At 12:37 AM 5/1/03 -0700, Bill Stewart wrote:

At 12:09 AM 04/30/2003 -0500, Jim Choate wrote:

...

On Tue, 29 Apr 2003, Major Variola (ret) wrote:

...

FWIW: There's a paragraph in the current _Science_ that mentions "Winnow", a program by some .il researchers that guesses the sex of authors by their writing. They claim 80% accuracy on general lit

Perhaps impressive, perhaps not. What's the actual percentage ratio? I bet it's decidedly male.

...

and 74% accuracy on 30 science papers.

Considering the known sex ration in science fields one can guess 'male' all the time and hit that level.

Yup. Weather forecasting in Ithaca New York was similarly simple. "50% chance of rain today" (Or snow, in the winter...)

Jeez Bill, I'd expect such cluelessness to be limited to JC.. Don't you think they'd pick their sample set to be more useful? If you're testing weather prediction algorithms then a desert is not a useful place. --- A stopped clock is right more often than one that runs fast (or slow)