Thinking about the impact of Rationalist writing after a small blog post made a big difference. Cross-posted from Putanumonit.

Changing British Minds

Even a year after the fact, it’s difficult to compile an accurate timeline of the UK Government’s thinking on COVID and how it changed in spring 2020.

The first COVID case in the UK was confirmed on January 31. Spread continued throughout February, and the British government published its first Coronavirus action plan on March 3. It focused on the tracing and isolation of known cases while saying that it “will aim to minimise the social and economic impact” and that “it may be that widespread exposure in the UK is inevitable”.

The decisions ultimately came down to PM Boris Johnson and his chief adviser, Dominic Cummings. At the beginning of March, according to the Sunday Times, Cummings leaned towards a strategy of herd immunity: having the majority of the UK’s population catch the virus throughout the summer months to protect the economy and put Britain in the clear by winter. The PM’s office denied this claim and alleged that the quotes attributed to Cummings by the Sunday Times were fabricated. In any case there were clearly heated debates within Johnson’s inner circle on the costs and benefits of mitigation.

Whatever the stance was in the first two weeks of March, by March 12 things began to shift. The strategic advisory group of experts warned that the UK would suffer 500,000 deaths if the virus was left unmitigated, and Cummings himself began pushing vigorously for a shut down. The UK Government began implementing mitigation measures on March 13, and on March 23 announced a full lockdown of the entire nation. The UK spring COVID wave peaked at 70 daily cases per million, significantly below the US and many European nations.

Oh, there’s one other thing that may have changed the mind of Dominic Cummings and the fate of Britain in early March.

Post Impact

The four hours I spent writing Seeing the Smoke on a whim were more impactful than projects I spent months of my life on. I calculated that if just 1% of the 20,000 people who read that post the week it came out avoided catching COVID and infecting others during a period when the curve was steepening, the post probably saved several lives. To learn that my post impacted policy for 67 million Britons is a whole other level of bonkers.

I’m struggling a bit to make sense of it all. My immediate motivation for writing is simply to get the thoughts out of my head, along with a cycle of anxiety about disappointing my readers if I go too long between posts. Aside from the satisfaction of having my post out there, my main reward for writing is making friends — even in 2020 I talked online or in person with dozens of people who I connected with in part because of the blog. On occasion I find out that some famous person mentioned one of my posts on a podcast or something. It’s always a pleasant surprise, but I’m never sure what to do with this. I don’t write to become famous and change the world. And yet somehow…

My first thoughts when Cummings mentioned that my post influenced British COVID policy was: holy shit! My second: if I had known you will read it, I would have written a better post! I would have done more research. I would have thought more about the impact of government interventions. I would have made some quantified predictions and recommended specific policies.

None of that would have made the post any better or more useful. In late February I didn’t know of a better policy beyond “shut everything down including travel and take some time to think this through”. By the middle of March I was convinced of the Hammer and Dance approach: suppress the pandemic by instituting complete shut downs (including travel) at the first sign of R0>1, since the timeliness of lockdown is the most important factor in their success. Then re-open equally quickly to maximize the proportion of time in non-lockdown over the 1-2 years it would take to make vaccines. In my understanding, this is basically the strategy that Australia and New Zealand successfully executed.

But on the day Hammer and Dance was published the UK already had 10 times New Zealand’s per capita case rate and R0 north of 2. As a result, the lockdown in the UK dragged into the summer instead of lasting only 3-4 weeks. The government had to make up for the economic damage with schemes like “eat out to help out” (that’s what she said) in the summer, then suffered a second COVID wave in the fall that inflicted all the damage the spring lockdown was supposed to avoid and likely gave rise to the more dangerous B1.1.7 variant.


Even with a year of hindsight and my current understanding of political pressures, lockdown fatigue, and equilibrating control systems, I’m not confident whether it was good or bad for the UK to lock down in March 2020. In the moment I was surely no wiser.

Would I have made a bigger impact by focusing more on COVID research and running the numbers? In March I was busy reading up on hydroxychloroquine, which turned out to be mostly a waste of time. In June I was doing math to convince people that it’s safe to meet up with small groups of friends, but all the people who hailed my genius in February now called me irresponsible and arrogant. It turns out that they needed permission to start worrying about COVID in the spring, but didn’t want permission to stop worrying in the summer.

In December I wrote a follow up post about the B1.1.7 variant, predicting a relatively high chance of massive infection wave in the spring. Judging by the recent uptick in cases I got the timing and importance of new strains right, as well as the reluctance to lock down. But I downplayed the possibility of vaccines having a big impact, which they have. William Kiely got the vaccine math spot-on in the comments, but I don’t know if many people read that or cared. The main impact of that post may have been to get people into an options trade that ended up losing all the money.

So: I’m not a COVID expert, and playing one on the internet is not a reliable way for me to have a major positive impact on the world. But the original post wasn’t really about COVID. It was public psychology, how people act as if appearing weird is worse than any disease, and that this is true not just of media consumers but also of the journalists and “experts” who produce it. What impact could I have through this deep and novel understanding of how everyone thinks?


Thinking that one understands how the public thinks and how to shape it invites the worst sort of hubris.

While the idea of benevolent rulers shaping the public mind goes back at least to Plato, it’s most recent incarnation goes back to the 2008 book Nudge by Richard Thaler and Cass Sunstein. The book repeats the Kahnemanian critique of “homo economicus” and gives some common sense advice that people knew long before Kahneman was born, such as that shoppers are more likely to pick the items displayed at eye level on the shelf. It then pivots to a defense of “libertarian paternalism” — a doctrine of shaping the public mind through control of the public’s choices by the government.

