Something Has to Stay Still

I have just participated in a failed project, and this post is the lesson learned.

Essentially, we faced a complex task that required assembling a group to build a deliverable. The details of the deliverable are not important. What is important is that the main group was around ten people, each responsible for their own field of expertise, and that it was the first project of this kind being delivered.

The project lead attempted to use "divide-and-conquer". They split the project into smaller chunks, let people work on the area they were best at, and hoped that the sum of those parts would be a deliverable. The project failed in a spectacular way. First of all — it was more expensive than the sum of the parts, because, as we discovered, there were overlaps, which each expert covered independently, and the same common parts were handled differently, setting up the team for conflicts. At the same time, the deliverable was not enough, as some areas were not addressed properly — nobody considered them "theirs". And since it was the first project of this kind, those areas were not spotted, because the lead had no experience. Besides, "divide-and-conquer" used to work.

And underneath all of that, the communication broke down completely. People going silent, avoiding any responsibility, questioning each other's competency all the time.

As I was reflecting on what really happened, I was reminded of the distinction between a team and a crew. Someone brought that topic to me a while back — and given this failure I just experienced, I find it to be a very underestimated subject.

TL;DR: Team is a group of people that figured out how to cooperate with each other. Crew is a group of people relying heavily on roles and procedures to sort out repeatable work.

Three arrangements, and only two of them work

Holly Arrow, Joseph McGrath and Jennifer Berdahl sorted working groups by a question that sounds too simple to be useful: what stays constant? In a team, the people stay and the work changes — you keep the same five and point them at whatever comes next. In a crew, the work stays and the people change — the roles, the tools and the procedures are fixed, so anyone trained into a slot can fill it. Both of these function, but for very different reasons.

After Arrow, McGrath & Berdahl (2000)

There is a third setup, which is the one we were in, and it isn't a weaker version of either. It's people who have no shared history doing work that has never been done before. This setup reliably creates the symptoms I experienced, and it creates them in groups of competent, well-intentioned adults, which is why everyone involved goes looking for a personality to blame.

Dividing could work for a crew. With defined roles and procedures, everyone would know what to do and where and how to communicate. A crew can do that because someone spent years defining where one job ends and the next begins. We had no such luxury.

What the silence actually was

Once you split work without splitting roles, the gaps between the pieces belong to nobody, and everything that goes wrong goes wrong there.

Then a second thing happens, and it's the one that felt personal at the time. A long-standing team carries a rough directory of who knows what — Ray Reagans, Linda Argote and Daria Brooks call it knowing who knows what and knowing how to work together, and they found it contributes to a surgical team's performance separately from anyone's individual skill. An ad-hoc group has no directory. So when someone goes quiet, you have no way to read it. In a team you'd know that's just how he works when he's deep in something. Here you ask a question, and the answer tells you nothing, and you conclude... he's lost.

That conclusion is cheap to reach and expensive to hold, and it runs in both directions at once. Everybody is being read by people with no data, and everybody can feel it, so everybody says less, which leaves even less to read. Nobody is hiding. There is just nothing left to go on.

And raising it is worse than saying nothing, because in a group with no shared history you have no track record to spend. "This isn't working" lands as a mood rather than a dependency — the difference I've written about before — and the person who says it becomes the problem, so the people who can see it clearest are the ones with the strongest reason to keep quiet.

The responsibility-avoidance is the same shape. When the seams have no owner, refusing to own one is not evasion. It is the correct reading of the arrangement. It is the structural point I keep coming back to: if you find yourself explaining a group's behaviour by the character of the people in it, look at what the setup was paying them to do.

Tackling "first of a kind" projects

Not divide and conquer. But a time-expensive, collaborative session to create an end-to-end vision of how the deliverable will be built, where the real responsibility limits are, where the unknowns are — and how communication will look during unknown discovery. So simple... and so... against corporate modus operandi.