Strokes

The problem it solves

There are teams where nobody says very much, where there are no arguments and no drama, and where the work goes out on time. People tend to leave those teams after a year or two, and the exit conversations keep producing the same sentence, which is that they never really knew whether they were doing all right.

Most organisations also contain somebody who is in a disagreement almost every week, who is exhausting to work with, and who is also by some distance the most engaged person in the room.

Neither pattern looks like a recognition problem, because nobody involved is complaining about recognition and nobody has ever asked to be praised more. The quiet team is having a problem it cannot name, and the difficult colleague has solved that same problem in the one way available to them.

There is a third pattern as well, and it is the one most managers who read about this end up creating, which is praise given constantly, warmly and honestly, and which after a quarter has stopped meaning anything to anybody.

The idea

Berne's starting claim was borrowed from mid-century infant research and stated a good deal more strongly than the evidence really allowed (it's the shakiest claim on this page, and the one most often quoted). Recognition, on his account, is an appetite in the same family as hunger, and people deprived of it deteriorate. You don't have to accept the biology to find the managerial consequence useful, because people who are not being noticed go and arrange it.

He called the unit a stroke, meaning anything that acknowledges another person exists, which covers a greeting, a name remembered, a nod across a room, a piece of praise and a piece of criticism.

Criticism counts, and that is the part that makes the idea worth something, because if recognition is a need and criticism supplies it then negative recognition beats none at all. A row is proof you are there and silence is not, so the engineer arguing in each planning session may be running the most reliable supply on offer, while the team that never argues may be starving quietly.

Sort recognition on two axes and it stops being one thing.

Enlarge: The same subject, four different objects, and the cell in the bottom left is the one with no legitimate use.

The same subject, four different objects, and the cell in the bottom left is the one with no legitimate use.Close

The same subject, four different objects, and the cell in the bottom left is the one with no legitimate use.

The vertical axis is the obvious one, positive or negative. The horizontal axis is the one that does the work, and it asks whether this is attached to something the person did or to who they are.

Attached to an act, recognition carries information, so it can be repeated on purpose or avoided on purpose. Attached to the person it carries none, and there is no action it points to in either direction.

That gives four things that behave very differently from one another, and one of them has no legitimate use at all. Criticism of a person is a verdict with no available response, whereas criticism of an act is most of the difficult half of a manager's job and belongs in it.

Claude Steiner extended this in 1971 with an argument about scarcity, which is that the shortage in most organisations is manufactured rather than natural, and maintained by unwritten rules against giving recognition freely, asking for it, accepting it when offered, or granting it to yourself. He called the result a stroke economy, and whatever you make of the politics he attached to it, the five rules remain the best description available of a silent team.

How to use it

Separate the two axes when you give difficult feedback. This is the whole technique and it takes about a second. "This team has got sloppy" and "the last three releases went out without a rollback plan" describe the same failure, and the first is a judgement about people that can't be answered while the second is a fact with an obvious next step. The second is more specific, not softer, and specificity is what makes criticism survivable.

Give the unconditional positive kind more often than feels right. It is the one that gets rationed, because it feels like something that ought to be earned (it isn't). It costs the manager almost no effort, it isn't praise for performance, and it is the thing that tells somebody they are not on probation. Asking how the weekend was and then listening to the answer is the entire mechanism.

Attach the positive conditional kind to something specific, or do not bother. Praise with no object decays into the noise a meeting ends with, and within a quarter it stops registering, while the person receiving it will be unable to answer the one question they have, which is whether any of this is working.

Then run the audit, which is the uncomfortable part. The question is who has had nothing at all this month, rather than who is difficult, because recognition follows attention, attention follows visibility, and visibility follows trouble. The quiet, the remote, the part-time and the reliable are the people a manager can go an entire quarter without addressing, and they are the last people who will mention it.

Where it breaks down

Every concept here has one. It is the section most summaries leave out.

The taxonomy is a teaching device that has never been tested as a classification. Nobody has shown that the four kinds behave as four distinct kinds, and fifty years of transactional analysis research has largely left the stroke alone as an empirical question. The nearest independent support is Mueller and Dweck's 1998 work showing that praising a child's intelligence and praising their effort produce measurably different behaviour after failure, which lines up with this distinction closely. It should be cited with its own caveat, since it belongs to the mindset literature, parts of which have had a difficult time in replication, and it is about children.

"Negative recognition beats none" is the most over-applied idea here. It licenses a comfortable conclusion, which is that the difficult person is starved rather than difficult, and that is sometimes true and sometimes an elaborate way of avoiding a conversation. The idea explains why an unrewarding behaviour persists without excusing it, and using it as a substitute for addressing the behaviour is worse than not knowing it at all.

It converts into a programme almost immediately, and the programme kills it. Recognition works because it is evidence that somebody noticed, and a weekly prompt instructing people to notice removes the evidence while keeping the words, which is why appreciation channels are enthusiastic for three weeks and embarrassing by week six. Anything that works when it is meant and stops working when it is required cannot be installed as a process.

Register and culture are doing more work than the model admits. The unconditional positive stroke that lands warmly in one team is intrusive in another and faintly alarming in a third. Berne's account does not address that, and its confident universality is a product of where and when it was written.

Steiner assumes all restraint is artificial, and some of it is proportion. A manager who finds everything excellent has removed the signal rather than liberating it. Scarcity that reflects a genuine distinction between good work and ordinary work is what makes recognition mean something when it arrives, which makes it a poor candidate for a rule to be broken.

It hands you a way to diagnose a colleague instead of answering them. Deciding that somebody is fishing for strokes turns a request for recognition into a symptom, and a symptom does not have to be met. They cannot argue their way out of it either, since objecting to the diagnosis reads as more evidence for it. The useful version of this idea describes what you are withholding rather than what they are after.

In one line

Attach criticism to the act and never to the person, give the unattached positive kind far more often than feels natural, and find out who has had nothing at all this month.