Interesting Links for 11-01-2022

[andrewducker]

What are the biggest non-diet causes of diabetes, cancer, heart disease, etc?

(tags:society sleep stress disease weight diabetes )

Leaked email shows 100 No 10 staff were invited to a party during lockdown restrictions

(tags:UK conservatives Pandemic law OhForFucksSake )

Repeat after me: Covid doesn't spread on surfaces

(tags:Pandemic cleanliness )

Adversarial Wordle

(tags:games language )

Serotonin selectively influences moral judgment and behavior through effects on harm aversion

(tags:morality serotonin psychology )

Male elite chess players are quicker to quit when dominated by a female player

(tags:men behaviour psychology games sexism )

"Why I went from being an actor to a clowndoctor"

(tags:acting children comedy )

In 1st, US surgeons transplant pig heart into human patient

(tags:pigs organs surgery )

UK health officials misinterpreted US guidance on Covid self-isolation period

(tags:pandemic )

Dolphin females have working clitoris

(tags:sex dolphins clitoris )

Beautiful trees

(tags:trees beautiful )

Impressive shepherding

(tags:dogs birds impressive training )

Comments

[cmcmck]

Nice to have something in common with female dolphins! :o)

suuuper low effort criticism

[hellofriendsiminthedark]

Is the author of "Covid doesn't spread on surfaces" credible?

His other articles all seem to pretty much rely on nitpicking a lot of pedantic points, appealing to status quos, and neglecting to source major claims: "climate change is overblown because the projections are based on pessimistic models"; "a book about deradicalization techniques fails because it takes a clear stance against conspiracy theories"; "not everybody fits the eligibility requirements to get covid boosted and this is somehow a point in and of itself?"; "implicit assumption that morbid obesity is caused by brains that cannot turn off hunger signals, without acknowledgement of cany of the other very legitimate causes"; "covid testing and contact tracing aren't 100% perfect so therefore we should 'return to as normal a life as possible' already and give up on all that stuff."

One thing that stood out to me was that he argues that there's a definitive meaningful difference between how concrete sexual orientation is versus how concrete gender is. He actively endorses the idea that autism or personality disorders are legitimate bases for trans people to be denied the right to informed consent and self-identity.

As for covid not spreading on surfaces? That's true insofar as putting your hand on an object that somebody coughed on isn't going to give you covid. And yet that says nothing about what happens when you subsequently rub your eyes with that same hand. He himself admits that most confirmed covid cases aren't fully traced. He also marvels at the novelty of seeing grown men wash their hands for once early in the pandemic, but then also bases his claims of "sanitization doesn't matter" by presuming that everybody was washing their hands responsibly before the pandemic anyways. While his ultimate points that particles in the air are much more relevant than particles on surfaces, and that ventilation is a much more relevant point of action than surface disinfection, there's something really insidious about only knowing how to make that case by traveling down the most inflammatory, reactionary, click bait-y, technically accurate path.

Surely good faith science journalism strives to inform, as opposed to obfuscate. My impression after reading about 10 of this author's articles is that he consistently starts with some wild claim, then leaps across several more spurious claims and tenuous assumptions in order to ultimately land on something even resembling a point which is not even terribly interesting in the end given that half his points are "I disagree with this word choice."

Re: suuuper low effort criticism

[andrewducker]

That's fair. I've seen a fair amount of his science journalism, and have generally considered him pretty reasonable. And this happens to be something that I mostly agree with him on, so I thought it was worth sharing. I'll keep an eye out for something better.

Re: suuuper low effort criticism

[hellofriendsiminthedark]

I had some food and decided to build a more precise critique for my own neurosis' sake--I hope that's alright. The tl;dr is that I don't disagree with him either on his grand takeaways, but the way he builds up his argument doesn't seem to actually lead to the points he says he's making.

I agree with his point that we should be "improving ventilation, keeping windows open and meeting outdoors where possible."

But the way that's worded in context is:

Instead, what we should be doing is improving ventilation, keeping windows open and meeting outdoors where possible.

The entire rest of the article is only relevant to this point when "sanitizing surfaces" and "improving ventilation" are positioned as mutually exclusive efforts--if it weren't a matter of one-or-the-other action, then the notion of an "instead" wouldn't hold.