“Libertarian paternalism” is really a rebranding of “manufacturing consent” for democracies with some fuzzy anecdotes about cab driver tips and a patina of scientism. Sunstein himself is quite supportive of manufacturing consent by direct means, free speech be damned. In any case, Nudge landed Sunstein a job as the administrator of the Office of Information and Regulatory Affairs under Obama, where presumably he proceeded to regulate and inform the nation according to his theories.

In 2020, every terrible decision made by the experts in charge was justified by appealing to the effects on public psychology. I don’t know if this was inspired or instigated by Sunstein directly, but I like to call this phenomenon nudgerism. I despise nudgerism.

Since one can invoke a “bias” for any side of any decision, you’d expect the universal application of nudgerism to have a 50% hit-rate. Yet somehow it seemed to do much worse than chance, probably because any non-terrible decision could be justified by actual evidence instead of the appeal to psychology.

COVID nudgerism started with Sunstein himself, of course. Other “experts” downplayed the risk of COVID to “prevent panic”. Then they told people not to wear masks or get tested to avoid “a false sense of security“. Scrubbing surfaces was still recommended, even though unlike masks and testing it probably doesn’t do shit for COVID. “Experts” said that challenge trials would reduce trust in the vaccine among normal people; normal people turned out to support challenge trials by an overwhelming 75-90% majority. “Experts” said that delaying the second dose of the vaccine or allowing the AstraZeneca shot would increase vaccine hesitancy. Britain delayed the second dose, is giving everyone AstraZeneca, and has the least vaccine hesitancy in the world. Probably the British government reads Marginal Revolution along with Putanumonit, and ignoring Sunstein.

I don’t think all these nudgers are wholly cynical about this. I think their inside view finds their theories of public psychology reasonable and valid. And while I will mock and deride them in any opportunity, I doubt I’d do better if I tried it myself. Not in small part because I’m an alien.

Coordinating Social Reality

A big theme in my writing since Seeing the Smoke has been coming to terms with the gargantuan gap between the way I think and the way most people do. It may even be wrong to use the word “think” for both. Michael Vassar says that what Rationalists call “thinking” is treated by most people as a rare technical ability (“design thinking”) that normal people can only pretend to do. What they call “thinking” we call “being depressed and anxious”. This sounded crazy when I first heard it, but the more I mulled it over the more it made sense and explained much of what has been happening in the last year.

Social reality is what is normal, accepted, cool, predictable, expected, rewarded, agreed upon. Physical reality is what is out there determining the outcomes of physical experiments, such as whether you get COVID or not if you wear a mask. When Rationalists say “thinking” they usually mean something like “using your effortful system 2 to determine something about physical reality”. It’s what I try to do when writing posts about COVID. Swimming in social reality is best done on feeling and intuition, not “thought”.

My experience is spending perhaps 97% of my time in social reality, swimming along with everyone else. 3% of the time I notice some confusion, an unexpected mismatch between my predictions and what physical reality hits me with, and try to think through a solution. 3% is enough to notice the difference between the two modes and to be able to switch between them on purpose.

I don’t think that this experience is typical.

With Vassar in mind, my best guess of the typical experience is being in social reality 99.9% of the time. The 0.1% are extreme shocks, cases when physical reality kicks someone so far off-script they are forced to confront it directly. These experiences are extremely unpleasant, and processing them appears as “depression and anxiety”. One looks at the first opportunity to dive back into the safety of social reality, in the form of a communal narrative that “makes sense” of what happened and suggests an appropriate course of action.

I think this explains why so many people don’t seem to notice or care that even institutions they consider to be “on their side”, like the CDC and the New York Times for the educated progressive tribe, are wrong or lying all the time. People look to those sources not for “truth” about physical reality but for coordination of social reality. The CDC’s job is to tell other institutions which policies they can implement and not get blamed for, not which policies will keep their clients healthy. People read the Times not to find out what happened where and when, but to find out who is to be comforted and who afflicted. People just want to be on the same page as their peers.

Seeing the Smoke came out during the 0.1% of the time when physical reality was manifesting and the institutions of social reality hadn’t reacted adequately yet (in spinning it into the narrative). By the summer, social reality reasserted itself. Whether someone was masking or not, locking down or protesting, eating fish tank cleaner or Lysol-sprayed Uber Eats was due mostly to their tribe membership, not to a physical model of the virus. No one cared about my microCOVID calculations.

Zvi, who is better than me at thinking about many things including COVID, explains why being good at thinking doesn’t mean one could be put in charge and change the world for the better. The entire post is worth reading, especially for young Rationalists just coming to grips with how non-Rationalist the rest of the world is. If I had to summarize it in one sentence: being correct is not the same as being good at coordinating social reality, and what those “in charge” really do is the latter.

The Bayesian Monastery

Zvi lays out a model of how Rationalists like him can influence policy:

Anna Salamon suggested a model of a rising sanity water line, but in the sense that this makes it harder to stay above water and thus directly sane. There’s a small and decreasing number of people who are still capable of synthesizing information and creating new hypotheses and interpretations. 

Then there’s those who are mostly no longer capable of doing that, things got too complicated and weird, and they can’t keep up, but they can read the first group and meaningfully distinguish between people claiming to be in it, and between their individual claims, ask questions and help provide feedback. To them, the first group is legible. This forms a larger second group that can synthesize the points from the first group, and turn it into something that can be read as an emerging new consensus, which in turn can be legible to a third much larger group. 

This third group can then be legible to the general public slash general elites, who learn that this is where good new ideas come from. Then the Responsible Authority Figures can feel under public pressure, or see what the emerging new ideas are, and run with the ball from there, and the loop continues.

The filtering process also acts as a selection for feasibility, as the second layer picks up things from the first that it thinks it can present legibly to the third, and so on.

