Still No Silver Bullets
On the first main conference day of XP2011, Esther Derby is speaking in the first keynote about the downsides of Agile transitions. She has these amazing hand-written slides… In fact, last night on the way to my room, I passed a person sitting in the lobby working on her slides. In the lift, I thought, you’ve seen that handwriting before… I went back, and as it turned out, it was in fact Esther:-) We had met on Twitter before, but not in person. I think this was the first time ever that I recognised a person I hadn’t met or talked to before by their handwriting.
She starts with the assessment, that agile has had substantial successes on the team level, but often fails in the large: ScrumBut, Scrummerfall… Michael Sahota lead a session about agile failures in the large at Play4Agile and wrote about it here, Esther’s points remind me of that discussion in multiple areas.
Differentiation of centralisation (visibility, coordinated action) vs. decentralisation (fast decisions) on the upside and (centralised) slow decisions, poor response to customers, (decentralised) reduced visibility, slipping control on the downside. She describes organisations as oscillating through these up- and downsides of central vs. distributed control. I find that view enlightening as it matches patterns that I’ve seen in practice, and I think many of the failures of lean/agile or misuse of practices I think is owed to the fact that they are used at the wrong place.
Managers are noticing feedback, but they are responding too late and too aggressively. We must manage the downside with more robust feedback loops. On the team level, we know how to do this, how to make teams robustly self-organising.
How do you differentiate a robust organisation? Esther asks for experience with American Airlines, and gives an example of an airline that was merged out of two. You can still tell which company a crew originally had been in, as employees of on of the airlines are happy, treated well and treat their passengers well, whereas the others… don’t.
Esther quotes Donella H. Meadows, who wrote “The original purpose of a hierarchy is always to help its originating subsystems do their jobs better.” Failures are often failures to balance the interest of the system with the interests of the subsystems, caused or reinforced by budgets, targets, etc. Targets alway make the work worse, she says… I can relate to that with numerous stories and have seen targets and incentives cause a number of common dysfunctions.
Esther sees a bifurcation of knowledge in organisations: Day2Day knowledge is spread in different areas from contextual knowledge, and the overlap is too small. The hierarchy filters and distorts information in both directions (top to bottom and bottom to top). That leads to local optimisation leads to poor and slow decision making and makes people on both sides look foolish. We need to increase the overlap to be more successful.