Who has worked with whom
What counts as having worked together
The bar has to be set once and applied to every pair. This is the bar used on this site, the cases that sit on the line, and why each one falls where it does.
The people this page is about, and which of them have ever worked together
- 01The team's managerThree of this person's five working pairs
- 02The senior engineer
- 03The second engineer
- 04The designerShared with two other teams
- 05The QA engineer
five people · 10 possible pairs · 4 of them have worked on something together · 2 have only been in the same meeting · 4 have never done either
Five of the eight, and only the pairs that run between those five.
Home / Who has worked with whom
To keep measurement transparent, teams can open the official information and agree together which information is useful, who can see it, and how long it should be retained.
For a separate perspective on teams and working practice, consult RAND.
The count is only as good as the rule, and the rule has to survive twenty-eight applications without drifting. This note sets out the one used here in enough detail to be reused, including the awkward cases.
The bar: one piece of work, inside the period, that needed both people, and that would have come out differently if either of them had not been there.
Three clauses, each doing a job. "A piece of work" excludes social contact. "Needed both" excludes one person watching another work. "Would have come out differently" excludes the formalities — the approval that was never going to be withheld, the notification copied to somebody who read it and had nothing to add.
Clear yes
A review that changed something. If one person read the other's work and the work changed as a result, that is a pair, and it is the single most common way pairs form in practice. The change can be small. It cannot be nothing.
An afternoon on one problem. Two people, one screen or one call, one thing neither of them had solved. Even if it was not solved.
A document with two authors in substance. Not two names on a cover page: two people who argued about what it should say.
A handover that required a conversation. Somebody took over somebody else's work and had to be told things that were not written down. The conversation is the work.
A question that took real effort to answer. Twenty minutes of somebody reconstructing why a decision was made counts. It transferred something and it cost them something.
Clear no
- Being in the same meeting, however many times. This is the whole argument of this site compressed into a sentence, and it is covered in its own note.
- Being on the same project. Staffing is an arrangement; it predicts pairs and is not one.
- Being in the same channel, thread or document. Proximity in a tool is not an exchange.
- Approving something without reading it. Common, usually harmless, and not a pair.
- Social contact of any amount. The quiz, the coffee, the virtual lunch. These may make the first real pair easier. They are not it.
On the line
A one-line review comment that was acted on. Yes, if the comment changed the work. The size of the contribution is not the test; the test is whether the output is different.
A question answered in thirty seconds. No, as a rule. Something that cheap did not require the asker to know anything about the answerer and does not leave either of them knowing anything new. Three such exchanges a week between the same two people is a different matter — that is a working relationship arriving in instalments, and it should be counted.
One person teaching another. Yes. Teaching is work that needs both people and comes out differently depending on who is in the room.
An incident both attended. Only if they did different things that depended on each other. Six people on a call watching one person fix something is one pair at most.
Pair programming or any deliberate pairing exercise. Yes, obviously, and worth noting because these are the only entries on the list that can be scheduled.
Work done through an intermediary. No. If two people each worked with the manager and never with each other, that is two pairs with the manager, and recording it as three is exactly the error that makes a wheel look like a network.
On the period
A pair that was real two years ago and has done nothing since is not drawn. This is the rule most people want to bend, because the relationship still feels present to both of them, and in the sense of trust and goodwill it probably is.
What has gone is the currency. They no longer know what the other is working on, what has changed in their area, what they now care about. The pair will re-form faster than a new one would, which is a real advantage, and it is not the same as existing.
Who adjudicates
The two people in the pair, and nobody else.
This sounds like a small procedural point and it decides whether the number means anything. If the person compiling the grid adjudicates, the bar drifts toward what they can remember, which is the work they were near. If the two people decide, the bar drifts toward what they experienced, which is the thing being measured.
The useful consequence is that a disagreement settles itself. If one person says a pair happened and the other does not recognise it, the collaboration was one-sided — somebody's work was used and the other person did not know they were in it — and that is not a working pair in any sense that will help next quarter.
Writing the rule down
Whatever you decide, write it in one paragraph at the top of the grid, and leave it there. The second count, three months later, is only comparable to the first if the rule did not move, and nobody — including the person who made it — remembers a rule that was never written down. Most of the counts that turn out to be useless are not wrong. They are incomparable.