This model doesn’t mean that self-identified Rationalists are always in the first group. Most of what I write, especially about COVID, is second or third layer (or just nonsense). I think that Rationality gives people two important things: the tools to evaluate original thinkers without relying on mere credentials, and the permission to occasionally shoot for first-level insight themselves. As a community, we are closer to the surface of the sanity waterline than most and thus, by necessity, farther from political power and institutional authority.

Byrne Hobart makes the same point in a post about how Rationalists got COVID right and early:

This puts the rationalists in a uniquely prosocial position. They’re a sort of distributed, mostly open-source monastic order, spending a lot of time contemplating the world and passing down important observations, but less time directly interacting with it. The influence of people who read rationalist blogs, but don’t self-identify as rationalists, is quite wide—the blogs are very widely followed in technology circles, and anecdotally have a large audience in the more quantitative branches of finance.

Byrne goes on to say that “identifying as a Rationalist is a losing move”, but I think that he presupposes that everyone is playing the same game. Joining a monastic order is a “losing move” if your goal is to inherit titles and command knights, but the life of a monk has much to recommend it over the life of the medieval court and battlefield. The pursuit of wisdom and the pursuit of power are usually at odds. Identifying as a Rationalist is a small way to nudge yourself (heh) toward the former.

A typical LessWrong meetup

In March 2020 Dominic Cummings was in a unique position to bridge the many layers between first-level analysis and “Responsible Authority Figures” by himself. He could take word from the monastery directly to the throne room. I just happened to be the monk that was on thinking duty that day.

But that doesn’t mean that I should now abandon the monastery and fancy myself a vizier. In general, I try to write about enduring topics like statistics and romance and how our brains work, not about breaking news. Instead of trying to coordinate social reality for the masses, I try to help people become slightly more independent of it. To avoid the mind-kill of political polarization, to get in touch with their own desires when deciding what to spend money on and whom to date, to spend some time in our monastery.

I hear individually from people that my writing is impacting them, and as long as those DMs come I’m content. I think my writing has some broader influence, but by the time it is passed down to actual influencers I would not be credited with it. And that’s fine! I’m really not looking for credit, and now that I got some for influencing the UK lockdown decision I don’t even know if it’s deserved or if the decision was the right one. I recommend that if anyone in power wants to hire a Rationalist advisor they do so secretly, and pay them based on the calibrated accuracy of their predictions only.

In the meantime, I’ll try to keep writing as if no one important will ever read it at all. Otherwise, the temptation grows to climb the simulacra levels away from reality, to signal loyalties and “nudge” the public and play 4D chess on a backgammon board. And that’s not how we do it in the monastery.

New Comment
41 comments, sorted by Click to highlight new comments since:

Re "nudgers", compare Alex Tabarrok in Ezra Klein's recent article:

In all of this, the same issue recurs: What should regulators do when there’s an idea that might work to save a large number of lives and appears to be safe in early testing but there isn’t time to run large studies? “People say things like, ‘You shouldn’t cut corners,’” Tabarrok told me. “But that’s stupid. Of course you should cut corners when you need to get somewhere fast. Ambulances go through red lights!”

One problem is no one, on either side of this debate, really knows what will and won’t destroy public trust. Britain, which has been one of the most flexible in its approach to vaccines, has less vaccine hesitancy than Germany or the United States. But is that because of regulatory decisions, policy decisions, population characteristics, history, political leadership or some other factor? Scientists and politicians are jointly managing public psychology, and they’re just guessing. If a faster, looser F.D.A. would lose public trust, that’s a good reason not to have a faster, looser F.D.A. But that’s a possibility, not a fact.

“My view is this was all psychology which no one really understood, so I just said, ‘Go with the expected value. Do the thing that’ll save the most lives and stick with it,’” Tabarrok said. “That’s a better rule than trying to figure out ‘If I do this, what will someone else do?’”

I don't think that nudgers are consequentialists who also try to accurately account for public psychology. I think 99% of the time they are doing something for non-consequentialist reasons, and using public psychology as a rationalization. Ezra Klein pretty explicitly cares about advancing various political factions above mere policy outcomes, IIRC on a recent 80,000 Hours podcast Rob was trying to talk about outcomes and Klein ignored him to say that it's bad politics.

This part:

People read the Times not to find out what happened where and when, but to find out who is to be comforted and who afflicted. People just want to be on the same page as their peers.

reminds me of Scott Alexander on the phatic:

Consider a very formulaic conservative radio show. Every week, the host talks about some scandal that liberals have been involved in. Then she explains why it means the country is going to hell. I don’t think the listeners really care that a school in Vermont has banned Christmas decorations or whatever. The point is to convey this vague undercurrent of “Hey, there are other people out there who think like you, we all agree with you, you’re a good person, you can just sit here and listen and feel reassured that you’re right.” Anything vaguely conservative in content will be equally effective, regardless of whether the listener cares about the particular issue.


my best guess of the typical experience is being in social reality 99.9% of the time. The 0.1% are extreme shocks, cases when physical reality kicks someone so far off-script they are forced to confront it directly. These experiences are extremely unpleasant, and processing them appears as “depression and anxiety”. One looks at the first opportunity to dive back into the safety of social reality, in the form of a communal narrative that “makes sense” of what happened and suggests an appropriate course of action.

Really? Shouldn't "typical experience" include small business owners running sales forecasts, truckers navigating new environments, and a contractor building a staircase? It seems to me that lots of normal people contend with novel situations in objective reality on a regular basis. What really seems noteworthy to me is how domain-specific that mode of thought tends to be. A guy who builds houses can tell when some new construction regulation is not reality-based, but he will not think twice about questionable statements from the CDC.

Notice how your examples are working class. The middle class or Venkat's aptly named clueless are maximally insulated from causal reality.

Social reality topics are often things in the news. When I was a student I realised I could just stop following the news because it almost never affected me. Only maybe once a year was there anything in the news that I needed to know, because it would affect my short-term actions. (COVID of course being a notable exception.)

