Thursday, April 30, 2020

Pandemic epistemology: discovery, feedback, ideological pomposity and banana peels.

I was going to forbear from posting on the Covid-19 pandemic, but this post by Arnold Kling prompted some more general observations about social dynamics.

He refers to a podcast by biologists Heather Heying and Bret Weinstein where they, in his words: 
... cite instances in which odd corners of the Internet are outperforming mainstream science and mainstream journalism. This comes through most in the last few minutes of [t]he podcast.
Those most against federalism or free speech (and there tends to be overlap in antipathy to each) tend to systematically under-rate the importance of discovery processes. This includes under-rating the dispersed nature of effective discovery processes. 

People are not all one thing. Someone can be batshit crazy in one area of life (a colloquialism with a bit more bite nowadays) and incredibly perceptive in another.  Sir Isaac Newton was deeply interested in alchemy and the weirder end of biblical exegesis. This does not stop him being a source of amazing breakthroughs in physics and mathematics (and, for that matter, coin production). 

Not only does one not preclude the other--being so wrong about X does not preclude being highly perceptive about Y--being willing to consider wild and wacky possibilities may actually help one be brilliantly creative, provided there is the requisite attention to evidence and careful reasoning (or whatever effectiveness constraints operate in the relevant domain). 

The discovery value of gentiles
So, those alienated from the mainstream whatever, for good or bad reasons, may well be more inclined to pick up things that the mainstream is blind to or weak on. In his excellent Nobel memorial lecture (seriously, if you haven't watched it, you really, really should) on how to do social science, Paul Krugman talks about the importance of "talking to the gentiles". Yes. (See also his essay here.)

Which is why the current penchant for identifying the gentiles, the "evil" infidels, and driving them out of public spaces is so dangerous. Our global civilisation is in utterly uncharted waters for our species and the last thing we need to be doing is seriously damaging our discovery processes, which is precisely what this penchant for cancelling the heretics does. Such burn-the-witch hunts are patently prestige-and-dominance plays but they are profoundly dangerous and destructive prestige-and-dominance plays. 

As an aside, I very highly recommend the online lectures available via the Center for Academic Research and Training in Anthropogeny (CARTA) at the University of California. Top scholars in the various fields germane to the study of our origins as a species lecturing to other scholars as cross-disciplinary exercises (so easily followable by a lay audience, because very discipline is a lay audience to other scholarly disciplines). It is fascinating, and profoundly informative. 

Problems with models
Arnold Kling also observes that:
... I bristle when someone says that based on a computer simulation, a certain policy for dealing with the virus can save X lives. I presume that there are some key causal assumptions that produce the results, and I want to know what those assumptions are and how they relate to what we know and don’t know about the virus.
The most widely-used models don’t differentiate the population by age. Blinded by these models, policy makers focus excessively on maintaining hospital capacity and inadequately on protecting the elderly.
We tend to selectively over-rate models. That their assumptions are often opaque helps with this process and can make their use rather too close to using maths and computing to replicate what previous ages did with sheep entrails.

Models in themselves are very weak discovery processes, as they discover the implications of the assumptions of the model, not reality. They have their uses, in working out what our assumptions imply and making our thinking more systematic. Alas, it is very easy to see them as doing something in themselves, without the testing against reality. Genuine discovery power always comes from exploring reality, which models do when they are tested against reality. If used for that, models can be profoundly useful, by forcing us to be consistent and systematic in our thinking. (Krugman discusses the importance of models for clear thinking in this essay.)

[A nice discussion of the performance of the Imperial College and University of Washington models is here.]

We know a lot more about the Covid-19 virus in late April than we did in late January. It is perfectly reasonable  to question whether decisions made early in the pandemic are still valid given what we now know. Particularly, how well the models used in those decisions have stood the test of reality. Unfortunately, there are all sorts of status considerations now built into those decisions, which may well be inhibiting effective use of the expanded knowledge. 

Having good feedback is vital to systems functioning properly. It is likely that much of the chronic health problems advanced societies are increasingly prone to are due to people developing  damaged or suppressed feedback regarding what we eat and drink. Aided and abetted by damaged, suppressed or pathological feedback systems in the provision of health and nutritional information. 

Feedback and incentives are deeply intertwined in human systems. Here is a question to think about: do the revenues of Western health departments go up if we get sicker or healthier? What incentives does that create? Then ask yourself if the answer to those questions, and considering what incentives health departments face, what the actual feedback systems they operate within are, affects how we might think about the response of those same health departments to the current pandemic. Such as what we did, and did not, have stockpiles of. Remembering that in most Western countries, health departments have (much) bigger budgets than defence departments.

Addiction to conflict narratives
The mainstream media sees itself as our central, and indispensable, information system. How well does it perform at that, really? Is mainstream media not somewhat addicted to conflict narratives, as they provide easy and "exciting" framings to present "news"? What does that do to the the signal-to-noise ratio in mainstream news?

