Bayesian probability

And I wanted to go and just do a deep dive on those drivers and try to understand, based on history, a very classic kind of Michael Mobizen approach.

What's the base rate probability of these things happening throughout history? And what would be the markers that, in this case, I could kind of use a Bayesian updating framework to update based on a base rate, based on what evidence I was seeing in the market, what the probability of these things were happening.

The type of analysis that if you try to do it by hand, just getting base rate probabilities on what I was looking at would have taken an enormous amount of work.

And AI was able to walk me through it.

And, you know, it's been, you know, that has been the right decision.

Now, whether or not I passed for the right reasons or the wrong reasons, maybe we can debate.

But, Base rate analysis, Bayesian type of updating.

It strikes me as that AI lends itself very well to that because it can go through an incredible speed and look at similar instances throughout history and develop base rate probabilities that we can then throw through a Bayesian kind of framework.

So the decision depends on the probability that the nodule is malignant, the response to chemotherapy, the molecular risk, and the consequences of being wrong in either direction.

This is essentially Bayesian oncology.

The initial CT sets a probability.

Response updates it.

It came from Bob, really.

that whereas you can use mathematical probability and statistical reasoning to understand certain patterns in human behavior, that the ideas of chaos theory and combining this through the filtration of Bayesian reasoning as a bridge that not only can investigators use, but if you happen to be playing a game of traitors yourself or in any real situation where you require following, an expanded pattern of human behavior interaction, then it's really for anyone that's been handed a statistic and then asked to apply it to someone in that kind of way.

Bob, what inspired this within you?

Well, I have been watching The Traitor, so I won't give any spoilers away.

And the chaos part of that is that it's just a small change that you've made to that situation, which has then pushed you somewhere else.

Yeah.

When I look at Bayesian reasoning within that, I go back to the applied behaviour analysis, the one they do with children.

They're probably the only real... um academic use of the word behavioral um analysis in that term um and they talk about um the abc and you see its behavior consequence and that is very similar to bayesian reasoning isn't it what do you think

Rationality training with a bit of socializing.

The game involves Fermi estimation, calibration, Bayesian updating, world modeling, and perspective taking.

You also get to know the other players better and find out how they think about all kinds of topics.

Materials Each player needs an equal stack of chips and writing supplies.

You have to map the cultural variance.

A fascinating algorithmic approach to this is called the Bayesian truth serum.

The Bayesian truth serum.

How does an algorithm map subjective cultural truth?

It uses game theory and probability to restructure the annotator's incentive model.

Normally, you just ask the human, "Is this comment toxic?" But with a Bayesian truth serum, you force the annotator to answer two distinct questions.

First, you ask their subjective opinion, "Is this toxic?" Second, you ask them to predict the group's behavior, "What percentage of the other annotators do you predict will say this is toxic?"

That is fascinating.

Okay.

And I'm gonna disagree with you about a few things about how you understand Bayesianism.

Um-

Good

Yeah.

Um, okay.

So let's start with, uh, just a basic overview of like what does Bayesianism assert? And then my follow-up is going to be how does this relate to like a material theory of induction? 'Cause I, I really like getting-

Right

The clinician is relying too heavily on the apparent resemblance and insufficiently on the underlying probabilities.

Bayesian reasoning offers a more systematic alternative.

Bayesian reasoning begins with a prior probability, such as the prevalence of a disorder, and updates that probability using additional evidence.

The resulting estimate is called the posterior probability.

For the EPPP, remember that a positive test result does not automatically establish a high probability of disease.

Which brings us to the core technical innovation of the paper.

Expected Bayesian loss, or EBL.

Yes.

Let's break down the mechanics of this.

It actually acknowledges the variance instead of ignoring it.

Exactly.

So if an experiment doesn't have enough data to prove it's safe, the expected Bayesian loss will inherently be high and the system will flag the launches too risky.

You've got it.