Okay, let's start.

Okay, so we probably have the right audio. Yes.

And, very good. So,

I've looked a bit at numbers. And one thing that struck me is that on Shudon,

we have 139 members that are not me and Zedalik. In Kampo, the exam is currently 60.

And the quizzes are 31, relatively constant. I know that this number is basically the number

wanted to look and see what's going on. That's fine. This description is slightly worrying

because there's a very good correlation between doing the quizzes and passing.

There's not quite as good a correlation between not doing the quizzes and not passing.

But it's still rather strong. Oh yes, and in class,

seven, today. It was a little bit higher than the rest. Okay, so typically those pass well.

These pass.

And here, not everybody passed. So,

especially for the Zoomies, or the not even Zoomies, the Watchies, later,

that is something you should think about. I am not very keen on

that. But I'm also not scared. Okay, so

make of the numbers what you want.

We want to go to the end of Fragment.

And nope, Fragment 2.

So, last week,

Oh,

last week, we had a relatively discussion heavy

session where we introduced this notion of a Tableau Machine,

which is fine because this Tableau Machine and the important

image is basically this one.

Because the Tableau Machine is the thing that is probably most important here because it models

or gives us kind of a framework for modeling semantic pragmatic analysis,

which is the difficult phase. We're going to talk about the grammar that defines the fragments and

how to do all of this and the logics. But whenever I say logic, I will from now on

assume that it kind of lives there.

Okay, so we're always, if we can, look at a model generation calculus and how it would look in that

particular logic. And otherwise, take it as a research project if we don't have it.

So the idea is do something

that is shaped like this, eventually. And there's a fully separate research question of

how should this exactly work and so on. Okay, and how would we go from, and that's, we've done a

couple of examples. We've

asked ourselves that correspond to

cognitive science, psycholinguistics, there's a full field of research that kind of tries to

open up the brain and study what is going on there when we are

understanding natural language. And we've talked about the results there. And

the other thing that's important to understand is that the Tableau Machine is interesting

because it gives us what you call the competence model. The competence model is what are the things,

what are the readings a human could in principle derive.

Rather than what a performance model does, make predictions about

which of them, and definitely only a subset, is actually derived.

So the Tableau Machine is kind of still relatively much at the

competency model side, but it kind of gives you some avenue how we could

make it work. And so,

but it kind of gives you some avenue how we could make it into a performance.

A model that we understand and can tinker with, that is different from all these neural models,