Expert Voice: The Limits of Decentralization

The whole point of the modern workplace is to pull people together so they can interact face-to-face. You get a knowledge spillover effect. If you get too dispersed, you lose a lot of the very reasons for having an organization. On the other hand, a highly centralized work force is obviously vulnerable.

The real issue is: How do you find the right balance? It’s not an either/or. You can have a completely dispersed work force without a lot of face-to-face interaction. Or you can take an intermediate approach with a small number of locations networked together.

The issue of granularity is key: Do you have a large number of small pieces or a small number of large pieces?

The requirements will be different for different types of organizations. University campuses have shown little tendency to do this. They’re highly concentrated. The reason is there’s a huge amount of creativity that goes on informally, conversations in corridors, bumping into each other. The more you can do to increase the density of those accidental encounters, the more creativity there is. On the other hand, a big financial services company might be able to disperse quite effectively, partly because the risks of concentration are so great.

There are some functions that are more important to decentralize than others. For example, it makes a lot of sense to disperse computer systems, to make sure there’s a physical duplication and backup. It’s probably important to disperse technical expertise. There’s no point in having an operational backup system if you don’t have the technical staff to run it.

But there are costs to dispersal. A lot is lost if there’s too much of it. A huge amount of informal interaction and transfer of knowledge takes place in the workplace. And no matter how good electronics get, some things are best done face-to-face, like very important discussions, things that require a very intense connection. At the highest level, a fair amount of what goes on depends on trust, on being known as a member of the community, and that requires being physically present.

As for concentrations of companies within one industry? That’s good in the sense that there is a knowledge spillover effect. You pick up on gossip or are seen as an insider. It’s important to be seen as an insider. One way to do that is by being there. If you want to be seen as a player in the computer industry, you better have a presence in Silicon Valley, for example. It’s also very important to be visible for your career advancement. It’s easier to be there visibly in physical space than by e-mail.

I think it’s unlikely that core high-level management tasks will be highly dispersed. They really do depend on trust, being seen as key to the organization and part of it. I think we’re likely to end up with a mixed situation—a lot of concentration at a high level, with a lot of dispersal of technical ability. I think the image of the beehive isn’t bad. You’ll have some kind of center, with many people buzzing around on the outside, sometimes coming in, then flying out.