Archie-a Darwinian development process

I was the manager of the small team of systems administrators and graduate students that developed and operated the Archie service, the first Internet-based search engine. I reflect upon some of the lessons we learned during the life-cycle of what began as a modest experimental development project. The seeds for the Archie system were sown in 1987, when faculty at McGill University's School of Computer Science in Montreal asked my team to investigate the possibility of connecting the school to the Internet. Our initial mandate was to provide less expensive access to specific research centers that were already connected, provided we could do so without impacting existing operations. There was no discussion of developing Internet-based search tools. In hindsight, that initial mandate seems modest indeed; but modest expectations, limited resources and a "Darwinian development process" of repeated research and release can lead to creative solutions. More importantly, this research environment was not goal-oriented, but rather process-oriented-an environment in which even a graduate student systems manager was allowed to pursue his research interests without interference, provided he and his team fulfilled their basic mandate.