Yeah, one interesting thing in the article was an offhand comment about how too much openness actually was counterproductive because people put on headphones, so didn't actually interact. Here probably, there is a 'right' amount of mixing with disparate groups. In the (admittedly only two) large groups I've worked in, you really needed to interact outside your local group to get anything useful done anyways.
Actually, this reminds of the network effects section of 'Where good things come from' (Steven Johnson), which actually points out there is a right amount of 'edge' density - too little and you don't see outside your local group enough for ideas to flow; too much and ideas flow but don't tend to die before they can reach critical mass.