Monday, November 20, 2006

The Agile Joke

Steve Yegge's epic rant against Agile is entertaining for more reasons than one. The follow up post even more so. Having being an unwilling participant for the most part and sometimes hostage in Agile projects, Yegge had me laughing until my sides ached. I can't claim to have read all the literature on Agile that is out there, so maybe someone has figured the way around to some problems I see with the methodology.

Metrics stands out as the biggest one. In the old fashioned, traditional way there used to be Earned Value Analysis to tell us how we were doing on the triple constraints and burn rate. The ratios meant the same thing to everyone and the project sponsor could monitor health at all times. I think that was a good thing.

In the Agile world, those measurements are hard to come by. Everything is relative. The complexity of a user story can be measured in gummy bear units and the team could have enough velocity to take on two hundred gummy bears worth of work in a sprint. The hapless customer has no freaking idea how that translates to the time, effort and money or easily digestible EVA numbers.

Scope and acceptance criteria of work is highly fluid as necessitated by the operating definition of Agile. Used to be the baselined project plan and its supporting artifacts were sacrosanct. There was predefined and mostly static success criteria as there was accountability at individual and team level.

Sure, scope crept, business requirements were found to be crap once UAT got under way, budgets were over-run by 200%. All of that happens in the Agile world as well. Only instead of being reported as a failure it gets misrepresented and misinterpreted as success because stories invariably get closed out at the end of every sprint and gummy bears are about as useful a measure for software development effort as cowrie shells are for making the down payment on a house.

No comments: