Evidence-led progression

Seen, done, adapted, transferred, integrated, organised

Evidence about a capability sits in one of seven states. They are ordered, so a later state means more than an earlier one, and they are not numbers: they cannot be averaged, added, or treated as if the distance between two of them were fixed. Take one capability, escaping the mount, and follow it through.

What changes between the states is the condition under which the capability was seen. The first state is exposure. The second is recognition under pressure, and it is separate because a grappler can execute an escape beautifully on cue and never notice the moment in a round. The third and fourth are execution and then adaptation, both inside declared bounds. The fifth is transfer: the same capability under materially changed conditions. The sixth is live integration, when the situation is not announced. The seventh is the athlete organising the capability’s own development. Locally, a coach watches for the same thing in five dimensions kept separate: recognition, execution, adaptation, retention and transfer.

Per dimension, never one score

States are assessed per capability, per system, per problem or per organisational dimension, never per athlete. There is no global score because a score would have to add states across dimensions, and the states do not add. A grappler who escapes mount live and cannot yet recognise a leg entanglement is two facts, and a total would destroy the more useful one. This is also why advanced ranks do not get new states. A brown belt and a blue belt can both hold the self-organised state; what differs is the object the evidence is about, and the conditions: specialist depth, adversarial pressure, opponent sophistication, context variation, novelty relative to the athlete, solution latency, transfer, persistence over time, and self-correction under contradiction.

What an observation has to be

An observation counts towards a state only if it was the right kind of observation for that state. Live execution, adaptation and transfer require representative interaction: a partner with real agency, in a task that puts the capability under the pressure the claim is about. A technical rehearsal cannot be relabelled as live transfer evidence, however clean it looked. Transfer needs materially changed conditions, such as a different partner, side, entry, surface, phase, opponent response, intensity or ruleset; repeating the same escape against the same partner from the same start is cosmetic repetition, not transfer. The chain the record keeps runs from exposure, through material observation, to current evidence, to any gate debt, to remediation or clearance, and on to retention and the floor the next level stands on.

Unknown stays unknown

If a capability has not been seen, the record says it has not been seen. It is not scored as zero and it is never read as poor performance. A blank in a table is easy to read as a bad mark, and the rule exists because that misreading is constant. Missing observation is kept distinct from observed failure, from evidence that was not independent of a cooperative rehearsal, from a context too narrow to count, from work that was blocked on safety, from evidence that has gone stale, and from absence, withdrawal or ineligibility. Each of those is a different situation with a different response, and one blank cell would erase the differences.

The practical consequence is that a gate can return “insufficient evidence” as its own outcome, and it is not a failure. A grappler with thin evidence and a grappler with negative evidence are in different places, and the coach’s next move differs accordingly: one needs more looking, the other needs more work.

Three clocks

Calendar time, training exposure and current admissible evidence are three separate quantities, and none of them is ever substituted for another. Someone who has attended for four years has a calendar figure. Someone who has trained six hundred hours has an exposure figure. Neither is evidence that they can do anything in particular, and only the third quantity feeds a level. No level has a universal minimum or maximum duration. A worked programme may use eight, twelve or sixteen weeks, a competition season, or an open-ended cycle, and the reference duration is never the credential. Fast progression is valid and slow progression is valid; the same evidence and the same safety gate apply to both, and neither is intrinsically better.

Safety cannot be traded

A blocking safety condition is never offset, averaged away or numerically erased by strength on any other dimension. A purple belt with a formidable passing game and a habit of cranking late on a partner’s neck has a blocking condition, and no amount of passing changes it. No advanced projection can average away unresolved terminal-safety debt, catastrophic unowned viability problems, recurrent unsafe partner behaviour, an inability to tell practice permission from competition legality, or a collapse of the floor a previous level established. This is the deepest reason there is no total: the moment a total exists, a good total can hide a blocking gap, and the arithmetic has done something no coach would do.

Retention, decay and re-observation

Clearance is historical and evidence is current. A capability cleared a year ago may now be current, due for a retention check, confirmed on re-observation, under decay pressure, or stale and unknown. A later failure to reproduce it creates decay pressure, current gate debt, targeted maintenance or a re-entry calibration; it does not erase the fact that the capability was once shown. The same logic runs the other way for an athlete returning from injury or absence: they are not reset to a beginner, and their unknowns are unknowns rather than failures.

Maintenance is not regression. A purple belt sent back to a simple, cooperative elbow-knee escape drill is being asked to retain and retrieve a capability, and that is a different objective from developing it; the record keeps the two apart so that a return to simpler tasks is never read as a demotion.

From evidence to level to belt

Capability plus current admissible evidence produces an internal level. The internal level carries a developmental meaning. Only after that, optionally, is anything external projected onto it. White means broad structured acquisition with increasing technical absorption; blue means technical depth, counter and proactive offence, late defence and increasing self-organisation; purple means self-organisation as the default, deepening into a specialist, robust advanced game; brown means robust expert integration becoming generative problem solving; black means durable generative expertise. Reading that ordering backwards, from the belt to the capability, inverts the whole architecture. Competition results are evidence of transfer, adaptation and robustness under score, time and boundary, and a result alone cannot create a level, because a raw result is confounded by age, size, athleticism, opponent quality, ruleset and opportunity. Competing is optional throughout.

Same rank, different games

Two people at the same level are not expected to look alike. A brown belt may be wrestling-first, guard-first, a leg-entanglement specialist, a passing and top-control specialist or a front-headlock and rear-control specialist, and every one of those profiles is legitimate. What the level protects is the floor: credible broad safety and viability and coherent management of the weaker domains, never symmetrical expertise everywhere.

What a gate can decide

At a gate the decision is one of a short list: clear the level; hold for targeted remediation; repeat part of the work; repeat all of it; compress or reorder the worked content where prior capability is genuinely present; maintain only, before the next level; pause on safety; pause for a reason that is not technical; recalibrate on re-entry; or record insufficient evidence, which is not failure. None of those is produced by a number, and the record behind them is kept sparse on purpose: the coach comes first and the record second, and nothing here asks for a dossier after every round.

Still to be measured

The architecture specifies what evidence is. Whether coaches apply the seven states consistently, what the safety outcomes are in practice, and how naturalistic evidence compares with evidence an assessor sets up, are questions that need people, and the theory records them as unanswered rather than assuming them. They are kept on the research page. What the evidence becomes over years is the subject of the pathway.