The kind of serious news I follow is about real-world events - albeit in politics, things happening in other countries, etc. - causally distant from me. Not local news, which may be the most likely to affect me, though in some trivial way (a new store opening or something).

Those who follow celebrity culture etc. are even less affected by the 'news' they follow, except I suppose insofar as it's about new films, albums etc. which they might see/buy. Indeed such people see the news more like what it is. Its main effect is in the meta (= social reality) realm, as a source of talking points. You 'need' to know the news in order to join in conversations with your friends about the news.

The relevance or truth of the news is beside the point. For the same applies to the 'need' to read Harry Potter, if everyone else is talking about it.

I think it's easy to denigrate (almost exclusive) trackers of social reality. But they are people who experienced conditioning that most of the variance in the outcomes they care about were controlled by social reality inputs. It makes sense that they constantly have to track changes in social reality because social reality is anti-inductive/adversarially optimizing out from under you constantly. Being able to ignore large swaths of social reality and thus bootstrap more permanent wins (since solutions in causal reality often stay wins) is something of a privilege. People mostly tracking causal reality also pay an often invisible cost related to having social reality treat you as a defector.


On a side issue, as you probably know but other readers may not, Dominic Cummings was central to another case of social reality. For he was subsequently turned into public enemy no. 1 by the British media, when he broke lockdown rules to drive his family across the country to his parents' home. Most of the public had never heard of Cummings, but he had apparently made enemies in the media (as well as government) by treating them with disdain, and this was their chance for payback.

And so, in a trial by media over several days, it was amazing to see how easily almost everyone in the UK was persuaded that Cummings was the devil incarnate, which continues to this day. (I broke ranks to post a defence of Cummings, or rather a criticism of the public's ill-founded view of him, on Facebook, which got a lively response.)

Just as amazing was the opposite attitude to the BLM protests in the UK soon after; I don't think it even occurred to 99% of the public that those were just as illegal as Cummings' trip. And a politician (Stephen Kinnock) who made a similar cross-country drive to Cummings on the very same day, to visit his famous father, former Labour leader Neil Kinnock, attracted almost no media coverage or criticism. This all showing that the law was quite beside the point - merely providing a pretext for demonizing Cummings.

Anyway, in the face of all this media and public pressure, Boris Johnson spent much political capital refusing to fire Cummings, as he was said to be 'Boris's brain' and by far the smartest person in Downing St. Boris even extraordinarily granted Cummings (a mere adviser) his own press conference in the Downing St garden, in which Cummings presented an implausible account of events exonerating himself, to general derision.*

The whole incident provided a pretext (that phrase again) for many Britons subsequently to break lockdown rules. The public mood changed immediately from a wartime spirit of doggedly following government advice to half-disregarding it - often explicitly saying "If Dominic Cummings can drive to Barnard Castle, then I don't see why I shouldn't do XYZ". Regrettably it's likely this has significantly increased COVID cases ever since.

(*I reckon it should have been played like this: Cummings should have admitted breaking the rules (perhaps inadvertently), and offered his resignation. But Boris should have reluctantly refused the resignation, on the grounds of not rocking the boat in a national emergency, and maybe accepted a fine as punishment.)


Cummings was already pretty unpopular before the Barnard Castle thing.

Some of the unpopularity was because of specific opinions and attitudes Cummings was alleged to have; for instance, he played a big role in the Brexit campaign, so Remainers (somewhere around half the country) didn't like him, and he was somewhat-credibly alleged to be in favour of a "herd immunity" strategy against Covid-19, which may or may not have been a reasonable idea given what was known at the time but a lot of people regarded as callous.

I'm pretty sure that quite a lot of it was because he was openly, unashamedly geeky, in various senses of that word. A distressing amount of the criticism of him included mockery of his physical appearance.[1] He got a lot of flak for calling for "weirdos and misfits with odd skills" to apply for jobs with him, which I think was again mostly a matter of anti-geek prejudice. (One of those weirdos and misfits who got hired turned out to have said some 

At any rate, his lockdown-dodging wasn't by any means the first thing he was attacked for; to whatever extent his vilification was payback for disdaining the media, they got started on it well before that. And I think he would always have been an easy target, even without specific media malice, because unfortunately the population at large doesn't like weird geeky people.

(I'm not altogether convinced by the parallel you draw between Cummings's trip and the BLM protests. Most of the complaints against Cummings weren't so much "how dare he break the rules????!!" as "someone so directly associated with the government should be being extra-careful and not breaking the rules". Caesar's wife, etc. For that reason, I don't think it's necessarily unreasonable for someone to think what Cummings did was worse than what BLM protestors did.)

[1] I think this is excusable (if and?) only if either the person being attacked has deliberately exploited their physical appearance for some kind of gain, or whatever it is about their appearance is actually directly relevant to what's being discussed. Usually neither of these is true, and in particular I don't believe either has ever been the case for Dominic Cummings.

I think he was only known & unpopular among those who follow politics closely. I expect 80% or 90% of the UK hadn't previously heard of him. The media coverage of the incident turned him from a niche suspect figure into a universal hate figure.

Re your point [1], people associate physical appearance with attitude. I overhead someone in the street at the time saying of Cummings' press conference: "He's so arrogant! Did you see how he was dressed?" I.e. that Cummings was and is deliberately slovenly to show two fingers to the press/Establishment - i.e. that he doesn't care what they think. Which is probably the case. Or at least, the geeky view that how you dress shouldn't matter - the two of course being closely related.

(One of those weirdos and misfits who got hired turned out to have said some 



Oh, whoops. I meant to delete that.

