This is a quick brain dump of the day for me – brief summaries and key highlights that stood out.
The conference kicked off with an interesting keynote by James Suroweicki, author of Wisdom of the Crowds. Essentiallly this is about how collective intelligence can be better than individual intelligence. For example, we had a live experiment where all attendees were asked to guess the number of lines of code in Visual Studio as part of the registration process. The average guess was 47 million and the actual answer was 43.2 million – only an 8 percent difference. In fact only 2 out of the 15000 attendees had better guesses. (The closest guess as actually by a Conchango colleage, Toby deBelder). Other examples given were Who Wants To Be a Millionaire’s Ask The Audience option, where the audience is correct 90% of the time, and horse racing betting, where the ‘favourite’, as determined by the crowd of gamblers, is invariable the winner. In order for the wisom of the crowds to work however, a number of conditions are needed, including cognitive diversity, in order to have different perspectives, and independence, to avoid group think. All of this makes perfect sense when relating it to cross functional teamwork as a basis for agile software development teams being successful.
The Feature Injection by tutorial Chris Matts was essentially a guide to how the business analysis role fits into Agile. Rather than trying to push every requirement into a system (via documentation) the BA ‘injects’ value-derived features using a pull model by describing examples. From a kanban perspective, Feature Injection describes how the features in the system are discovered and scheduled.
In the afternoon I was part of an Open Jam on the Evolution of the Agile Model. This was a follow up to a similar conversation last year, planning on how we can be more active in taking things forward next year. More to follow soon hopefully.
Finally, I went to the talk on Code Metrics & Analysis for Agile Projects by Neal Ford and Ram Singaram. The metrics themselves are really useful, but its important to remember that they are just numbers – information to be shared and used to inform decisions about the code base. They need to be “read like tea leaves” as they are context dependant. One really cool things I found was the use of tree maps to visualise the state of the code base.