The tacit remainder

A method that works has an arithmetic problem waiting for it. One practitioner, a finite number of hours, a waiting list that grows faster than the hours do. The answer the professions have reached for over the last century is codification. Write the method down, build a training around the writing, certify other people to deliver it, and let the manual do what the person cannot be in enough rooms to do. It works, in the plain sense that the thing spreads. It also fails in one particular way, and the failure repeats across fields that share no vocabulary, no regulator, and no professional lineage, which is what turns it from bad luck with licensees into a property of codification itself.

The pattern is this. Codification reliably transfers a method’s structure and imperfectly transfers its judgment. Everything about the method that can be stated arrives at the far end intact: the sequence, the agenda, the diagnostic categories, the words. What does not arrive is the thing the originator does when the room stops matching the manual. We are going to call that residue the tacit remainder, because it is not a defect in any particular training program. It is what is left over after the best possible writing-down, and every serious licensor of a method is quietly managing it.

What a manual can carry

Look at the methodologies that have been documented and licensed most carefully, the ones you can research from outside without knowing anyone. The Entrepreneurial Operating System runs a franchise, with a total investment in the low six figures by the public franchise disclosure summaries, ongoing fees of roughly fourteen hundred dollars a month, and a ladder of implementer designations that reads exactly like a quality system. Even the number of implementers is a small lesson in what codification produces: the public tallies do not agree with each other, ranging from a few hundred to several hundred more depending on which aggregator is counting what. What the franchise transfers is real. A meeting agenda, a scorecard, a quarterly cadence, a shared vocabulary that lets two strangers describe the same organizational problem in the same words by Tuesday afternoon. That is not nothing. It is most of what a manual can be.

Now look at what the same market does when fidelity matters more than reach. Dare to Lead, built on Brené Brown’s research, has kept its certified facilitator population small by design, fewer than three hundred by the count the facilitators themselves publish, and at present the certification is closed, with no new trainings scheduled. That is a licensor choosing the other lever. If the judgment will not travel through the training, restrict who gets trained and accept the ceiling on scale. Internal Family Systems sits between the two, with a long post-training consultation requirement standing between Level 1 attendance and certification, which is the same admission made structurally: the training is where the structure transfers, and the years afterward are where something else is supposed to.

The clearest case is the one that refuses the whole apparatus. The North American Reggio Emilia Alliance states flatly that the approach cannot be considered a method or a pre-established curriculum, that the municipality of Reggio Emilia does not approve of any certifications in it, and that there are no Reggio Emilia schools outside the city of Reggio Emilia. Schools elsewhere call themselves Reggio-inspired, and the Alliance is candid that this phrasing is self-determined and that the resulting practices vary enormously. Montessori made the opposite choice and built certifying bodies. Neither choice dissolves the remainder. Reggio names it and declines to pretend, Montessori certifies against it and lives with variance, and the two together are a natural experiment in what a credential can and cannot promise.

The four dimensions

Our own research into how these licensing models actually fail converged on four places where the gap sits, and the four are worth stating separately because they fail differently and are learned differently.

Decision trees, first, meaning what to do when the person in front of you does not do the expected thing. The manual has a step for resistance. It does not have the twelve variants of resistance an experienced practitioner can distinguish on sight, or the knowledge that this one is fear and that one is a test and the third is simple fatigue and should be met with a break rather than a technique.

Micro-timing, second. When to probe and when to leave the silence alone. Every method has a moment where the same intervention is correct at one second and destructive at the next, and the difference is not in the intervention. Nothing in a written sequence can carry a variable that fine, because the sequence is a list of steps and this is a property of the space between them.

Interpretive frameworks, third, meaning what a behavior means inside the method’s worldview. Two practitioners watch the same client cross their arms. One sees defense, one sees cold, and the method’s own theory says which reading is available and which is not, but the theory was learned as prose and applied as perception, and the conversion from prose to perception happens somewhere the training did not go.

