Entropic Thoughts

Unburden yourself of the backlog

Unburden yourself of the backlog

unburden-yourself-of-the-backlog.jpg

David Marquet was a submarine commander who was abruptly assigned to work on a boat he had no training on, and no idea how it worked. To compensate for his lack of technical knowledge, he adopted a transparent leadership style, which allowed the competence of the rest of the crew to fill in for his gaps.

During this process, he discovered that transparent leadership produced outsize performance improvements compared to the more traditional leadership style of the Navy. You should read the book1 Turn the Ship Around!; Marquet; Penguin Books Ltd.; 2001. he wrote about this experience. It’s well written and says many good things.

The boat-wide tickler

There’s one thing in particular I keep coming back to all the time.2 And I want to reference it in a draft for a future article, so I’m sharing it now. In the following transcription, I have snipped out some unrelated flavour to get to the core of the chapter.

On board Santa Fe (152 days to deployment)

In addition to observing the ship perform its tactical manoeuvers, the inspection team also looks at administrative issues. In this case, they identified that Santa Fe had not responded to several messages we owed to the squadron, subpac, and the maintenance facility.

I asked the xo about the missed items, and he brought out the “tickler”, a three-inch binder maintained by his yeomen that had all of the messages – such as this one – that we received. They were sorted by department and due date. He looked in the book and proudly reported that, sure enough, we were tracking this particular message and knew we hadn’t responded.

So we had a system that was focused on understanding the status instead of actually getting the work done. Unfortunately, everyone was too busy to look at the binder, and in any event, it was stuck up in a locker in the xo’s stateroom. Like every other submarine we would have weekly tickler meetings where the department chiefs and department heads would sit in the wardroom for an hour or more going through the binder page by page. Of course, none of this activity actually resulted in getting any of the work done; it simply allowed us to catalogue what we were supposed to do, and what we were delinquent on. It sucked up a lot of time, valuable supervisory time.

This was how everyone did it, always had.

There is no requirement to maintain a tickler, only a requirement to get the work done. We set out to invent an even more efficient way to do it, a new way.

The discussion went like this.

“Weps, who’s responsible for your department?”

“I am, Sir.”

“Not the xo?”

“No.”

“Then why should he spend time keeping a tickler for you and have you sit in all these agonising tickler meetings?”

“He shouldn’t.”

“Okay. But here’s the deal: you guys need to get the work done.”

“We will.”

Weapons Officer Chief David Steele reminded me that his department had worked through the night while we were in San Diego so we could participate in the battle group-wide exercise. What I didn’t know was that Chief Steele had box seats at the Padres game and had given those up. No one told him he had to do it or ordered him to do it. It was needed to support the mission and he just did it.

“So here’s what we are going to do. You are all going to monitor your own departments and whatever is due. You are responsible, not me and not the xo, for getting it done.”

And with that, we unburdened ourselves of the effort of maintaining the tickler.

This had two advantages. First, it would be more efficient because the work would be getting done without the overhead of maintaining the tickler and those darned tickler meetings. Second, there would be no illusion about who was responsible for the performance of the various departments: the department heads were!

Sure, I was worried that a lot of stuff would slip through the cracks and Santa Fe would get a reputation for not getting the work done, but that didn’t happen. I won’t say that we never again received a message zinging us for not reporting something, but they were easily remedied and not that important. What was incredibly powerful was the idea that everyone was responsible for their own performance and the performance of their departments; that we weren’t going to spend a lot of effort telling them what to do.

Marquet goes on to discuss the power of self-directed work as a way of instilling a sense of ownership, and the importance of retaining monitoring but without accidentally transforming it into control. Those things are important, but I got stuck on something earlier in the chapter: when I read about the tickler meetings, I was thrown back into all the backlog refinement meetings I had sat in.

Why backlog refinement meetings are junk

Most organisations I’ve been a part of have had some form of backlog refinement meeting. Here’s what’s happened in those meetings.

  • Mostly, the team just stares blankly at a projected screen as one person drags issues around to adjust their priority order.
  • Sometimes, the team listens to the product manager interviewing one programmer to clarify requirements on a task they would be assigned.
  • A vanishingly rare occurrence is the team actually discussing assumptions about users, and having productive conflicts on desgin. This is so rare it doesn’t pay for the cost of the meetings. It would be better to call a short meeting specifically for those tasks where these discussions are needed.

To stave off the obvious argument, I’m not saying prioritisation is bad. I am, however, saying the only level of prioritisation that truly matters is “What’s the next most important thing to work on?” Any relative ranking beyond that is not providing value right now, and risks going stale by the time it would. I’m also not saying the team shouldn’t get an opportunity to interject when the priorities are incorrect, but they can do that without the meeting.

I’ll grant that it’s possible the team learns something by listening in to the one-on-one interviews, but I think there are more efficient ways to learn those things, such as writing them down in a brief decision record.3 This also makes it possible to later answer the question, “Why did we do it this way, again?” which is often very hard to answer a few years into a project.

Life without a shared backlog

I firmly believe that, just like a boat-wide tickler wasn’t necessary on the Santa Fe, a backlog is not necessary in a software product development organisation. An organisation that works in sprints may want to have a shared plan for the current sprint, but that’s a small, bounded list of tasks. There’s no reason to have a shared backlog that stretches beyond that.

What matters is knowing at all times what the next most important thing is. That tends to already be fresh in the heads of the people who think about those things. It doesn’t need to be written down and ranked relative to 50 lesser ideas for improvements.

To be clear, this doesn’t mean individuals can’t maintain their own wishlists. If Carlos wants to keep a personal backlog to organise his ideas for improving the product, I won’t have an opinion on that. He can do his job the way he finds most efficient for him. What I disagree with is the idea that Jennifer, Kenji, Mohammed, Amy, Pierre, Raj, and Lisa shall all be forced to spend their valuable minutes gawking at Carlos’ wishlist.