Where do the managers (and team members for that matter) fit in who have tried telecommuting experiments, seen productivity drop, moved back to co-located teams, and seen productivity rise back to previous levels again?
Well, it works for some, so either the team of this theoretical person or manager was a unique snowflake or there were people problems causing this distinction (e.g. people who abuse it, crappy manager who was using stupid metrics that prove their biases).
Wasn't trying to say that "telecommuting doesn't work". It obviously can - and there are often good reasons for it (hell - I do it myself most of the time).
I was just pointing out that people choosing not to telecommute isn't necessarily caused by management idiocy. There is, indeed, a lot of evidence (see http://news.ycombinator.com/item?id=4207309) that the optimum for productivity is a radically co-located team.
Of course there are other issues. Telecommuting is a great option to have and solves some problems. Sometimes the best folk cannot co-locate. Sometimes quality of life issues are more important. Sometimes the on-site work environment can't be made effective due to issues outside the relevant people's control.
However, a lot of geek folk seem to automatically assume that telecommuting == always more productive, and management that prefer co-location == always idiots. This isn't backed up by the facts.