Via
InfoQ ,
Jason Yip
Tobias Mayer recently described
some successful adaptations he's made
to Scrum .
Jason Yip summarised:
- Don't separate the technical and non-technical members of the team.
- Shorten iterations (1 month is too long) and focus on end-to-end cycle time.
- Use smaller tasks instead of task estimation.
- Make tracking visible in the workspace.
- Quality engineering practices are not optional.
- Process leaders should make themselves unnecessary.
You can't beat 1-week iterations . They force everything to be smaller. Complexity is broken down by smaller user stories and they can be estimated with less inaccuracy. A great visual indicator of progress is a steady rise in running tested features as stories hit the done column every day. Planning games are shorter because you do less in a week. And by the time you start you need to be thinking about finishing because iteration reviews come fast, so communication is more intense. You get to inspect and adapt every 7 days, and celebrate success 4 times in a month as opposed to once. And this builds momentum .
I don't like planning with tasks because, when user stories are small (typically 1 to 2 days of effort), tasks belong to the developer-pair. They form a 'private' to-do list that reflects how the developers will get the story done. Planning and tracking these tasks is micro-management.
These adaptations do not compromise agility . They amplify it.
I'm not sure that the Scrum Master role isn't always necessary, though. I agree that the coaching/process-leader role can become redundant when a team achieves the ability to self-organise. But the facilitator role that the Scrum Master also fulfils remains important even within a self-organising team .
