Recently the Lean community has been enamored with two hot topics now that elimination of waste and value stream mapping are passe. I made my feelings known about Genchi Gembutsu (going to the source) back in March twice. The other hot topic is set based design. The robot above belongs to Katherine Radeka who was the chair of the recent Lean Design and Development conference. Katherine uses the robot in a new set based design theory training class she teaches to (mostly) software people.

This box of parts is used to create the alternative architectures for the problem. The original robot is not designed to pick up the axel in the top photo. It has a new problem to solve - pick up the axel and place it between the goal posts. A set of alternatives will be developed and tested for performance, manufacturability, cost, and so forth.

The idea with set based design is that multiple designs are created and kept alive until gradually they are eliminated and a best choice is made. Best choice will come from a combination of economics, manufacturability, performance, function and so forth. Set based design is a logical extension from the Lean concept of "decide at the last responsible moment." In this case, it is the architecture that is being locked down at the last responsible moment. Actually, in manufacturing they have another name for this, "postponement." It is the idea that you postpone decisions as late in the value chain as possible. In some cases postponement goes all the way to the point of retail sale. Remember removable face plates for cell phones? Or how about paint mixed in the store at the point of sale? Remember when stores carried inventory of colors and color fashion changed with the years?

However, the new set based design hype worries me a little because for years we've had a body of knowledge in our business around developing alternative architectures. The Software Engineering Institute offers us ATAM. While on MSF we picked up work by Jeromy Carriere at Microsoft on Lightweight Alternative Architecture Assessment Method (LAAAM) that offered an agile alternative to ATAM. The idea that we develop alternative architectures and assess them and choose the best solution based on a range of criteria is not new. In fact making architectural alternatives is required for CMMI Level 3 as part of Decision Analysis and Resolution (DAR).

What I think set based design is bringing to the picture is the concept of real options theory and the postponed decision making. This is not really present in ATAM or LAAAM that have a single early lifecycle decision point.

Despite the hype from the Lean movement, there is a gap. No one with real software engineering systems and architecture experience has occupied the space and explained how to synthesize postponement and real options theory with alternative architecture assessment. Technorati tag: Agile, David+Anderson, ATAM, SEI, CMMI, Set+Based+Design, Lean