Consider two doctors who own and run various clinics in California talking about their experience of* [now available here] the pandemic and what they, as relevant experts, glean from the available data and talking to their colleagues. This is a discovery briefing. The journalists the doctors are briefing, however, are not in discovery mode, as is revealed by the tone and content of their questions. They are in identifying-conflict mode. Discovery is messy, identifying conflict simplifies and excites. They don't want messy discovery, they want simple, exciting conflict.

[ADDENDA: *YouTube took down the video of two doctors briefing journalists and reporters about their clinical experience of Covid-19. There is an obvious irony for such a link in an essay on feedback and discovery.]

Moreover, it is a very easy shift to go from being addicted to conflict narratives to moralising about (and then within) those conflict narratives. It is very easy to turn conflict narratives into goodies-versus-baddies stories, with the journalists and reporters both identifying "the goodies", and identifying with and as "goodies". They then become part of the conflict narratives themselves, and the signal-to-noise ratio gets way worse.

There is a reason why public trust in the media has become so disastrously varied. The Donald's approval rates as US President were rather poor and are now consistently mediocre, in accordance with my view that he is demonstrably an electorally weak candidate. (He seems rather obviously personally high in (dis)agreeability; a wildly unusual characteristic for a senior elected political figure, though rather more common among those highly effective in other spheres of life.) And a disagreeable President makes an unusually potent figure in conflict narratives. Even more so in moralised conflict narratives. The noise-to-signal ratio in mainstream media coverage of The Donald's Presidency has rarely been less than toxically high.

Shifting to US public opinion of the media, the standing of the media as a source of news is relatively good among Democrat voters, poor among independent voters and abysmal among Republican voters. In terms of operation as a shared feedback system, this is a disastrous pattern. Not only is the mainstream media not trusted by large parts of the US populace, but the mainstream media are so often patently participants in their adopted goodies-versus-baddies conflict narratives, which actively encourages them to be generators of more noise and less signal. To be actively hostile to the processes of discovery -- seeking to block information which undermines the goodie-v-baddie narrative they have inserted themselves into while elevating information that feeds it -- against more careful considerations of significance and accuracy. 

Ben Goldacre's book Bad Science provides depressing chapter and verse on how very bad the media can be in reporting science, just from problems of not understanding science and statistics, limitations in human cognitive patterns, and the media's addiction to conflict and 'ghee whiz' narratives. Indeed, the media tend to be particularly bad on nutrition as that combines (1) obvious public interest, (2) deeply vested corporate and other interests, (3) the benefits of publicity for scam artists, along with all the above problems that reporting on science already has.  

The media's self-insertion into goodie-versus-baddie conflict narratives, and already poor performance in science reporting, is not a good pattern in general, and particularly not in a heavily science relevant matter such as a global pandemic. 

Discovery and feedback
Discovery and feedback systems matter. Both in response to short term events and in long term prospects for our civilisation and our species. Does this help or hinder discovery processes?, help or hinder effective feedback systems?, are good questions. And if you are not even asking the questions, that is a problem. Indeed, there is an excellent likelihood you are part of the problem.

If you are asking and answering the questions in terms of a goodies-versus-baddies narrative, you are probably not really asking the questions and are very likely to be part of the problem. As Bret Weinstein observes (at 23minutes), utopianism (which tends to be a goodie-versus-baddie narrative set to a maximum) is perhaps the most disastrous idea Homo sapiens have ever had precisely because it is so intrinsically hostile to discovery and feedback. This is a result of absolutely prioritising a single value (for, as he says, that then creates "incredibly large costs for every other value") and because they "tend to imagine they know what the future state should look like", short-circuiting (indeed, typically blocking) open discovery processes. This combination is compatible with ruthless selection for what works for seizing and monopolising power and disastrous selection in who gains power and how they use it. As a series of tyrannies, and millions of corpses, demonstrate.

The Hurley model of humour says that humour comes out of our cognitive error identification mechanisms. This is why ideologues are so often humourless--they are unable to accept the possibility of error. (Cue that great definition of a fanatic--a person who can't change his mind and won't change the subject.) Extremists have a crippled epistemology that blocks discovery and feedback.

Ideologies tend to be pompous, they inflate themselves beyond the possibility of error, particularly errors of significance. They are the cognitive equivalent of the pompous fat man unable to see the possibility of the banana peel. The slipping-on-a-banana peel joke works so much better if the pompous man is fat because he is less likely to see the banana peel, his pomposity takes up more space, and he is more likely to bounce (boing, boing, boing ...).

Admit it, you laughed.

It was better if he was a man, because when the trope was established, male pomposity had further to fall. And if your reaction is to point-and-shriek "fat shaming!" un-ironically, you just outed yourself as a humourless ideologue.

The more morally grand one's vision of what one is about, the more entitled one can feel to suppress the views of those who disagree. (Herbert Marcuse's iconic essay on repressive tolerance rests on belief that some group reliably has such knowledge.) But such suppression automatically involve suppressing any discovery that might thereby be revealed. One's sense of moral conviction, precisely because it is so emotionally powerful and because moral concerns have inherent trumping value over other concerns, can be a profound barrier to discovery and to effective feedback.