Changes

Lately I have been in a contemplative mood. There are several good reasons why. I have started mentoring (via email) a female undergraduate at a university in the Midwest. The project I have spent the last 3 1/2 years on is increasingly under its own management and direction. In April, my wife is expecting our first children (twins!). And, worst of all, I am getting really cranky about the software I use every day.

From 1983 to 1996 I worked in computer science research at Bell Labs, beside some of the best in our field (Ritchie, Thompson, Kernighan, McIlroy, Pike, . . . ). Even better, I got to use their software! Eighth, ninth and tenth editions of Research UNIX, and the various editions of Plan 9. Best of all, we had Dave Presotto riding herd on the two greatest fiascos of modern computer systems, mailers and networking. During this time I was quite productive in my work, largely due to the quality and smoothness of my computing environment, especially the effective pervasive user interface and window system done by Rob Pike.

Strangely, I was often pitied by outsiders as stagnating in my own insular backwater. I was missing out on the software revolution, the brave new world of the PC with its stunning, fabulous diversity of really keen products to help me do my work and increase my productivity. When I moved from Bell Labs to AT&T Labs, I undertook to experience the mass market full on! I tried to live on the frontier, using Windows 95 and X. Furthermore, my projects involved production systems, so I got to see and use real software (not that fake stuff we had in Research). After three years, I have learned some lessons and thought I'd pass them along. Of course, these are just one person's opinions; I would expect your mileage to vary.

1. As far as I can tell, the only useful outcome of the PC business is high-performance commodity hardware. The software sucks. Even at its worst moments, Plan 9 was more reliable than Windows 95. Of course, it worked okay (meaning it only crashed every week or two) if you didn't do much networking or open many windows or install any new software. But this sounds like giving up to me.

2. But what about the fabulous marketplace of software to do my bidding? Well, I tried that. I wanted some JPEG/MPEG viewers. So I followed the new order and went to <www.tucows.com> and downloaded every freeware/shareware viewer they had. Of the fifteen I tried, five destroyed my Windows software (complete reload required), and eight coredumped or caused the system to crash. But two actually worked! One of these had one of the worst user interfaces I've seen in some time; one was actually usable.

3. So I am giving up on the whole Windows thing. I am switching to Linux or a BSD system; it should be much better, and I can't imagine it being worse.

4. Perhaps more important, I think I am shifting to the open source camp. Perhaps I was unlucky, but in my current project we used five distinct software products on the production machine (other than what we developed). Three were commercial offerings (the system software, a backup suite, and a file transfer suite), and two were supported by folks in AT&T Labs - Research (sorting and searching programs). After fifteen months, the story was:
   • Searching: 0 bugs, high performance
   • Sorting: 5 bugs, two months to fix
   • Backup: 10+ bugs, several tapes' worth of data lost, mediocre performance, 1 fix delivered during severity-1 mode
   • File transfer: 5 serious bugs, 3 fixed by patches within a couple of months, 2 still unresolved after a year
   • System software: 10+ serious bugs, all but 1 fixed within 2-5 months

I seem to have come full circle back to where I started in 1975 -- running UNIX and software developed by a worldwide collaborative community. And for pretty much the original reasons: it's a very effective way to use the hardware, and the software (and software environment) is simply better. (Of course, I'm still hoping for another release of Plan 9!) Although some part of me is depressed at the apparent lack of progress in 24 years, the rest of me is excited at getting back to an environment where I am much more effective.