Feedback loops, fourth and most neglected. The originator knows they got it wrong because they have run the method thousands of times and have a felt sense of what a session that is drifting feels like at minute twelve. A newly certified practitioner has no such instrument. They will learn it, if the work gives them enough repetitions and honest information about outcomes, which is exactly the thing that shrinks when a method is delivered at scale by people who see each client once.

Notice that all four are discriminations rather than procedures. The manual answers what the method does. All four of these answer which, and when, and how much, on this occasion, with this person, and those are the questions the manual was never in a position to answer, because the answers were never in sentence form even in the originator’s own head.

The oldest statement of it

None of this is new, and the cleanest statement of it is sixty years old. Michael Polanyi opened The Tacit Dimension in 1966 by proposing to reconsider human knowledge starting from the fact that we can know more than we can tell. His examples were the ordinary ones, riding a bicycle and recognizing a face, where a person performs flawlessly and cannot produce the rule they are following, and his argument was that this is the normal structure of skilled knowing rather than an exception at its edges. The economist David Autor gave the idea its modern name in 2014, calling it Polanyi’s paradox in an argument about why computerization had swallowed some jobs whole and barely touched others: what we cannot describe, we have not been able to instruct a machine to do.

The implementation science literature reached the same place from measurement rather than philosophy. Carroll and colleagues, writing in Implementation Science in 2007, proposed a framework in which fidelity is assessed as adherence to the intervention’s content, coverage, and frequency, moderated by quality of delivery and participant responsiveness. Read that as a confession and it is a candid one. If delivering all of the components at the right dose were the same as delivering the intervention, no moderators would be needed. The moderators are the field’s own bookkeeping for the tacit remainder, which is why fidelity monitoring in the wider literature so consistently records what was delivered and so rarely records how well.

What codification is for

Which leaves the question our deskilling pieces circled and did not answer. If judgment does not transfer, and codification is what we do when we want a method to outlive one person’s calendar, what exactly is codification for?

The wrong answer is that it is a lossy transmission of expertise, a copy that came out at eighty percent and would be better at ninety-five. That framing makes the remainder into a defect and sets everyone chasing a better manual, which is how a field ends up with certification ladders that grow more elaborate as their graduates get no better. The remainder does not shrink much when the manual improves, because the remainder is not made of missing sentences.

The better answer is that codification manufactures a smaller and more locatable problem. Before the method is written down, a newcomer must acquire everything at once, structure and judgment together, by proximity to someone who has both, which is apprenticeship and costs years. After it is written down, the structure is free and the newcomer’s whole remaining task is the remainder, which is now four named dimensions instead of an undifferentiated fog. That is an enormous gain, and it is a completely different gain from the one certification usually advertises. The manual does not transmit judgment. It clears the ground around the judgment so that a person can see what they still have to learn, and so that whoever is responsible for their formation knows what to spend the scarce hours on.

Three things follow, and they are practical. Grade any codification by how honestly it marks its own edges, because a manual that says nothing about where its authority stops is teaching a false confidence along with its content, and confident wrongness at the boundary is the failure mode the certified practitioner will actually produce. Put the human where the remainder is rather than where the volume is, since the four dimensions tell you exactly which moments in a method are load bearing for judgment and which are procedure that anything can run. And budget for apprenticeship as a line item, because the remainder was always transferred by proximity and repetition, and it still is, and nothing that has happened to the cost of writing things down has changed that at all.

There is a version of this that reads as a limit and a version that reads as an inheritance, and the second one is closer to true. Every method worth licensing began as one person’s accumulated discrimination, and the part of it that resists writing is not the part that failed to make the transfer. It is the part that is still alive, still being learned by somebody this week in a room with somebody else, still capable of being wrong in an interesting way and corrected. A method fully captured in a document would be a method that had stopped developing. What we have instead is a structure anyone can pick up in a weekend and a remainder that takes years, and the remainder is where the people are.