(It had been going to be about Andrew Sabisky, who was one of Cummings's "weirdos and misfits" and resigned after it turned out that he had said a lot of politically very unpalatable things about race, eugenics, and the like. I'd thought I remembered that a lot of the complaints about Sabisky were attacking his weirdness and geekiness as much as his controversial opinions. But when I went back and checked the discussions I was thinking of, that didn't after all seem to be so, so I cut that bit out. Except that I somehow failed to cut all of it out.)

PS I just realized, one of the main reasons the general public fell in with demonizing Cummings was the very one you identified that delayed reaction to COVID: he seems weird, and reacting to a faroff disease which everyone else is ignoring would seem weird. And seeming weird is the worst thing in the world.


People don't like unelected power-behind-the-throne types either. Mandelson was Blair's Cummings, and he was pretty unpopular.

Characterising the reaction to Cummings as about being about people overreacting to a small violation of the rules is misleading. The issue wasn't the initial rule violation, it was that the initial denial and lack of even token punishment was symbolic of a wider issue in the Johnson government with corruption and cronyism. Caring about hypocrisy and corruption among leaders is entirely rational as it is indicative of how they will make other decisions in the future. 

This seems like a post-rationalization. IIRC the way it played out over a number of days was that initially it wasn't clear what the facts were, and hence what if anything Cummings had done wrong (e.g. whether his journey had been legal, or at least justified). And even if he had done something wrong, I heard one pundit point out that as Cummings wasn't a minister or public-facing figure there was no requirement for him to resign or be fired (rather than apologise or be disciplined in some way).

But nonetheless the media picture right from the start was that this maverick egg-head weirdo must be guilty of something, even if they weren't sure what exactly. And the public reacted accordingly.

For example, 3 days before Cummings' press conference (which IIRC was the first time his side of the story was fully set out) I heard a radio phone-in about what an evil character Cummings must be, in which callers were mostly accusing him of risking his parents' health by going to stay with them. Or saying he must have stopped at a petrol station and so risked people there (he denied this). It later turned out he hadn't even stayed in his parents' house, or had close contact with them, but stayed in another building nearby.

So then it was a question of, was his main journey illegal (with much detailed media analysis of the fine points of the law)? Or if not, how about the short trip to Barnard Castle? Which is what most people - the narrative - have now settled on.

What this all shows is that in this trial by media, Cummings was presumed guilty from the start; and then it was just a matter of finding some crime to pin on him. And once something was found that seemed enough like one, everyone could congratulate themselves that they'd 'known' all along, and so their outrage had always been justified.

(I can't recall which cognitive bias this is - but quite a typical example.)

(To avoid doubt, as I turned out I think it's very likely he broke the rules and adjusted his story to try to exonerate himself. And clearly Boris mishandled it badly. But my point isn't about whether he/they turned out to be in the wrong, it's about the fact the media had it in for Cummings, and had no trouble swaying the public accordingly.)

An interesting postscript: Cummings recently gave 7 hours evidence to the UK parliament about the government's COVID response, criticizing it heavily (my shortform summary here).

In this he gave more detail on his trip out of London that had attracted all the vilification. Apparently it followed a crowd of people gathering at his house threatening to kill his wife & children (after a false negative media story about him). Boris Johnson told him to leave London for his own safety, regardless of the rules. This effectively exonerates Cummings (for the main trip).

If this is true, I assume he didn't state this at the time to protect Boris, and one of the reasons Boris protected him was because Boris knew what had actually happened.

Incidentally in the BBC Radio 4 news report of Cummings' evidence, none of this was mentioned! Though they found time to interview locals around Barnard Castle about why they think Cummings is so nasty.

Hi Jacob, 

I think your Seeing the Smoke was interesting and the conclusions about human nature are right - particularly the point that most people will fail to do the obvious thing like leave the smoke-filled room out of fear of looking weird. That said, I really wish Cummings had drawn a different conclusion from your blogpost, because I strongly believe that lockdowns were the wrong response to Covid. Specifically, I would prefer Cummings had read Hans Rosling's excellent book Factfulness, especially chapter 10 on the urgency instinct: 

Rosling was investigating a disease in a remote area of Mozambique. The mayor of the nearest city asked Rosling if he should institute a roadblock to prevent sick people from coming to the city. Rosling wasn't even sure the disease was contagious, but he thought better safe than sorry and said yes, put up the roadblock. When the village women, some of them carrying infants, found the road to market blocked they asked local fishermen to carry them instead. The overloaded boats sank causing the death of about 20 people including the infants. Rosling later found that the disease was caused by eating improperly-prepared cassava, and wasn't contagious at all, meaning that the roadblock had killed those people for no reason.

Rosling blamed himself for those deaths for years afterwards. As he puts it: "Back in Nacala in 1981, I spent several days carefully investigating the disease but less than a minute of thinking about the consequences of closing the road. Urgency, fear, and a single-minded focus on the risks of the pandemic shut down my ability to think things through. In the rush to do something, I did something terrible." [My bold.]

I believe that the decision to lockdown followed the exact same error mode as Rosling's roadblock. Policy-makers got tunnel vision and focused on one thing - the pandemic - and ignored all other consequences including a mental health epidemic, mass unemployment/furlough, serious harm to children's development, and many other issues. More generally, I suggest that the typical failure mode is to over-react to any threat which is reported in the media (a combination of the availability heuristic making that threat more salient, and asymmetric justice meaning that politicians are more scared of being accused of inaction than they are of wrong action.) The failure mode can be summarised as: we must do something, this is something, therefore we must do it.

Pretty much everyone who has tried to calculate a cost-benefit analysis of lockdowns finds that they cost more years of life than they save. In developed countries, the typical estimate is 10x. (For example, Prof Philip Thomas of Bristol Uni estimating for the UK, or Dr Ari Joffe estimating for Canada.) For developing countries with poorer, younger populations, the ratio is worse. A multi-disciplinary South African team estimated that their lockdown would cost 29x more years of life than it saved. 

