Pair programming on the C3 project

Abstract

Chrysler's Comprehensive Compensation (C3) project was one of the first large-scale IT projects on which Extreme Programming (XP) precepts were attempted. As XP was being invented, we didn't really think of it as a theoretical or methodological breakthrough; it was simply an opportunity to get the job done. The "theory" came about later because of the practice's success. In my opinion, XP is not a theory, but a cogent descriptive body of successful praxis. Some people consider pair programming, an XP technique, to be difficult, unworkable, or downright weird. After all, what does the workstation represent, if not finally having your own computer? The idea of sharing your machine with someone seems like a lot of bother without much reward. How can you talk if you're trying to think? And how can you think if you first have to explain what you're thinking about? Having participated on the C3 project as a performance-tuning consultant beginning in 1996, I can attest to the success of pair programming. At the same time, I can show instances where it did not work or was not emphasized in this early XP project. In a sense, this pragmatism is the strength of XP: Use a technique where it works, ignore it where it doesn't. XP has never been prescribed as a panacea. XP's pair programming helps programmers synthesize their individual expertise into an effective combination.