It’s been ten weeks since we started our 20% time experiment at Atlassian, and about eight weeks since the first time someone from marketing asked me “Have we done anything yet?”
I think that among some people there was an expectation that we would just turn on the 20% spigot, and out would come a steady stream of new ideas and implementations. Of course, that’s not going to happen. 20% of ten weeks gives each developer at most ten working days, and with the effort of having to fit them around already-tight release schedules, I doubt anyone’s used up more than half their allotment so far. Most teams were a month or so into the programme before they could spare any significant developer time.
Still, we’ve had some successes already. Two different projects are already deployed on our internal Jira instances, ready for inclusion in the product’s next major release. With that in mind, I sent out a quick survey to a few of the developers taking part.
*Why did you pick the project you are working on?*
Some representative responses were:
bq. “Pure selfishness. I chose to implement [something] I personally wanted to use.”
“While I had a long list of potential new and exciting ideas I’ve been really depressed lately about the body of work that I’ve left incomplete. I’ve decided that I’m going to use my time to actually deliver some of the things that I’ve started in the past that never made it all of the way.”
“Wanted to write an application for the iPhone but haven’t found a good reason. Now I have one.”
“I proved the concept with my ShipIt entry, garnered customer interest, and made a commitment that I’d produce something.”
*How much time have you had to work on it? How much time do you think is left?*
This survey was sent out after we had officially had eight weeks of 20% time, which would mean the theoretical maximum any developer might have spent on a project was eight weeks. In fact, almost all the responses were in the 2-3 day mark, and none higher.
Time remaining was more varied. Some developers had chosen to start with projects that could be completed in a couple of days, others were more ambitious.
*Why do you think that working on your 20% project is more important than doing your scheduled dev work?*
To me, this was the most interesting question, which is why I hid it in the middle of the survey. It was also the biggest surprise. Only two developers believed that their 20% project was as important as their scheduled development work:
bq. “I think I got good bang for buck. I spent very little time to improve Jira in a way I knew was useful.”
“I don’t think it is more important, just equally important but to a different set of stakeholders.”
Interestingly enough, both of these projects were the ones that are scheduled to be included in the next major release of Jira. Still, this didn’t mean that the remaining developers thought they were wasting their time entirely:
bq. “I believe that 20% time is very important though; I don’t like the way we’ve stagnated a little in the innovation and “scratch that itch” stakes”
*How easy/hard is it to find time to do 20% amongst all other work?*
As we expected, the biggest problem for developers finding time to work on their projects was fitting it around their scheduled work:
bq. “We are really trying to get a feature done on time. So far its going really well but […] I kinda feel like I would rather keep forward progress on my feature”
“…we made the decision to delay 20% time until after we shipped 2.8. It hasn’t been that hard to take time when I’ve wanted since then.”
“Our team leader has embraced 20% time, and there is no difficulty in taking it. […] The feature I am currently working on is running on time, so this probably helps. If we were running behind time, it may be different.”
A common refrain was that the smaller the team a developer was on, the harder it was to find time:
bq. “Pretty hard for me lately. The customer comes first.”
“Very difficult on a small team like Bamboo. It was easy during the post-2.0 lull. But pretty much impossible now, because I want to be involved in the planning for 2.1.”
*If you could improve 20% time in any way, how would you?*
The most open-ended question. Most developers didn’t feel they’d had enough time to form an opinion, but of those who did:
bq. “My only concern is that a 6 months trial may turn out to be a rather small amount of time. We are already 2 months into it, and it seems a lot of people are only now starting to put together their ideas of what to do. I would hate to get to the end of 6 months, and go “hmmm…. no-one did anything, lets forget about it”.
“Mandate that particpants have to spend 2 days a fortnight on it, otherwise it’s difficult to keep the pace up to 20%.”
“One thing I fear in the Jira team at least, is that a lot of the projects may end up being “engine-room updates”, or code-cleanups. This should not be considered a failure of 20% time, but more a reflection that our normal scheduling doesn’t allow enough time for these type of changes.”
“I like it so far. I think we should make it permanent.”
Next time… Five questions for the team leads.