One of the things that dismays me is the pathological bureaucratisation of some of the organisations with which I have to deal, so this perceptive observation by Clay Shirky struck a chord.
Process is the feature creep of organizations. In the same way software has to have features, groups have to have process. But like software, process creep in groups is insidious — each additional check in or form seems to cost little and add much, but over time, the cumulative overhead of process can hamstring an organization, almost without their noticing.
Six or seven years ago, ATT asked me to spend some time helping them figure out their web hosting offerings, and after some preliminary work, it became clear that there would be no mainstream hosting business, because the cost to the customer would be too high to be competitive. This was not because ATT was buying expensive hardware; it was because their minimum hosting processes imagined layers and layers of dev, stage, and live servers, and a complex array of user management interfaces. When ATT asked how the existing hosting companies could provide their services so cheaply, I said that the competition was simply offering shell access, and that people could FTP anything they liked to the server or telnet in and write stuff directly on a live box.
ATT was aghast, of course, at such laxity, but in fact, it was this kind of simple, process-lite attitude that helped the net spread generally, and it was ATT’s “Quality of Service” attitude that marginalized them.
This is many stories, of course, hundreds of stories, thousands of stories. It’s why Berners-Lee succeeded while Nelson failed, it’s why markets work better than central planning, because central planning is process made supreme, and it’s even why Open Source works though it has less process than commercial firms. This is not to say that there is not a process to Open Source efforts, but rather that it is considerably simpler than the process adopted by Serious Commercial Software Firms®, who for years misunderstood Open Source, because they assumed no one could build software with that little process.
