Zach's Blog

Bugzilla Team Evaluations: The Fallout Continues

Update: At the time of posting, I did not realize that the evaluations were posted to the reviewers list and not the developers list. This changes things somewhat although my points still stand. On the other hand, since the discussion spilled over into the developers list, I'll keep this post here for now. I'd rather not clutter up the reviewers list with more meta-discussion since it would just take away from the main goal at the moment: the releases. On that not, a new version of Bundle::Bugzilla will be on it's way soon.

Note to self: do not ever post pictures to my blog again. My bandwidth usage for those couple days on zachlipton.com, where the photos were posted, (thanks to planet.mozilla.org) was 3 times what I had in all of November.

Well the flame war continues over the Bugzilla evaluations (see my above post). Thankfully, Gerv tactfully arranged for a truce so actual work can happen. Personally, I think the idea of looking at our past practices and trying to improve is a noble goal, but there were some problems with the approach:

• Too many evaluations: Criticism is a useful tool, but it needs to be given in moderation. There were so many evaluations that it became a constant wave of criticism. In such a situation, the only possible outcome is for everyone to get defensive, which is exactly what happened. It would be better to focus on specific problems one at a time and pick your battles for what is important instead of releasing a torrant of criticism in one big wave.
• Too personal: The evaluations generally seemed to focus around specific people, especially Dave. In a project such as Bugzilla, where everyone involved is a part of the project because they want to be, not because they are required to be as part of their job, public evaluations that focus too harshly on one person just aren't a good idea. If someone has a specific piece of advice for a specific person, that's a matter for a private email. Save the mailing lists for issues that relate to the team as a whole. Also, as we have learned the hard way, the evaluations shouldn't blame people. It would certainly be better to present the situation, give some examples, and then propose solutions rather then to blame individual people for the problem. Make them evaluations of situations, not people. A good start would be not to indicate the person being evaluated in the subject line of the email, it sets the wrong tone.
• Too little self evaluation: Self evaluation is perhaps the most useful and natural tool known to mankind. It's one of our most valuable tools, a caveman tries to move a rock up a cliff by shoving it upwards, finds that it isn't working, then reflects on his action and takes steps to improve it. Without self evaluation, we're just doing the same thing over and over again, hoping for a different result. In other words: insanity. The Bugzilla evaluations end up being completely focused on external criticism, putting the recipiant of the evaluation on the defensive. As such, there's no real opportunity for improvement, the purpose of these evaluations, since the evaluee never really tries to adjust and improve.

These evaluations are a useful tool and really unique in open-source projects and, with a larger dose of tact, are a good idea to continue as part of a constant rethinking of processes and actions. At the same time, something a little less personal and a little more situational and progressive would be certainly a more useful and less provocative tool.

Posted by zach at January 8, 2005 6:24 PM