Thursday, August 19, 2004

Cambridge philosopher Peter Smith, in a manual for a LaTeX presentation package: The philosopher Stephanie Lewis, in a stage whisper during a conference talk: ‘Power corrupts: PowerPoint corrupts absolutely.’ Didn't Tufte say this first? Actually, I find PowerPoint (and even more so Keynote) quite usable for technical talks and lectures if I follow a few simple rules: Include what you need to show, not what you should say important equations graphs quotations visualizations Simple color scheme, without fancy background No special effects Don't overcrowd slides Tufte as well as most linguists I know prefer handouts, but these are really impractical for large audiences. 3:18:17 PM

Unemployment drops, optimism rises for computer professionals. Is it because the unhappy ones leave the field? [CNET News.com] What a confused article. Lamentations of the poor state of IT employment always ignore differences among specialties, and especially differences in performance among workers. You may well have an aggregate slump coexisting with unfilled jobs at the highest levels. That's what happens in other areas with big individual performance spreads, for example professional and semi-professional sports. We also see that in computer science graduate student admissions. Year after year, we have more applicants just from the US than we could possibly admit. There are two reasons the demand far exceeds the offer. First, we try to identify those candidates who have real research potential, who are a small fraction of the formally qualified applicants. Even among successful applicants who go on to finish their degrees, there are big differences in performance. It's the nature of this line of work. Second, graduate fellowships are funded for the most part by federal research dollars, which in long-run average grow slowly. If we were to double stipends as the article suggests, we would have to half the number of admits. The fact that there are so many more applicants than admits, even from the US, indicates that stipends are not a big issue. The reason is that, except for some economists, journalists, and union organizers, everyone realizes that graduate study is not just a line of technical work with similar market properties to, say, writing code for a software company. Graduate students in computer science have a high degree of autonomy in choosing their research direction. They do contribute to ongoing research projects, but much of their time is rightly devoted to learning from courses, seminars, their advisers, and other students, and defining their thesis research. They know (as I did when I was in graduate school) that they are doing this because they want to learn how to advance the field, and that is worth quite a bit to them, overwhelming whatever difference in long-term earnings there is between them and similarly capable people who did not go to graduate school. 11:55:19 AM