27th
Brief thoughts: Computer science versus statistical programming
One key reason I adore coding in computer science, and seem to come close to loathing it in statistics/econometrics, is that the sense of accomplishment is radically different:
In CS, you build something, and each step you have another function completed (another side of the house, another piece in the puzzle, another routine you never need to repeat again) that just works.
In statistical programming, however, you run a regression, and it does work, first go around (barring blatant syntax errors). The problem is, your output could be crap and still look quite similar to how it ought to look. The fact that it works doesn’t mean you’ve accomplished anything. And once you’ve done all the work in the world, you still don’t know if it really means anything. It’s a constant feeling of insecurity; a persistent lack of satiation; an unceasing uncertainty even in the most reasonably-feasible of certainties.
(I ought to note here I’m intentionally avoiding a statement linking this to the postmodern condition. This is a post about coding, after all.)
So screw R; I’ll stick to web.py and Django. I have enough existential angst as it stands.
