[s1artibartfast]

Im not sure what your criticism is directed at.

In the real world, we can study and predict things. We can study the weather and predict if it will rain tomorrow. We can asses how accurate our predictions have been in the past.

We can evaluate if it is true that smoking increases lung cancer, or that aspirin increases autism.

Sometimes we dont have good answers or predictions, and that too can be studied.

[pdonis]

> We can evaluate if it is true that smoking increases lung cancer, or that aspirin increases autism.

We can use real-world data to see if there are correlations. We can sometimes run controlled studies to get a better understanding of causal mechanisms. But we can't prove by logical argument based on "fundamental axioms of reasoning" that these things are "true".

[s1artibartfast]

I see. I agree that there are major limitations to purely logical arguments.

This is why unlike the rationists of the 17th century, modern "rationalists" predominately focus on empiricism (ironic i know). They are practically obsessed with Bayesian reasoning and probability of truth. This has the knock on effect of interest in prediction markets.

Is your impression otherwise?

[pdonis]

> They are practically obsessed with Bayesian reasoning and probability of truth.

Which is still math, not the real world. Bayesian reasoning is useful, but it's not a drop-in replacement for your brain. "Rationalists" often seem to talk as though they think it is--just follow this recipe and your thinking will automatically give the right answer. And anyone who tries to talk about limitations of this is obviously a crackpot.

> Is your impression otherwise?

Not with respect to obsession with Bayesian reasoning, no. But I seem to have a different take on that than you do. See above.

[s1artibartfast]

Im genuinely curious how you are differentiating "using your brain". What does that look like for the real world?

[pdonis]

> Im genuinely curious how you are differentiating "using your brain".

I just mean that your brain can do lots of other things besides Bayesian reasoning, and most of them can't be reduced to Bayesian reasoning.

[s1artibartfast]

Your response is dead for some reason and my vouch isnt enough to fix it.

Why cant the model be updated to include information about the concert? This is how basically every human works anyways.

It seems that yes, a bad model is not recoverable if you refuse to consider concerts or other data.

[s1artibartfast]

Like what are you thinking of that is so different? Can you give an example?