--On Tuesday, September 14, 2004 12:15 AM -1000 Aaron Kagawa <[EMAIL PROTECTED]> wrote:
I guess the whole point of my previous email was to try to narrow down the scope of what Cedric was trying to do.
That's an admirable effort!
My opinion is that the first step for Cedric's thesis is for him to not worry about defining Quality or Productivity as such, but rather to gain expertise at using telemetry-based information to provide novel and interesting perspectives on development that are not otherwise apparent. As a simple example, one could think about development "hot spots". You can think about hot spots in various ways: areas of the system that are being worked on intensively by a single person, areas that are being worked on by multiple people simultaneously, or people who are working intensively in one way or another, or areas that are being tested intensively, etc. Now think about looking at hot spots over time: are they moving? Is there more than one hot spot present at a single time? If we had a suite of Hot Spot measures at our fingertips (for example, a Telemetry Wall Scene), and could basically watch them while we worked, I believe that we would start seeing some interesting relationships between Hot Spots and other factors in development more directly related to Productivity or Quality. We might find, for example, that Hot Spots can "collide", and when they do we need to do reviews in that region in order to keep everyone on the same page and coordinated.
Hot spots are just one conceptual entity to explore--Cedric and I sat down one day and came up with others including "stability", "progress", and the relationship between the team architecture and the software architecture. Dan has been working on the concept of "project volatility" for a while, which would also be interesting to explore. All of these can be operationalized in different ways, and once operationalized then Telemetry lets us look at their dynamic behavior over time and watch for interactions and effects.
Our telemetry infrastructure is like a brand new power tool that came without an instruction manual. Cedric needs to experiment with this tool intensively and start figuring out how to turn it on, how to hold it, and to start building things with it. For now, my hope is that Cedric will be presenting to us a new Telemetry Scene or two every Wednesday on the Wall that tells us something interesting that we didn't know before about Hackystat development. Once he can do that reliably and regularly, then the narrowing down part of his thesis will fall out quickly and naturally.
Cheers, Philip
