Specifically, Cooper wants to allow the interaction designers to work by themselves for a while to create their design. Beck wants them to interact with the programmers. I'm a big believer in Conway's law -- that the communication within software maps to the communication within the organization that built it. If you want software with excellent interaction with its users, you need to have excellent interaction in your team.
There is a lot of confusion and mistrust among testers about XP. I was suspicious from the start, myself, but also intrigued. I've seen software evolve and was intrigued by the attempt to maximize the speed of software evolution. My column published this week at Stickyminds is my attempt to try and dissolve some of this confusion.
I spoke about agile testing at several locations last year. But this weekend's workshop has already given me some new insights. Here's one.
I've been recently doing some test-first programming in Ruby. While doing this, i found that it was often easier to implement the code in with the tests -- often in the setup method -- rather than write a separate class. This seemed a bit like cheating, but Rob Mee's presentation of using Fit to implement acceptance tests did the same thing. What's interesting about this is that it really makes it easier to evolve the form of the code. He's promising to write more about Fit and acceptance-test-driven development. I'm eager to see it.
Several of us have decided to blog about the work we are doing together. Ward, who invented the Wiki, recently was invited to speak at a conference on social software. There were a lot of bloggers there, and he came away enthusiastic about blogging. I plan to be writing more frequent updates here, with links to the blogs that the others will be starting. Stay tuned.