I’m in Orland this week attending the 2010 StarEast conference for software testing, and was surprised at how much hadn’t changed. Its been over a decade since my last tech conference, and while the names have changed, the actual recommended practices haven’t changed much at all. Speakers are still preaching the values of early testing, automated test suites, and automated build processes that I learned as a young engineer at Sun Microsystems 20 years ago, but now they’re called “agile” and “ATDD“.
The only controversial statement of the day was in Jeff Payne’s opening keynote, where he made the observation that if testing was considered a position senior to coding, then the quality team would be able to mentor the development teams on quality throughout the entire development process. He recommended that testers who don’t know how to program learn how, if only so they can speak the language of development and better communicate with the engineering teams, which lit up twitter with disagreement.
He’s spot on, of course. It makes no sense to consider test team positions as junior to development when they’re just as important to the success of the project, and in fact require more experience. I didn’t know it at the time, but Sun had it all right 20 years ago: the testing and documentation people were integrated into project teams from its inception, and were respected as critical to its success. Not only that, but no project could get far without an extensive review from a separate quality team that was made up of the most senior engineers who had proven themselves experts by their peers in other companies. You heard right: the very best engineers, as defined by others in their fields, were the ones who provided feedback on quality.
My favorite presentation of the day, though, was from Alan Page of Microsoft for his talk Stop Guessing How Customers Use Your Software. He was the only person I saw today who knew how to give a presentation with Powerpoint. It was exactly as Tufte described: a presentation should be a compelling story told well, and the slides should only reinforce the story, not distract from it. The slides Page did show were information graphics that illustrated his points wonderfully.
The main point of his talk, hammered home with one effective demonstration after another, was that its impossible to guess how users interact with your software, and we should use actual metrics instead of guessing. The most effective example was a group exercise where he challenged the audience to pick the most effective GUI design in an A/B comparison. Out of the hundreds of people there there only five got the first two comparisons right, and no one got all three.
Michael Czeiszperger
Founder, Web Performance, Inc.
1 Comment
13 October 2010 Alan
Hi Michael – I’m quite late to respond to this post, but I just discovered the kind words and wanted to say thank you. I had a good time giving the talk, and I’m glad that you enjoyed it.