Although still under the weather, I was having an absolutely wonderful day here in Madison. Then I return to the place where I'm staying and, rather than read either of the dissertations or either of the articles for review in my bag, I decided to tackle this strange problem with one of my Stata programs (-asprvalue-, for fans of discrete choice models with alternative-specific data). The problem took me over an hour to figure out, and it turned out to be the result of a brand-new bug introduced in some recent update of Stata, rather than anything wrong with my own program. If my irritation was converted into energy and emitted from your monitor right now, you and everything at least thirty feet behind you would be in flames. Nothing personal.
Maintaining these programs sometimes feels like it takes way too much time. I am not a programmer! (I do really enjoy programming, the creative parts of it and sometimes even the detective work of debugging, but figuring out that an error you presumed was your own was really Stata's is extremely frustrating.)
Sometimes people ask how I learned how to program Stata. The story: In fourth grade, I didn't have an Atari like the cool kids, and decided that what I really wanted was a Commodore VIC-20 anyway. My parents did not have much money, but father got caught up in this moment of grandeur when I entered the Frontier Days Spelling Bee in Fort Dodge, Iowa, and told me that if I won, he would buy me a VIC-20. Even in my deepest and most tortured moments of intellectual self-doubt, I know this: I can spell. So I won and got my VIC-20* from K-Mart and this led later to them getting me a used Commodore 64 and by the time I was in eighth grade I was really good for someone who was entirely self-taught from looking at code and articles in Commodore magazines. But then, around this time the Commodores went south and PCs because more expensive and I discovered girls or Vonnegut or something, I don't know, but the upshot is that I didn't really do much more programming until I decided a few years into graduate school that, despite various declarations to the contrary, I wanted to do quantitative social research after all.
At first I worked with SPSS, which I found bewildering because it seemed designed to encourage users to do things in cumbersome and self-defeating way. Then, one bright day, I discovered Stata 5, and after like a forty-some-hour manuals-and-keyboarding-binge, all was bliss.
* The Commodore VIC-20 had 3581 available bytes of memory. If I added another paragraph to this post, I'd be over the limit.