The tagline of the article is "Will we ever move beyond outdated advice?" but the case is not adequately made that the advice is outdated:

A Lancet Infectious Diseases paper way back in August 2020 pointed out that the claims about how long Covid lasted on surfaces was overstated, based on studies that used unrealistic concentrations of viral particles.

As far as I know, there have been no solidly confirmed cases of fomite transmission (although this somewhat gross “snot-oral transmission” incident is a possible). No doubt there have actually been many tens of thousands — most cases are never fully traced — but they are rare.

The reader is expected to roll with "this is outdated advice" based on an article which only reaches "overstated" but not "outdated," a personal speculation, and the admission that fomite transmission may be possible and probably has happened. For somebody who purports in other articles to care strongly about the precise meanings of words (including "conversion therapy"), this seems like intellectual negligence.

People can have unfortunate takes on transness and still know what they're talking about when it comes to epidemiology; my intent is to point to the author's own standards for integrity purposes, not character purposes.

The understanding that "outdated" is being used as a rhetorical inflation of "overstated" changes the nature of the subsequent points being made; there's a meaningful difference between a call to action for governments to stop reiterating outdated information versus to stop reiterating information meant to mitigate overstated risks.

And then there's this paragraph, towards the very end, which seems to make unfounded assumptions about the purpose of public health initiatives:

This isn’t to say that you shouldn’t clean things or you shouldn’t wash your hands. You should! The prevention of faecal-oral transmission of pathogens, through hand-washing and increased hygiene, is one of the great reasons why waterborne illnesses are so reduced in the Western world (along with improved sewerage). But you presumably did that before the pandemic, and you didn’t feel the need to sanitise your hands every time you went into Uniqlo or disinfect every surface in your house twice a day.

I have a hard time understanding what "you should, but you presumably already did" is supposed to demonstrate. For example, you also should cough into your arm instead of your palm, but presumably you already did that, and yet that's still an important piece of public health advice for government organizations to communicate. It seems to me that "this is a mechanism to reduce pathogen transmission and illnesses" and "you didn't feel the need to do this before the pandemic" doesn't naturally lead to the conclusion of "therefore it needs to stop being said."

There's also the fact that the cited advice is not "disinfect every surface in your house twice a day," but rather "frequently touched surfaces should be wiped down twice a day." And pandemic or otherwise, that doesn't seem that unreasonable of a suggestion, given the aforementioned utility of regularly sanitizing things for non-covid reasons.

I agree with the premise that fixating on overstated risk factors as opposed to more realistic ones doesn't do much to aid in pandemic-curbing efforts. I agree that in the grand scheme of things, the current trend of constant deep cleaning only produces a negligible difference in outcomes at best. I agree that improving ventilation and promoting cultural changes in where we gather are extremely important goals backed by current science about likely modes of transmission. But all those valid points are implied at best, plus heavily obfuscated by clever rhetoric and disingenuous arguments. There are reasonable takeaways to be had, but the journalistic integrity seems to be lacking.

Re: suuuper low effort criticism

[andrewducker]

Thank you, I appreciate the hard work there!

Re: suuuper low effort criticism

[cmcmck]

I'm trans myself and worked with young people with autism who were also trans (interestingly always trans women) and clearly well aware of the fact!

Another 'expert' who doesn't know a lot!

Re: suuuper low effort criticism

[andrewducker]

I hadn't actually spotted the trans stuff. Which I should make it clear I very-much don't agree with.

Re: suuuper low effort criticism

[cmcmck]

I'd never have expected anything else! :o)

Re: suuuper low effort criticism

[stormclouds]

He also claims that in a study that followed 64 people with borderline personality disorder for 27 years, 92% of them no longer met the diagnostic criteria. But as 10.3% of the 64 had died by suicide I don't understand his maths.

I have BPD and I'm autistic and I'm non-binary. And I'm not liking this Tom guy.

[simont]

I was curious about the "Adversarial Wordle" article because I could think of two different things the title might mean.

