Comments on: Exploring a Development Team's Momentum

By: john

john — Sun, 29 Mar 2009 05:05:47 +0000

very cool stuff rob…interesting insight that speed of a project’s completion is dependent on marrying velocity and mass to overcome distracting collisions using momentum…

i wonder if the analogy might play out in a couple of other ways too…for example, maybe mass could refer to the quality of a collaboration by a production team…in which case a small highly collaborative team would typically generate less friction and thus the most mass, and thus achieve a faster completion rate than larger & better-resourced but less nimble and more resistant teams, which is sort of counter-intuitive…though it lines up with some thoughts I’ve had lately too…not to mention the 37signals credo…

By: brett tackaberry

brett tackaberry — Mon, 02 Mar 2009 00:29:51 +0000

Alain, that would work for a dev shop, but the model doesn’t fit a multidisciplinary shop. You need to expand the metric a bit – this doesn’t take into account the work that isn’t coding such as planning, design, research, consultation. It has to be measured by progress against a task. Progress is measured in time (distance), and the validity (direction) of a task would be measured in performance (equivalent to elevation off zero). In the case of some agencies, the task is approved to be invoiced or not. The units would be hours of successful work over time.

I am interested in the discussion about potential energy.

By: viller

viller — Fri, 27 Feb 2009 20:12:45 +0000

Don’t forget your bug count metrics. Alain, sounds like we have a project we need to start.

By: Allain Lalonde

Allain Lalonde — Fri, 27 Feb 2009 14:46:29 +0000

Interesting idea, I can’t help but wonder how you would go about quantifying this stuff.

Maybe Velocity as: (# of new lines – # of modified lines) / time

Got me thinking now.

It’d be pretty cool to get some metrics that could be retroactively analyze a subversion repository. That way you could test out the metrics against the Quality of the source code produced to see if the are good predictors.