Those estimates all assume that lockdowns are highly effective at preventing Covid deaths, which was intuitive as of March 2020. However, reality is not always intuitive and the experience of places like Sweden, and those US states which didn't have winter lockdowns, suggests that the actual benefits of lockdowns are far smaller than thought. Studies like the Stanford University one find small-to-zero benefits of lockdowns vs less-restrictive measures.

For those of us who are more interested in physical truth than in the socially-constructed narrative, we need to start persuading the wider world that lockdowns do more harm than good. I hope you and the rest of our virtual monastery will join me in this effort.

Building a narrative around lockdowns being more harmful than beneficial is simplistic at best, and dangerously misleading at worst.  The lockdown is a powerful weapon, and for that reason, it should be used wisely (but it should be used when needed!). We seem to forget sometimes that some countries have indeed provided valuable examples on how to succeed at curbing the pandemics and the lockdowns have been one of the main tools they employed, but not the only one. 

 A very important point is that judging lockdowns in isolation is wrong.  The reason why some countries have managed to successfully stop the spreading is not because of the lockdowns alone, but because multiple interventions taking place simultaneously.

Take the example of Australia. The lockdown in Melbourne lasted longer than in most places, but in the meantime, the borders were closed for most people, except for those Australians coming back to the country that had to go through mandatory quarantines in specifically designated hotels.  Australia has also been very effective at contact tracing but that is something that works very well only if you have few cases (and what is the prerequisite to have few cases?). In Europe or in US is not that the lockdowns did not work (they work very well at doing what they have to do, stopping momentarily the spreading of the virus), it is simply that other measures were not taken simultaneously or they simply failed at implementing them. 

Lockdowns are more harmful than beneficial with the few exceptions of those countries like New Zealand that successfully kept the virus out. For any country where the virus is already endemic, the damage done by lockdowns was immense, and the benefits relatively limited. Remember that the counterfactual is not 'do nothing'. It's 'enact some more reasonable set of restrictions'. 

Prof Douglas Allen of SFU just did a really good takedown of bad arguments in favour of lockdown. In his most unrealistic extreme scenario intended to steelman the pro-lockdown case, he finds that lockdowns cost 3.6x more than their benefits, at the opposite end of the spectrum, they might have cost as much as 282x more than their benefits. (His figures for Canada, but the argument should generalise to any developed country.)

This image is a good example of how distorted pro-lockdown arguments are. It's taken from Neil Ferguson's Imperial College model used to argue for lockdown. Pro-lockdown cost-benefit analyses generally compare the blue line below (full lockdown) with the black line (do nothing) for an estimate of 120 lives saved per 100,000 people. It would be far more reasonable to compare that blue line to the brown line below (assumes case isolation and household quarantine but not lockdown measures) which immediately halves your assumed benefit for lockdown measures. Then remember that the Imperial College model is grossly overstated and assumes no public behaviour change in the absence of a lockdown. You get the picture. 

I really do encourage you to read the whole study:

The report you're linking to contains this:  

>Estimates of the IFR have continued to fall over the year. The latest meta-study by Ioannidis (March 2021) estimates the average global IFR at 0.15%.

That's completely off, and so obviously and indefensibly so that it discredits the entire thing, IMO. Maybe there are economic arguments that suggest that alternatives to lockdown could be better, but it would be irresponsible to update on that based on arguments made by a person who cites Ioannidis's IFR estimates favorably. Ioannidis is a crackpot when it comes to Covid. It's ironic that you write "This image is a good example of how distorted pro-lockdown arguments are."

I have looked into IFR estimates quite a lot when I was following Covid and I won a large forecasting tournament (and got 3rd in the year-long version):

Also, from the anti-lockdown side I've always wanted to know how to justify letting hospitals get so overwhelmed that people will die of appendicitis – basic health care collapses for at least 2 weeks. Do we really want that if it's avoidable? How would anyone feel as a doctor, nurse, caretaker, etc. if the government expects you to do triage under insane conditions when it's totally avoidable? The anti-lockdown side has to engage with that argument. If you say the IFR is low enough that hospitals wouldn't get overwhelmed without lockdowns, that's simply not true and you're engaging in wishful thinking or ideologically clouded thinking. I'm open to arguments that we should have a breakdown of civilization for 2+ weeks (and probably several times) if [edit] "more hidden" consequences are extremely catastrophic otherwise, but then one has to be honest about the costs of a no-lockdown policy.

Edit to add: It's a strawman that policymakers compare lockdown to "do nothing." And by now, even the people who initially got it wrong have understood that there are control systems, that many people will stop taking risks as they read about hospitals being overwhelmed. However, there's a 2-week lag from infections to the peak of hospital overwhelm and if the government isn't strict enough, you overshoot things really quickly. It can happen extremely fast. You cannot assume that people will always time their behavior the correct way to anticipate hospital overstrain that's 2 weeks ahead. That's what government is for. 

Your arguments about health care systems collapsing in the absence of lockdowns are highly emotive but not factual. The experience from Sweden, from Serbia, from the US states which didn't do second lockdowns, is that the health care systems didn't collapse. You are arguing from the unexamined assumption that no lockdown means failed health care system and your assumption is empirically, provably false. Please update your views accordingly. 

Now you're moving goalposts. Of course you can find places that didn't need lockdowns. I thought your position was that lockdowns were almost never/nowhere worth it. If your position is just "some locations didn't need lockdown (e.g., the ones where governments decided not to do it)" – that's extremely different. Whether lockdowns make sense is to be assessed case-by-case, because the virus (and new variants of concern) affected different locations differently.

