[{"data":1,"prerenderedAt":18},["ShallowReactive",2],{"post-one-mind-repeated":3},{"slug":4,"title":5,"seoTitle":6,"excerpt":7,"content":8,"tags":9,"readingTime":15,"publishedAt":16,"published":17},"one-mind-repeated","A team of similar people is one mind, repeated, or: an open question I have not closed","One Mind, Repeated","Should a team be built out of people who resemble each other, or out of people who do not. My lean is toward the second, and this piece is mostly an attempt to work out how much weight that lean can carry. The literature turns out to agree with half of what I believed and to push back fairly hard on the other half, which is the more useful outcome of the two.","\n## The question\n\nThis is a shorter piece than usual, and it ends without landing, because I have been trying to answer something for a few weeks and have not managed it.\n\nShould a team be built out of people who resemble one another, or out of people who do not.\n\nMy lean is toward the second, and most of what I have behind that lean is one person's experience of a handful of small teams. It is also worth saying at the top that the lean is convenient for me. I have a career path that jumps between fields, so the theory in which range pays is the theory that flatters my own history, and I would rather flag that than have it operate quietly.\n\n## What I already think about the single person\n\nThe older belief, which I have held for a long time and written around the edges of in [[generalists-outperform-on-every-task|the note on generalists]], is that a person should not stay inside one field. Some competence in several, real depth in more than one.\n\nWhere my version goes further than most versions I hear is the part about distance. The size of the gap between the fields is the variable I care about. Two adjacent technical disciplines make somebody more useful; a technical discipline and a literary one seem to do something different in kind. Sitting with an equation, reading a second language closely, writing code, looking at pictures for long enough to know why one works — these leave different residues in a person, and my claim has always been that the residues interact rather than stack.\n\nStated at full strength: the wider the span between the fields a person holds, the larger the change in how that person thinks. Not more knowledge. Different thinking.\n\n## Where the strong form breaks\n\nThe strong form is where the evidence stops agreeing with me, and I would rather write that down than skip past it.\n\nTransfer between domains is the most studied version of my claim, and the finding has been stubborn for a century: near transfer happens, far transfer is rare and unreliable. What somebody learns in one field does not walk over into another on its own. Dedre Gentner's work on analogy sharpens the disappointment considerably — what makes an analogy valuable is structural similarity between two situations, but what makes a person actually retrieve an analogy from memory is surface similarity. Which means the distant analogy, the one that would be worth the most, is precisely the one least likely to arrive when it is needed.\n\nSo the leap is not automatic, and I had been treating it as though it were. Range supplies material. It does not supply the mapping, and the mapping is a habit rather than a possession — the habit of stopping on a problem and asking what else it resembles. David Epstein's *Range* makes the case for breadth well, and the qualification I would attach to it is that breadth without that habit is a wider library nobody walks into.\n\nThe honest revision, then. A wide span raises the ceiling on what a person is able to connect, and does nothing on its own to raise the floor.\n\n## From one person to a team\n\nA team is the cheap version of the same idea. Nobody has to spend six years acquiring a second field if somebody two desks away already has it, and the wiring between two people is a conversation rather than a decade.\n\nThis is where the argument for difference gets strong. A team of similar people is one mind repeated at several times the salary. Repetition buys throughput, and past a point it buys very little else, because it also reproduces the same blind spot in every seat.\n\nThere is a distinction here I want to keep straight, because I have argued the opposite-sounding thing elsewhere. In [[redundancy-is-not-waste|the note on redundancy]] the claim was that slack is not waste, that the good outcome is the tail of ninety-nine failed experiments. That is redundancy of time, and it is fuel. Redundancy of perspective is a different object wearing the same word, and it is duplication. Spare hours pay. Spare copies of the same judgment do not.\n\nOne more link, to the piece I wrote a day ago. A hiring seat with nothing pushing back on it drifts toward candidates who resemble whoever is sitting in it, which is the argument I made in [[the-seat-closest-to-the-rules|the HR piece]]. Homogeneous teams are mostly not chosen. They accumulate, one reasonable-feeling decision at a time, and by the time anybody notices, the composition looks like a preference rather than a residue.\n\n## What I suspect the real variable is\n\nHere is the part I am least certain about, and it is the part that keeps the question open.\n\nI started this thinking about personality, and I no longer think personality is the thing I mean. The evidence on personality composition points somewhere slightly uncomfortable: what predicts team performance is the team's *mean* on conscientiousness and agreeableness rather than its variance, and on those traits variance tends to hurt, because a single careless or combative member is enough to cost the group. Barrick and colleagues found the weakest-link pattern fairly clearly. A team of clashing temperaments is not the good version of the thing I am arguing for.