[hdjjhhvvhga]

I read the article and while the approach seems reasonable, in real life it doesn't always work.

My colleague tried a similar method recently and the result was a total disaster. In stage I, collecting the inputs, he gathered everybody's opinions. In stage II, consensus, everybody was happy. Why? Because person A said we should do A, person B said, "Fine, but we should also do B", person C said, "that's perfect, but we must also do C". The result was a huge monster that everybody thought was possible to realize, except person Z who was responsible for QA and knew very well it's going to fail. Now, in phase III, obviously problems started to pile up. Person Z was overworked and in the end he left the company.

So, basically you need to be smart and know when it makes sense to use the methodology and when it makes no sense. As a rule of thumb, if you work with a small group of smart people who also have a rough idea of what others are doing, it's worth taking the risk.

[Jtsummers]

In your example there is no consensus as Ousterhout would define it. From the article:

“The second possible outcome is that the poll did not produce a clear consensus. For example, an 80-20 split is a fairly clear consensus, but 60-40 is not a consensus. A 5-2 vote seems on the surface like a consensus, but if all of the votes in the majority were from the Engineering team and both of the votes in the minority were from Marketing then there is no consensus: there's a departmental conflict. If the consensus isn't clear to everyone in the room, it's better to follow the steps below than to pretend that there was a consensus.“

QA, in your example, hasn’t consented. Therefore no consensus.

It’s important to distinguish between consensus and majority rule. They are not the same concepts, though seemingly similar.

[hdjjhhvvhga]

I'm not sure. There was just one representative from each department/team, and there were around 20 of them. Only one objected. So if "80-20 split is a fairly clear consensus", 95-5 is even more so.

I remember this story as the approach seemed quite reasonable to me initially. However, the weak point in this case was that people cared only about only a limited set of items, mostly related to their work, and were unable to grasp how the aggregation of everything agreed upon would influence the outcome.

That's the point where a good manager should step in and say something like, "B and C, I agree that what you propose is important, but we can't possibly realize everything everybody is asking for. Therefore we will have to consider your requests on another occasion."

[Jtsummers]

> if all of the votes in the majority were from the Engineering team and both of the votes in the minority were from Marketing then there is no consensus

This is the critical part. If the QA objects there is no consensus. Doesn’t mean management won’t decide in favor of the others, but no consensus has happened if a major group objects.