In your other comment, you attribute a claim to me that I haven't made ("you have provided zero support for your own claim that lockdowns do more good than harm"). All I did was saying that I'm already skeptical since you were making the opposite claim with extremely poor and flawed arguments; I didn't say I confidently disagreed with your conclusion. Pointing out the favorable mention of Ioannnidis's  0.15% IFR estimate isn't "nitpicking of your evidence." It's damning that you rely on a source that does this – it's off by a factor of three to seven. After more than a year of the pandemic, you simply cannot be off about the IFR by this much without looking quite poorly. If someone (the person you were citing/recommending) writes an entire report on how bad lockdowns are but thinks the virus is at least three times less deadly than it actually is, this person seems incompetent and I cannot trust their reasoning enough to buy into the conclusion.

I will drop out of this discussion now.

I see a lot of nit-picking of my evidence, but you have provided zero support for your own claim that lockdowns do more good than harm. I challenge you to come up with a published cost-benefit analysis that proves the same. 

What would a good cost-benefit analysis include? There are a lot of harms caused by lockdowns. Some of them are difficult to quantify (eg my last point), but I think it's reasonable to demand a cost-benefit analysis takes into account at least three of the following six harms (which are far from an exhaustive list): 

  • Increased poverty is directly correlated to lower life expectancy, so we should measure the lost years of life from increased poverty. This is a very long-term effect, which will be doing harm for years to come.
  • Unemployment and financial problems are both 'scarring' (it takes a long time to dig yourself out) and both cause mental health effects. We should include the long-term mental health cost of increased unemployment and individual financial problems.
  • Where lockdowns include school closures, the long-term effect on children's development and socialisation is extreme. We should include estimates of the lifetime impact on children, which will include shorter life expectancy. (Yes, shorter educations are correlated to future life expectancy.)
  • Additional mental health effects directly attributable to the lockdown including elevated rates of depression, stress and anxiety. Proper acknowledgement that some of these effects (eg increased alcoholism) have long-term effects which should be accounted for.
  • Lockdown sucks for everyone, even if still employed, not in school, and not suffering a formal mental health condition. We should acknowledge that this is a widespread disutility and deserves to be considered.
  • Lockdowns have set a deeply disturbing precedent that governments can remove almost all civil liberties when they declare there's an emergency. This directly harms democracy and raises the risk of future loss of freedoms. (Yes this one is the hardest to quantify, but it's important and I would like to see more people acknowledge it.)

An ideal cost-benefit analysis would acknowledge that benefits from lockdowns in terms of lives saved are uncertain and include a range of estimates, but if you find one that properly considers at least 3 of the above 6 points, I'll accept it even if it has a point estimate for benefits. (Those using the initial Imperial College models should be at the top end of the range because the Imperial College figures are too high for all the reasons I've already said.) However, since you've said that "It's a strawman that policymakers compare lockdown to "do nothing."" then I do expect your superior cost benefit analysis will compare lockdown to more reasonable restrictions, rather than a do nothing option. 

So there it is. I challenge you: bring evidence or go home.


PS: If I seem to be beating the drum of long-term effects too hard, it's because I'm still angry at the UK government's (belated, poor-quality) excuse for a cost-benefit analysis which, among its many other failures, looked only at harm done over the next five years.

Thanks for the link. I have read the first part of the study. I tell you the things that make me suspicious about it:

  1. First, it starts with a very broad definition of lockdown: "is used to generically refer to state actions that imposed various forms of non-pharmaceutical interventions". Then, it proceeds to redefine the lockdowns of countries such as New Zealand as something else, "isolation". To me, the thing that comes to my mind when we talk about a lockdown is that you forbid people to leave their houses for an X number of days. Correct me here if I am mistaken, I am not an English native speaker.
  2. Convenient ignore that China has been one of those successful countries dealing with the Covid (not a single mention to it in the full document). I see that many times. New Zealand is used always as an example by people sceptical of lockdowns, for the reason that it comes with a built-in justification: it is an island.  Well, China is much more densely populated than the US and they also succeeded, and it is not an island. Would have been capable of replicating the success without enforcing lockdowns? Why is China not discussed more in general?
  3. Wrong use of counterfactuals. Those counterfactuals nicely generated do they really present realistic counterfactuals? Take for instance the blue line.  It suggests that by taking some social distancing measures the number simply go down in a matter of ~3 months and then everything is over. That's far from true.  As soon as there is a single case out of control (and this is what countries like NZ, Australia or China understood very well), the pandemic is not under control, you go back right to the beginning.  If you do apply strict lockdowns early you can aim to effectively eliminate the virus (but as I said before, you need to enforce mandatory quarantines and keep a very high level of testing and contact tracing). Is this counterfactual considered? 

serious harm to children's development

Why do you think the lockdowns caused serious harm to children's development?

Closure of schools. There's a mountain of evidence that taking kids out of school is harmful. It's not just the loss of education - although that doesn't help - but also the loss of socialisation. Less education is directly correlated with shorter life expectancy - a US study found that just that effect was enough to mean that closing schools would cost more years of life than it saved, with 98% probability. That's before adding in the burden from significantly higher rates of mental health problems in children who have been deprived of school. 

Close of schools is disproportionately harmful to those who already come from deprived backgrounds - think about the difference between a middle-class family where every child has their own iPad and the educated parents will help with homework, compared to a lower-class family which has one phone to share among everyone and the single-parent doesn't have time to help kids and also work. Then consider that closure of schools means loss of free school meal schemes - this caused chaos and serious hardship even in the UK and will have been worse in less developed countries.

Then there's the extreme cases: school is an escape for kids who live in homes where there's domestic violence. Teachers can also look for signs that kids need help or are suffering for abuse - if they're physically present. Closure of schools means that kids in abusive situations are trapped 24/7 with their abusers - whose own behaviour may become worse due to stress of unemployment or isolation. 

