Thursday, October 04, 2007

Alcohol

Very funny xkcd.com cartoon.

Since alcohol slows reflexes, one would think that it would slow thinking in general. As programming requires some of most intense thinking anywhere, one would expect that imbibing would categorically be a bad thing. Yet, there are two pieces of evidence that i'm aware of that contradict this.

I worked with a guy who preferred working from home, over a slow modem, to working in the office. At home, he could work with a six pack of beer (brand not specified). He said it was less painful. Let's say he sees a bug. When sober, his reaction was Damn! Another bug. But after a beer or so, his was reaction was Ha ha ha, another bug, and he'd get right to it. So, apparently, psychology can be an important factor for programming.

But i also did an experiment, mostly by accident. In the mid 80's, i was at this keg party of mostly geeks. Thing about keg parties is that i'm a really cheap drunk. It just doesn't take much to get a buzz. So, i can't really count cups (they were plastic, and so can't be properly called glasses) after one. I've no idea how much i had. Enough so the Universe spun, but not enough to bring me to the ground. Anyway, this guy, whose name i don't recall, was talking to me, and i mentioned this graphing package that i'd just written. We had forty something devices capable of plotting results, and they were all different. Our plotting packages only knew about three or four of these each, so you got your output on one of those. If you wanted output on a better one, you'd start from scratch using another package. So, i invented an intermediate format, and a series of filters that could translate to and from it. This guy said he had access to a new laser printer, and had the manual for it. How long would it take to create a filter? I'd have said a couple hours. You know, just take a filter that looks like it, and modify it. He wanted to do it right away.

Now my judgement was nearly entirely impaired, so despite the time, maybe two in the morning, we started right away. We staggered to my office, and started hacking at it. I don't remember the details. But when i got in to work later in the morning, the resulting filter still worked. Further, the code wasn't bad. Over time, the bugs will eventually show themselves. But no maintenance was needed for this filter. Ever.

Windows ME really was pretty awful. But it seems more likely that alcohol was involved in the decision to release it rather than in development.

No comments: