mathbabe

Although it’s wonderful to hear Jon’s sane and sensible voice and views, I just wish commentators, media and the public would stop describing prediction markets as accessing “the wisdom of the crowd” which, as Francis Galton originally observed, relies on regression to the mean of a random sampling. That regression fails if any member is aware and influenced by others’ bets, or if some bets have greater weight than others. When these requirements fail, the “wisdom” is not only lost, it can turn into its very opposite, a market bubble.

Two further consequences of this muddying of Galton’s insight: prediction markets discredit the true wisdom of crowds that perhaps alone can offer a bottom-up solution to the freedom of speech dilemma; and regulating or banning prediction markets from media provides a false sense of security by ignoring the underlying market failure of top-down, “expertise” in news reporting.

Teaching the sociology of false beliefs over the past few years has taught me that the dilemma of free speech — the Scylla of allowing lies or the Charybdis of censorship — can only be solved from the bottom-up, not from media sources that are captured by their audience. Unfortunately, accurate information is tied to that market so information, even when true, is tagged with socio-economic status. NYTimes readers are never persuaded by Fox News perspectives even when they are accurate, and Fox News watchers are never persuaded by NYTimes perspectives. On the contrary, NYTimes reader finds Fox information or spin as evidence of its depravity and the Fox watcher finds the Times’ spin equally depraved and misguided. It’s all a game of socio-economic confirmation/myside bias. The media’s problem is not its relation to accuracy, it’s a market failure of its audience capture — the consumer drives the market, even when information itself is the commodity.

Untampered regression to the mean and crowd-sourcing are the only means out of that market failure. I mentioned this to Jon some years ago, but I think he assumed I was suggesting that there needed to be some means of controlling or regulating lies. Far from it. The sources of influence on the public can only be understood by looking at the motivated reasoning differences between the educated elite respected class — the professional class members that reap the most from their society, economy and governance and are most invested in its society’s success — and the non elite class, the class disrespected by the elite and by the prestige values established in the mainstream of culture, the class seeking mutual approval by rejecting elite spin and embracing complete distrust of elite information sources. Conspiracy theories give them street cred as being smarter and more insightful than the elites who believe mainstream “propaganda”. The audience of conspiracy theories often view voters as chumps. Regulating media, suing them in court, implies top-down censorship, but who fact-checks the fact checker? And when mainstream media is duly chastised, its audience will turn to social media.

I want to say, the free speech dilemma will never be solved through improving media, Jon. News media is a market failure. A solution can only begin with understanding the zero-sum game of class difference: the sociology of false beliefs, whether those beliefs be leftist nostrums or rightwing conspiracies, the self-interest of the virtue-signaling class at the economic top or the solidarity among the respect-signaling class at the cultural bottom. Above all, it begins not through concourse with one’s ideological peers but socializing extensively with the cultural, uneducated bottom.

I wrote more in detail on my blog deperplex (dot) blogspot (dot) com in posts on “the obvious fact showing propaganda persuades no one”, “the Gates-Musk paradox: conspiracy theories are not about what the theorists think they are”, “the sociology of false beliefs”, “The Fool’s Errand Attachment: a cognitive bias”, “corporations are followers, not drivers”.

Btw — in the 90’s when I was getting my PhD in linguistics, neural networks were the Big New Thing and a serious challenge to the Chomsky computational model of mind that buried the behaviorist-empiricist-positivist of the first half of the 20th century after Wittgenstein and Quine. I like to call neural network learning the zombie revival of behaviorism since the Shannon model of learning is essentially empiricist-behavioral. In May, I gave a presentation at the International Linguistics Association on “Why neural networks succeeded where generativism failed: The syntactic and semantic limits of neural network behaviorism (Yoneda lemma limit)”. It’s not political at all — it’s about a math problem in learnability: how a neural network learns NOT to extrapolate from its training set, the problem called “the poverty of the stimulus”, a problem with important implications for empiricism vs rationalism and tech vs science in the context of Turing vs Shannon, — but there are implications for Category Theory that you might find interesting. It was based on two blog posts, mostly from “complexity and AI” but also “art, craft and game-theoretic cognition in machine learning” which predicted some of the failures of LLMs prior to the appearance of ChatGPT.

LikeLike

Just to be clear: it’s not Galton’s “wisdom of the crowd” (regression to the mean, a robust and established result of probability) that needs debunking, but the widespread public mistake that prediction markets tap into that “wisdom of the crowd”.

The important point, stated more briefly(?): in a class-distinct society, we all live in an information-confirmation bubble based on social location; the professional class in its bubble assumes improving media will improve civil discourse by reforming media in the professional class’ own image. Could be they have better information than the Right. However, in our zero-sum polarized society, members have already chosen what they want to believe based on their social location, not based on truth. (Just look at what the Fox watcher thinks of mainstream media.) So “improving” media will merely drive an audience to other sources, as we’ve already seen in social media with Joe Rogan, Carlson, Fuentes, Yarvin … too long to list, which says something important about the market for it.

A solution must come from the distributed bottom, not censorship from the top or from the professional class. I don’t know whether Galton’s original observation of the wisdom of the crowd plays a role in the solution, if there is one, but surely we should be looking for such bottom-up solutions. Observers smarter than me should be able to come up with possible solutions to try, if they would just focus on it.

OTOH, go to Cairo Egypt and you will find no political polarization. Everyone knows their government is a military dictatorship that serves the military and the military owns everything. The consequence: ordinary Egyptians take care of each other, as if they were one big family. When times are good, they tolerate Sisi. When inflation rises, everyone agrees Sisi is an idiot. I see something of this among Russians. They’re not happy, but they’re not hating each other the way the secular middle-class Istanbula hate the religious conservatives. The pretense of participatory democracy is not an unalloyed good. And if it’s a mere pretense, then it’s surely a net bad.

Full disclosure of hypocrisy: I’m a monthly donor to The Intercept because if we lose investigative journalism, then nothing remains but propaganda and clickbait. But I support investigative journalism only on the left-liberal side. I recognize my bias but have to accept it: it’s me.

LikeLike