Re: [exim] Bayesian filtering idea

Tony Finch wrote:
>
> You're dealing with P(spam | date) and P(spam | no date) which are not
> directly comparable (though reference to an A-Level probability textbook
> should help).

Good point. I'll stop making pronouncements on stats before my third
cup of coffee in the morning, and stick to far fetched security analysis. :)

- Marc