What they've done is one of the things I had in mind: change your mind as the game progresses about what word you're considering to be the answer, so as to prolong the game as much as possible without ever invalidating any response you've already given. I once wondered about making my puzzle-collection game Guess (an implementation of standard Mastermind) do the same thing, not in order to be "adversarial", but because I thought it might make the game more satisfying to at least some players – I find it perversely annoying when I've carefully constructed a maximally informative guess and come up with a plan for each branch it might send me down, and that guess turns out to be 100% right and end the game completely, and I never get to use any of those painstakingly worked out followup strategies!

(But I never got round to it. Guess does the totally obvious thing, of choosing a secret at the start of the game and never varying it.)

The other thing they might have meant is: vary the probability distribution of the choice of secret word, so as to maximise the expected number of moves taken by even the best strategies. This too is a concept that would apply to Mastermind, and it always struck me as odd that I'd never seen any maths articles focusing on that aspect of the game. There's been a fair amount of research into optimal Mastermind strategies for the guesser, based on the assumption that the setter will have chosen equiprobably from all options. But the setter has no requirement to do that: they could perfectly well choose with a skewed probability distribution, if they thought it would be a good idea. And if you're taking turns being the setter, and adding up scores over multiple rounds, then it is in your interest as the setter to try to pick patterns that the guesser will take more moves over. So the question becomes: what distribution of choices by the setter maximises the number of moves taken by even the best guesser strategies?

In Mastermind, this idea is limited quite a lot by symmetry. The whole game is symmetric in permuting the peg positions and/or the colours. So there's no real advantage in choosing different elements of the same symmetry orbit with different probabilities; that would help against a guesser you knew to be colour- or position-biased (say, always starts by guessing RRYY), but a guesser wise to that trick can render it pointless by applying a random symmetry to their whole strategy at the start of each round. So the only interesting variations the setter can do would be to vary the relative probabilities of the symmetry orbits themselves, e.g. bias towards or away from 'three of one colour' or 'two and two' or that kind of thing. And there may not be all that much you can do with that, since there are only five symmetry orbits in any case! But even so, I always wanted to see someone actually do the analysis and decide whether there was a marginal advantage to the setter in choosing from those orbits in a skewed fashion.

But in Wordle, there's no symmetry at all because of the highly non-orthogonal English word list, so this kind of skewed setter behaviour has a lot more possibilities to choose from!

[ingreatwaters]

Having tried the wordle thing, I'm not quite sure which side is adversarial, because the player seems to have quite a lot of power over the word - you don't quite force it not to use the letters you start with (when I started with STEAL the E did end up in the final answer), but you do strongly discourage it, so starting with something like QUICK/BROWN/JUMPS clears away a lot of the less common letters.

[andrewducker]

"The other thing they might have meant is: vary the probability distribution of the choice of secret word, so as to maximise the expected number of moves taken by even the best strategies"

Could you explain what that might look like? I thought that Adversarial Wordle already maximised the expected number of moves. How would this extend it more than that?

[simont]

I didn't say it would extend it by more. This second suggestion of mine is playing within the rules to a greater extent than AW is, so you'd expect it to be less effective at increasing the number of moves required.

(At least, unless ingreatwaters's counterstrategy turns out to be extremely effective.)

But as an example, suppose we were playing Wordle using a dictionary consisting of just the words AAAAB, AAAAC, AAAAD, AAAAE and ABCDE. Clearly, as the guesser, I minimise the maximum number of moves by guessing ABCDE first, because if that isn't the right answer itself, it will always let me identify the real right answer on the second move. Whereas if I guessed one of the others, say AAAAB, then I wouldn't get enough information back from guess #1 to tell the difference between AAAAC, AAAAD and AAAAE.

So, if the setter picks their word equiprobably from the five available ones, then I win in one move 1/5 of the time and in two moves 4/5 of the time, for an average of 1.8 moves to win.

But if the setter wants to make my average as bad as possible, they should never pick ABCDE! Then my strategy will hit its worst case of 2 moves 100% of the time.

[andrewducker]

Ah, gotcha. So picking a word, but one that's going to be harder to guess (possibly based on the likely choices of the player). Thank you, that makes sense.