Monday, October 17, 2011

Density and productivity

I just learned (silly me for not knowing) that it takes density to get decent productivity. And policies that discourage density have a likely unintended consequence of squeezing off improved productivity. This seems to be the message from Ryan Avent

In cities and urban areas, his advice is don't discourage growth, or least make it affordable for more people to join in the high production center. Dis-affordability, we are told, ultimately leads to declining, or least not improving productivity.

So, now transfer this to the project domain. (Notice: switching domains is often problematic and solutions often simply don't transfer.  However, Avent himself brings up the topic of work productivity vs distribution, so perhaps I'm not too far off base). The thing that jumps to mind is the co-located team, a favorite if not a mandate for the agile community, and the distributed or virtual team, a favorite if not a mandate for the project at enterprise scale.

The conclusion is begged by the set-up: we should either expect a certain productivity premium for the co-located team, or on the flip side, we should expect to discount the productivity of a distributed or virtual team. Whether discount or premium depends on where the baseline is. From my experience, working mostly in larger scale enterprises, the baseline is the distributed team, so I see a premium for the co-location case.

So, in the context of "density begets productivity", I'm probably taking the right view. Co-locating, where possible, show pay a productivity dividend. However, reversing seven decades of ever larger scale--since the onset of WW II in 1940--is no small matter, and probably not desirable or practical. If we going to "go big" as the popular refrain goes, we're going to need big, distributed teams. We're unlikely to use agile methods to build Hoover Dam or the Golden Gate bridge (projects in the 1930's, and unprecedented in scope for the time, and pretty impressive, even today)

What's to do?  Distributed teams are less productive, but necessary in most cases to get true scale.  The important thing is have the right reference class for estimating.  Obviously, the reference class should be one that likely has all the productivity discounts built in, else an underestimate is almost a foregone conclusion.

 Bookmark this on Delicious  
Are you on LinkedIn?    Share this article with your network by clicking on the link.