood ratios for updating are independent of the prior so a weakly-held probability should update exactly as a confidently-held one. Is there a way to quantifiy the “strongness” with which one holds a probability?
Imagine you have a coin of unknown bias (taken to be uniform on [0,1]).
If you flip this coin and get a heads (an event of initial probability 1⁄2), you update the prior strongly and your probability of heads on the next flip is 2⁄3.
Now suppose instead you have already flipped the coin two million times, and got a million heads and a million tails. The probability of heads on the next flip is still 1⁄2; however, you will barely update on that, and the probability of another heads after that is barely above 1⁄2[1].
In the first case you have no evidence either way, in the second case you have strong evidence either way, and so things update less.
In terms of odds ratios, let H be your hypothesis (with negative ¬H), B your past observation, and B’ your future observation.
Then O(H|B’,B) = P(B’|H,B) / P(B’|¬H,B) * O(H|B).
The Bayes factor is P(B’|H,B) / P(B’|¬H,B). If you’ve made a lot of observations in B, then this odds ratio might be close to 1. It’s not the same thing as P(B’|H) / P(B’|¬H), which might be very different from 1. Why? Because P(B’|H,B) / P(B’|¬H,B) measures how likely B’ is, given H and B versus how likely it is, given ¬H and B. The B might completely screen off the effect of H versus ¬H.
In a court case, for example, if you’ve already established a witness is untrustworthy (B), then their claims (B’) have little weight, and are pretty independent of guilt or not (H vs ¬H) - even if the claims would have weight if you didn’t know their trustworthiness.
Note you can still get massive updates if B’ is pretty independent of B. So if someone brings in camera footage of the crime, that has no connection with the previous witness’s trustworthiness, and can throw the odds strongly in one direction or another (in equation, independence means that P(B’|H,B) / P(B’|¬H,B) = P(B’|H) / P(B’|¬H)).
So:
This means that they expect that it’s quite likely that there is evidence out there that could change their mind (which makes sense, as they haven’t looked carefully). They would have a strongly held probability if they had looked at all the available evidence and converged on 40% at the end of weighing it all up; it’s unlikely that there’s anything major they missed, so they don’t expect anything new to change their estimate much.
It’s (106+2)/(2∗106+3)=0.50000025, I believe.
Thanks, I think this is the crucial point for me. I was implicitly operating under the assumption that the evidence is uncorrelated which is of course not warranted in most cases.
So if we have already updated on a lot of evidence, it is often reasonable to expect that part of what future evidence can tell us is already included in these updates. I think I wouldn’t say that the likelihood ratio is independent of the prior anymore. In most cases, they have a common dependency on past evidence.
Yep, that seems to be right. One minor caveat; instead of
I’d say something like:
“Past evidence affects how we interpret future evidence, sometimes weakening its impact.”
Thinking of the untrustworthy witness example, I wouldn’t say that “the witness’s testimony is already included in the fact that they are untrustworthy” (=”part of B’ already included in B”), but I would say “the fact they are untrustworthy affects how we interpret their testimony” (=”B affects how we interpret B’ ”).
But that’s a minor caveat.
This is kind of technically true, but not in a practical sense. As you learn more about most systems, the likelihood ratio should likely go down for each additional point of evidence. The likelihood ratio for an event X is after all P(X|E):P(¬X|E) where the E refers to all the previous observations you’ve made that are now integrated in your prior.
Usually when referring to “updating on En we use the likelihood ratio
P(En|E1,E2,...,En−1):P(¬En|E1,E2,...,En−1)which kind of makes it clear that this will depend on the order of the different Ei.
I’d be interested to see the assumptions which go into this. As Stuart has pointed out, it’s got to do with how correlated the evidence is. And for fat-tailed distributions we probably should expect to be surprised at a constant rate.
I’ve been recommending the Rootclaim article on the Covid origin question for a while as an example of Bayesian reasoning (with likelihood ratios, priors and posteriors, etc.) that features reasoning so transparent that it seems valuable irrespective of whether it’s correct or not. That is, if your priors on the various Covid origin hypotheses differ, or your interpretation of various pieces of evidence differs, then your conclusions will also differ, but at least you could argue fruitfully with the authors of the piece.
Very interesting page! I think the most Bayesian flaw is not conditioning on earlier evidence in some places. The equivalent of trying to evaluate P(virus with properties A, B, and C) as P(A) x P(B) x P(C), rather than the correct method P(A) x P(B given A) x P(C given A and B).
I’m specifically thinking of how the genetics of cov-2 are not independent of its adaptedness to humans. Once you condition on the unlikely genetics, it shouldn’t be as unlikely that you see transmission in humans, and vice versa.
Substantively, I’m not sure whether coincidences are being cherrypicked here—there are lots of cleavage sites in viruses, and lots of other viruses to compare chunks of genes to. How many similar coincidences should we expect just due to chance? Basically if you update on the coincidences you see but never correct for the number of possible coincidences you could have seen, you’ll overrate how much evidence coincidences give.
I also feel like they put a lot of thought into evidence capable of testing a normal zoonotic origin ( as is proper—https://www.lesswrong.com/posts/rmAbiEKQDpDnZzcRf/positive-bias-look-into-the-dark ), but didn’t put forth similar effort into what goes against the genetic engineering hypothesis. How likely is it that a research lab finds a new bat coronavirus and then before publishing anything about it, decides that it’s the perfect testbed for dramatic gain of function research? This likelihood could be evaluated by checking for prior examples of ambitious research using a microbe before anything about it had been published. This sort of thing is missing.
A hell of a lot of what they say about furin is simply wrong. Insertions and deletions happen ALL THE TIME in nature and while rarer than SNPs are not uncommon, around the cleavage site loop there are other indels in related viruses, the furin site is DECIDEDLY suboptimal for cleavage such that all the fun lineages that are more contagious are optimizing it and creating actual canonical sites rather than something just good enough, it is generated out of frame in in a way that no sane biologist ever would, and the glycosylation pattern nearby weakly suggests it appeared in the context of an actual immune system rather than cell culture.
The alignments they point to arguing for other differences in the cleavage sites in other viruses being due to SNPs rather than insertions are laughable, and from people who I have found to have a poor grasp of the underlying science.
On top of all that when conditioning on something that successfully changes species you will enrich for things that broaden their tropism, which that kind of cleavage site will do, so of course things that are rarer but helpful will pop up more than if you were looking at the space of all possible mutations without conditioning.
In China? We’re talking about a virus based on RNA, which mutates more easily, making it harder to control. China’s government craves control to the point of trying to censor all mentions of Winnie the Pooh, possibly because a loss of control could mean the guillotine.
I think this is too cynical.
FiveThirtyEight is analyzing a bunch of existing polls. I expect that none of them asked questions more specific than overall assessment of how Biden was doing on COVID-19. If someone did do a poll asking more specific subquestions—does he do messaging better, do they credit him with beating his vaccine rollout target, etc—you’d probably see some details emerge.
(His messaging hasn’t been ideal, of course, but it’s raised the bar from the last guy. And his rollout speed was pretty good, but he also managed expectations smartly. Etc.)
Of course, the biggest reason that people rate him highly is just that, well, things are going Back To Normal and that means he’s doing a good job on it. That’s not a super sophisticated analysis on their part, but there are worse things for low-info voters to do than to support the ruling party when things go well and oppose it when things go badly, without trying to divine the causality.
A synchronized theory.
Conspiracy theory:
A) There is a conspiracy
B) There is a conspiracy to spread this fake theory (with no evidence)
Here’s the thing about B—those do actually exist. For example, scams.