The way spambayes estimates the probability that a msg is spam given that it contains a specific word is thoroughly Bayesian, as described on Gary Robinson's web page, and in his March "Linux Journal" article.
The way spambayes combines probabilities ("chi-squared combining") is indeed not Bayesian at all. The probability combining scheme Paul Graham suggested isn't correctly Bayesian either, unless you assume the universe consists of equal numbers of ham and spam messages (so that the prior probability of spam is 0.5).
> Guido is a fairly discrete person, and it's
> always kinda difficult to know what he thinks
That's why I get the Big Bux to channel him.
> about functional programming
No secret: lambda, map(), filter() and reduce() were contributed code. He used to call them minor conveniences. Now he calls them minor annoyances, as they "opened the door" to people trying to use them as the foundation of a programming style Python wasn't designed to support. Guido still uses them himself as minor conveniences, but his brain isn't wired to use them for more than that. The point isn't theoretical purity (either pro or con), but plain effectiveness in practice for programmers of ordinary skill. List comprehensions (a la Haskell or SETL) were added much more recently, and on the latter basis: in the absence of dogmatic abuse, they're *clearer* than the loop-based alternatives. map() can be too, of course, but listcomps are even clearer than map. OTOH, I've never seen a Python program using reduce() that wasn't better written via a loop instead.
> typing
Yes, he types. I don't think he's even tried speech recognition yet .
> Python/Java complementarity or rivalry
Complementary. Indeed, the happiest Java programmers in the world may well be those who use Jython as a front end for prototyping, managing, exploring and testing their sprawling Java code. Java and Jython work very well together!
Slashdot Mirror
The way spambayes estimates the probability that a msg is spam given that it contains a specific word is thoroughly Bayesian, as described on Gary Robinson's web page, and in his March "Linux Journal" article.
The way spambayes combines probabilities ("chi-squared combining") is indeed not Bayesian at all. The probability combining scheme Paul Graham suggested isn't correctly Bayesian either, unless you assume the universe consists of equal numbers of ham and spam messages (so that the prior probability of spam is 0.5).
> Guido is a fairly discrete person, and it's
> always kinda difficult to know what he thinks
That's why I get the Big Bux to channel him.
> about functional programming
No secret: lambda, map(), filter() and reduce() were contributed code. He used to call them minor conveniences. Now he calls them minor annoyances, as they "opened the door" to people trying to use them as the foundation of a programming style Python wasn't designed to support. Guido still uses them himself as minor conveniences, but his brain isn't wired to use them for more than that. The point isn't theoretical purity (either pro or con), but plain effectiveness in practice for programmers of ordinary skill. List comprehensions (a la Haskell or SETL) were added much more recently, and on the latter basis: in the absence of dogmatic abuse, they're *clearer* than the loop-based alternatives. map() can be too, of course, but listcomps are even clearer than map. OTOH, I've never seen a Python program using reduce() that wasn't better written via a loop instead.
> typing
Yes, he types. I don't think he's even tried speech recognition yet .
> Python/Java complementarity or rivalry
Complementary. Indeed, the happiest Java programmers in the world may well be those who use Jython as a front end for prototyping, managing, exploring and testing their sprawling Java code. Java and Jython work very well together!