I disagree.

You can approximate OKC as a two-person game: Weird = Honest, Polished=Dishonest. U(WW)=+3/+3, U(WP)=0/+4, U(PP)=0/0, then you have the usual Prisoner-Dilemma payoff (motivation: being honest will generate more long-term utility).

This is a bad approximation, as OkCupid is a multi-player-game, so it's more complicated than the classical 2-player Prisoner's Dilemma. That's where the tragedy of the commons comes in. In an environment where nearly everyone plays defect-bot, a lot of utility is destroyed. But, tit-for-tat players have an advantage if they meet another TFT player.

That's how I interpreted the use of concepts in the article, did you understand it differently?

Yes, I completely agree with the weaker formulation "irreducible using only THESE means", like e.g. Polynomials, MPTs, First-Order Logic etc.

I think the answer to the question "are there irreducibly complex statistical models?" is yes.

I agree that there are some sources of irreducible complexity, like 'truely random' events.

To me, the field of cognition does not pattern-match to 'irrecducibly complex', but more to 'We don't have good models. Yet, growth mindset'. So, unless you have some patterns where you can prove that they are irrreducible, I will stick with my priors I guess. The example you gave me,

For a very simple example, if you're trying to fit a continuous curve based on a finite number of data points, you can make the problem arbitrarily hard with functions that are nowhere differentiable.

falls squarely in the 'our models are bad'-category, e.g. the Weierstrass function can be stated pretty compactly with analyitic formulas.

But also, of course I can't prove the non-existence of such irreducible, important processes in the brain.

and my answer there is more like "I don't know, but I could believe so."

Fair enough.

These are heuristic descriptions; these essays don’t make explicit how to test whether a model is interpretable or not. I think it probably has something to do with model size; is the model reducible to one with fewer parameters, or not?

If you use e.g. the Akaike Information Criterion for model evaluation, you get around the size problem in theory. Model size is then something you score explicitely.

Personally, I still have intuitive problems with this approach though: many phenomenological theories in physics are easier to interpret than Quantum Mechanics, and seem to be intuitively less complex, but are more complex in a formal sense (and thus get a worse AIC score, even if they predict the same thing).

How much of human thought and behavior is “irreducible” in this way, resembling the huge black-box models of contemporary machine learning? Plausibly a lot

'Irreducible' is a pretty strong stance. I agree that many things will be hard for humans to describe ein a way that other humans find satisfying. But do you think that an epistemically rational entity with unlimited computational power (something like a Solomonov Inductor) would be unable to do that?

I am highly suspicious of any explanation that includes the claim "then every understood this, and so they behaved differently"

I agree. Can you point to where the author uses this in his model?

(I have no strong opinion on the model, I mostly pattern-matched when I read the call for more models)

I like your models. Do you think you can make them into quantitative ones?

Here are some caveats:

Model 1: Escalating Asks and Rewards

• I have experienced several communites with a 'not me'-mentality, where people are on bord for the fun, but do not want any work. This is strong evidence against universality of Model 1.

• On the other hand, I have experinced communities where this escalation happened and people told stories of how important their work was. So I definitely think this happens, but I can't think of a good predictor for whon or when model 1 applies.

Model 2: Imperfect Containers

Most humans seek some form of novelty; most humans desire growth.

This smells like the typical mind/sex-life fallcy. People in my filter bubble? Yes. My (grand)parents? (Much) less so.

I think this assumption (and thus Model 2) is only valid for a small subset of humans: Ambition and novelty-seeking should be a good predictor for whom Model 2 applies to, age and conservativeness should correlate negatively with model accuracy for these subgroups.

Here's an informal model I saw a while ago, with some similarities to Model 2:

"Subcultures were the main creative cultural force from roughly 1975 to 2000, when they stopped working. Why?

One reason—among several—is that as soon as subcultures start getting really interesting, they get invaded by muggles, who ruin them. Subcultures have a predictable lifecycle, in which popularity causes death. Eventually—around 2000—everyone understood this, and gave up hoping some subculture could somehow escape this dynamic."

https://meaningness.com/geeks-mops-sociopaths

I think this is strongly connected to the Typical Mind Fallacy.

I did a quick inventory of distortions that I recognize often [I live in a leftist techy-academic bubble full of socially and sexually permissive people].

Other people's properties that I overestimate because of my bubble:

• Intelligence

• Comptence (Imposter-Syndrome)

• Knowledge (expecting short inferential distances)

• Physical fitness

• Available amount of leisure

• Susceptibility to arguments and evidence

Stuff that is (relatively) easy for people in my bubble, but seems to be hard outside:

• Reading a novel

• Looking at the water

• Discussing sexuality & relationship styles

• Casual non-sexual touching (hugging, cuddling etc.)

And these are just properties where I deviate pretty strongly from mainstream society.

The rule follows: for things that are private and rarely discussed, there may be a good deal of unacknowledged diversity.

I agree. The Hamming-style question I now ask myself is:

'Which unacknowledged diversity is creating the most problems for me in social interactions?'