There are way too many variables for that to imply "programmers get fatter over time."

1) Maybe t-shirts sizes themselves are not the same between 1999 and 2008 2) Maybe the average age of the attender has not changed, so there's really no age component 3) Maybe a different demographic is now attending the conference 4) Maybe all Americans are fatter on average and programmers are just part of that trend

Interesting data point, but it doesn't support the thesis that programmers get fatter the more they've programmed unless you can show the attendees at the Linux symposium are older/more experienced now than in the past.

You'd probably also want some external data as I'm pretty sure people as a whole are getting fatter.