Closure of schools. There's a mountain of evidence that taking kids out of school is harmful. It's not just the loss of education - although that doesn't help - but also the loss of socialisation.

What if socialization is relative?

This explanation of the experience of Rationalist vs non-Rationalist thinking accurately describes a lot of my experiences in recent years. The following are the first examples that come to mind, of interactions with very smart people, good at thinking, who don't identify with the Rationalist community.

Something my wife last month: "Is this how you think about politics all the time? No wonder you're depressed."

Something I told a coworker two years ago: "Most people really, truly aren't consequentialists. They don't do things because they expect a certain outcome, they do them because that's what's customary in that situation in their community, full stop. That results in behavior that looks like they're implementing something like separate magisteria for each context."

Something I told a different coworker three years ago: "Most people don't actually know how to think. They do something that superficially looks like thinking, but isn't."

Something I told yet another coworker four years ago, "The client asked me X, which is the wrong question for what they're trying to accomplish, but it's his boss that made him ask it, and he's not socially allowed to challenge it, so I answered Y, which hopefully will trickle back in a way that gets the message across about what they actually need to ask, which is Z." Result: they came back and asked Z a few months later. Note: this is dangerously close to trying to "nudge the public" and I'd much prefer to have have to do things like that.

Something my wife last month: "Is this how you think about politics all the time? No wonder you're depressed."

I'm not quite sure that the "this" is in that sentence. You think about politics all the time how?

Yeah, I totally left that part out. I don't remember the specific situation, but it had to do with starting from a base assumption of factors like institutional inadequacy meaning I expect lots of seriously suboptimal decisions that lead to bad results that no one wanted, and public figures often being incompetent at their supposed jobs because they're picked by selection criteria force them to optimize for something way different from the supposed job requirements, and everyone just constantly talking past each other without even trying to really understand the other side (either due to ignorance, lack of interest, or various forms of group identity signaling).

For context, on an individual level, she's vastly better than me at intuiting what other people are thinking and how they're likely to act. And she does understand the social psychology of groups of people very well. She just doesn't instinctively consider politics in terms of the dynamics and evolution of systems.

Also note: after years of grappling with ideas like that, I've gotten much closer to not always being depressed by this kind of thing, or seeing it as an inescapable trap (and trying, whenever possible, to focus on the side of "Wow, look what we managed to accomplish anyway!"). But it definitely had that effect on me for a long time.

I'd much prefer to [not] have to do things like that.

The Nudgerism section seems to be mushing together various psychology-related things which don't have much to do with nudging.

Things like downplaying risks in order to prevent panic are at most very loosely related to nudging, and at least as ancient as the practice of placing objects at eye-level. Seems like an over-extension of focusing on "morale" and other Leaders of Men style attributes.

The main overlaps between the book Nudge and the awful The Cognitive Bias That Makes Us Panic About Coronavirus Bloomberg article are 1) they were both written by Cass Sunstein and 2) the one intervention that's explicitly recommended in the Bloomberg article is publicizing accurate information about coronavirus risk probabilities.

One of the main themes of the nudge movement is that human behavior is an empirical field that can be studied, and one of the main flaws of the thing being called "nudgerism" is making up ungrounded (and often inaccurate) stories about how people will behave (such as what things will induce a "false sense of security"). These stories often are made by people without relevant expertise who don't even seem to be trying very hard to make accurate predictions.

The British government has a Behavioural Insights Team which is colloquially known as the Nudge Unit; I'd guess that they didn't have much to do with the screwups that are being called "nudgerism."

I had nudging cached in my memory as, more or less, a UX movement.

Want to increase charity donation at your company? Make it opt-out, rather than opt-in. Want to increase completion rates of your survey? Make it shorter. 

And so forth. 

So I was surprised by Jacob Falkovich claiming that nudgerism caused the elaborate psychological theorising used to inform covid policy. Many such policies mostly seemed to be about oddly specific, second-order claims. Like, in the case of expected resistance to challenge trials, or vaccine hesitancy. Those arguments venture more heavily into psychoanalysing people; rather than appealing to simple behavioural economics and basic UX. 

(My cached memory of the nudge movement might be too narrow, though)

Want to increase completion rates of your survey? Make it shorter. 

Why not send out 'surveys' to a large audience which have a random subset of questions included, randomized separately for each person?

During COVID the UK government has been heavily advised by the SAGE committee (an emergency committee of scientists), including a subcommittee of behavioural scientists who advised on what the reaction to measures like lockdowns might be. I don't know how reliable behavioural science is at the moment (with the replication crisis) but this seemed like a reasonable move - being guided by them rather than politicians' own hunches.

Hmm, I do think I honestly believe that behavioral scientists might be worse than the average politician at predicting public response. Like, I am not totally confident, but I think I would take a 50% bet. So this strikes me as overall mildly bad (though not very bad, I don't expect either of these two groups to be very good at doing this).

Habryka, is the reasoning that politicians have a real incentive to accurately predict public response -- because it entirely determines whether they remain in power -- whereas behavioral scientists have a much weaker incentive, compared to the dominant incentive of publishing significant results? 

Why not do polls?

They do, but the committee was trying to predict future reactions.

My jaw dropped when I got to the screenshot of the tweets, and it took me about a minute and a few double checks before I was able to put it back in place.

The monastery section made me think of Marcus Aurelius, he is considered one of histories greatest leaders, and from the little I know about him, it seems justified. He also seems to be an example of someone who was both a "monk" and was on the throne. I'm not sure what to take away from this, but at least it suggests that it's not impossible? (of course, there are quite a few differences between our world and Aurelius' world).

What other examples are there of the rationalist community influencing things outside of it? Seems worthy of a wiki page / tag

A good friend of mine, who I know is making absurd returns on altcoins, is still bummed about that losing his entire investment on that options trade. Losing 100% is pretty demoralizing I guess.