\n\nWhat I actually want is difference in cognitive kit — the perspectives people bring, the heuristics they reach for first, the way each of them represents a problem before starting on it. Personality was standing in for that because personality is visible in a week and a cognitive toolkit takes a quarter to see. Having spent several pieces here complaining about proxies, I notice I reached for one at the first opportunity.\n\nThe second condition is the one I keep underrating. Difference only converts into anything where disagreement is cheap. In a room where saying I see it differently costs something, variety turns into silence plus friction, and a homogeneous team will beat it comfortably. The diverse team is not better by construction. It has a higher ceiling and a lower floor, and which of the two you get is decided by the manager rather than by the hires.\n\n## The principle, stated tentatively\n\n> A team of similar people is one mind, repeated, and repetition is not resilience. But difference pays only in a room where disagreeing is cheap, and building that room is the manager's work, not the new hire's.\n\n## What I am not saying\n\nI am not saying hire for contrast as a rule. Assembling opposites and hoping the friction produces something is not a method, and I have seen the failure mode of that too — a team where every conversation restarts from first principles and nothing reaches a decision.\n\nI am not saying run personality tests. The evidence I trust least in this whole area is the evidence behind the instruments companies buy for this purpose.\n\nI am not saying difference is free. It has real coordination costs, and the honest reading of the literature is that varied teams are slower to agreement and better at the answer, which is a trade rather than a gift. On an easy problem the trade is bad.\n\nI am not saying my own range proves the point. One person is not evidence, and a conveniently self-flattering theory deserves more scrutiny from the person it flatters, not less.\n\n## Where this connects\n\nThis sits next to [[generalists-outperform-on-every-task|the argument for generalists]] and underneath [[the-two-image-week|the two-image week]], where one of the four principles was to keep role boundaries deliberately blurred. I would now say that piece was making a version of this argument without noticing: blurred boundaries work because they let different kits meet on the same problem.\n\nIt also connects backwards to [[the-seat-closest-to-the-rules|the hiring seat]]. Whatever the answer to the question turns out to be, the mechanism that decides it in most companies is not a strategy about team composition. It is one person's sense of who feels like a good fit, repeated forty times.\n\n## What the literature already says\n\n**Scott E. Page, *The Difference*.** Page gives the mathematical version: on hard problems, a group of people with diverse perspectives and heuristics can outperform a group of individually higher-ability people, because the strong performers tend to share the same approach and get stuck in the same places. What I take from it is that this is a theorem with premises attached — the problem has to be hard, the diversity has to be relevant to the problem, and the group has to be large enough. It is not a slogan about mixed teams being nicer.\n\n**Meredith Belbin, *Management Teams: Why They Succeed or Fail*.** Belbin's Apollo teams were built from the most analytically able people available, and they lost, repeatedly, to ordinary mixed teams. They spent their time competing to occupy the same role. The finding is about role composition rather than personality in the loose sense, and it is the closest thing I have found to a direct answer to my question.\n\n**Arthur Koestler, *The Act of Creation*.** Koestler's bisociation is the best account I know of why the span matters: a new idea arrives when two internally consistent frames that are not habitually connected get held against each other. That is my belief about cross-domain knowledge, stated forty years before I arrived at it and stated better. Frans Johansson's *The Medici Effect* is the readable trade version of the same claim.\n\n**Anita Woolley and colleagues, on collective intelligence.** Their finding is that a group's performance across varied tasks is predicted less by the average or maximum intelligence of its members than by social sensitivity and by how evenly the group takes turns talking. Read next to Amy Edmondson's work on psychological safety, it says the same thing my second condition says: the composition sets the ceiling and the conversational conditions decide whether anybody reaches it.\n\n## A postscript\n\nI know roughly what would change my mind, which is a better place to end than a conclusion I do not have. If I watch a varied team lose to an aligned one on the same problem, and the gap turns out to be the quality of the answer rather than the speed of agreement, then the ceiling argument is wrong and I should update hard. So far what I have seen is the other pattern — slower to agree, better once agreed — but I have seen it in small teams doing creative work, which is the friendliest possible setting for my own conclusion.\n\nI am still watching this one.\n    ",[10,11,12,13,14],"Team Design","Org Design","Creative Process","Process","Note to self",7,"2026-09-02",true,1788356310713]