Where Human Intelligence Debt and the Attribution Gap arrive at the same place — offered as an observation, not a proof.
Document Status — Field Note · Series: The Attribution Gap and Capability Loss
Tegrity.AI · The Integral Management Society · Iván Abril Palma
This bridges two independently developed lines of work: the Human Intelligence Debt series — an account of why automatable human work persists, rooted in the scarcity of governance decision capacity — and the Attribution Gap theory — an account of why firms destroy the capabilities they most depend on, rooted in the divergence between contribution and credit. The two were not built to meet. This note records where they do, keeps each theory’s formal machinery in its own house, and is deliberately modest about what the meeting proves. The convergence is offered as something worth attention — a coincidence between two roads — not as evidence for either theory. The case for that restraint is the last section, and it is the point of the note.
1. Two roads, two foundations
The two theories begin in different places, and should be allowed to keep doing so.
The first road begins with a question about people: why do capable people spend their working lives doing what a machine could do? Its answer is that automating such work requires a prior frame-defining decision — who owns this, what the target is, what is retired — and that the authority to make such decisions is structurally scarce. Hierarchy concentrates it; entanglement forces it to be centralized; the few who hold it have no time and rationally defer. The binding constraint is not the capability of the technology but the capacity of the organization to decide. Human intelligence is the capacity to make frames; the debt is that capacity, stranded. On this road, the central failure is that the right decision is never made.
The second road begins with a question about value: why do firms cut the capabilities they most depend on? Its answer is that the credit a capability receives — its attribution — drifts away from the value it adds — its contribution — because keeping a dependency implicit protects the rent of whoever sits on top of it, and because a firm prices the capabilities it buys from outside while capturing, unpriced, the ones it grows within. The most depended-on capability becomes the least recognized and the least owned, and is removed, usually disguised as retiring legacy. On this road, the central failure is that the decision that is made is systematically wrong.
Different foundations. One is about the scarcity of decision; the other about the corruption of the value signal that any decision would need. I have kept their formalisms apart on purpose: the recovery coefficient and the contribution ratios of the first do not appear in the second; the contribution deficit and the capture surplus of the second do not appear in the first. Nothing below merges them.
2. Where the roads meet
And yet, walked to their ends, the two roads arrive in the same room, at the same body, around the same evidence. Four meetings are worth naming.
The same victim. Both roads end at a particular kind of capability: the small, load-bearing, high-value thing the organization depends on and cannot properly see. The first calls it the capability stranded beneath an un-made decision; the second calls it the diamond cut because it is unowned. It is the same thing, described from two sides.
The same fingerprint. Both reduce, independently, to one observable: the gap between how many parties claim or depend on a thing and how few will own it — claimants per owner, or in the plainer field version, the room in which everyone affirms a system’s importance and no one answers when asked who owns it. The first theory predicts that gap throttles the decision; the second predicts that gap predicts the cut. They are measuring the same number.
The same entropy. Neither uses the word in its physical sense. Both define it as the indeterminacy of the answerability ledger — the condition of being unable to say, about your own estate, what a thing costs, who is accountable for it, what depends on it. The first calls this lost architectural self-knowledge; the second calls it accountability entropy. The same definition, reached twice.
The same irreversibility, and the same cure-and-trap. Both find that the death happens in the dark, before any formal decision, and that re-architecting afterwards produces a new order, never the lost one. And both reach the same uncomfortable corollary about the remedy: clarity must come from outside the local incentive system, and value must be re-established before clarity is imposed — because clarity alone, resolving who owns what without first re-valuing what is owned, resolves the ambiguity in favour of the cut and accelerates the very loss it was meant to prevent.
3. The bridge, stated plainly
If the foundations differ and the destinations coincide, the bridge is simply this: the two theories describe the two failure modes of a single act — the governance decision over a capability — and the failure modes compound.
The first explains why that decision cannot be made well: there are too few who can make it, and they have no time. The second explains why, when it is made, it is made wrong: the information reaching the few is already distorted, the valuable thing already mis-credited and unowned. Put them together and the picture is worse than either alone. The scarce decision-maker, with two minutes and a single line on a slide, decides the fate of a capability whose value has already been hidden from him by the time the slide is written. The bottleneck guarantees the decision is rushed and centralized; the attribution gap guarantees the input to it is corrupt. The diamond dies twice over — once for want of a decider, once for want of a true number.
That is the whole bridge. It is, by design, an observation about where two independent analyses of the same firms land — not a third theory built on top of them.
4. Why this is a note, and not a proof
It would be easy, and wrong, to present this convergence as evidence that either theory is true. A convergence between two lines of reasoning is evidence only to the degree that the lines were independent in fact — and the temptation, once you notice they agree, is to quietly let each lean on the other. At that point the independence that made the agreement interesting is gone, and what looked like two witnesses becomes one author talking to himself.
So the discipline is to state the convergence and then refuse to use it. The two theories keep their separate foundations, their separate machinery, and their separate falsification programs. Each must stand or fall on its own evidence — the first on whether governance decision capacity predicts persistent manual work; the second on whether dependency diverges from attribution and the divergence predicts the cut. If both survive their own tests, the fact that they were built on different ground and met anyway will begin to mean something. Until then the meeting is a coincidence I find striking and want on the record — no more, and no less.
I record it for one reason. When the same conclusion is reached twice, from different starting assumptions, by someone who was each time trying to do something else, it raises the prior that there is a real object underneath both descriptions — a force in the way organizations govern, recognize, and discard their own capabilities — worth looking at directly. That is the only claim this note makes: not that the bridge carries weight, but that it points somewhere worth the attention.
— Iván Abril Palma