Re: Proposed Universal data types

From: Daniel Radetsky (
Date: Thu May 26 2005 - 00:07:22 MDT

On Thu, 26 May 2005 13:52:13 +1200 (NZST)
Marc Geddes <> wrote:

> > Seriously, you have figured out by now that it does
> > fit in, right?
> >
> No. No, I can't see how deductive reasoning fits into
> Bayesian Induction at all. Please entertain me.

Consider the probability calculus with propositions with probability equal to 1
or 0.

> There is inductive (associative) reasoning, where one
> is making probabilistic generalizations which don't,
> strictly speaking, follow from the premises.

But in Bayesian reasoning, near as I understand it, you make a generalization
with a degree of confidence, where the degree of confidence *does* follow from
the premises. I think you may be confusing standard induction with Bayesian

> Then there is deductive (conceptual) reasoning, where
> linearly ordered logical steps follow from axioms with
> (near) certainty.

I think that from {a -> b, a}, b follows with complete certainty. There's no
"near" about it.

> Now perhaps you might try to argue that deductive
> reasoning is just a special case of induction, but I
> don't see this.

look harder.

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:51 MDT