Data Mining Idiosyncrasy

Hmmm ... a year or two ago I suggested that it was useless to try to turn Google-style statistical learning algorithms on the data of a single individual. I figured that the idiosyncracies of the individual meant that there was way too little data available to make viable models and predictions compared to tackling the population at large.

To emphasize my point I claimed that Google, sensibly, weren't trying to apply their algorithms to the contents of my Gmail mailbox.

Except, now, it seems they are.

I'll wait to see if I'm proved  wrong and they can make a success of this.