The Constraints Led Approach

By OSJJ

Open Source Jiu-Jitsu

THE CURRICULUM SERIES ยท PART FIVE

What Is the Constraints-Led Approach (CLA)?

Everyone in Jiu-Jitsu seems to have a position on CLA. Almost nobody has said plainly what it is. So let us stop arguing and answer the question properly.

Justin Hall / Co-Founder, Open Source Jiu-Jitsu

What is the constraints-led approach? Ask around and you will get answers that cannot all be true at once. Few topics in modern Jiu-Jitsu generate more heat than CLA. One camp treats it as the future of coaching and close to a replacement for everything that came before, teach the games and let the technique take care of itself. Another camp waves it off as a fad with a fancy name, nothing you were not already doing. A third says integrate it, blend it in with everything else. Everyone has planted a flag. But notice what almost none of them do first: actually say what CLA is. They are arguing about where it belongs and how much to use it without ever pinning down what the thing itself even is. You cannot answer where it fits until you answer what it is, and that is the question this piece is going to take seriously.

This series happens to give us a clean way to answer it. In the earlier parts we defined, in concrete terms, what a curriculum is and what a system is. Those definitions are fixed standards, set before CLA was ever in the room, which means we can do something the debate almost never does. We can run a real test. We can hold CLA up against the definition of a curriculum, and against the definition of a system, and see whether it meets either one. A test is only honest if a failure counts as much as a pass, so let us agree up front to follow the result wherever it goes, and find out together what CLA actually is.

The two tests, and why we are running them

Before we test CLA we need the standards it will be tested against, and they are not new. They are the same definitions this series built in the earlier parts, which is the only thing that makes this scientific rather than rhetorical. If I invented fresh criteria designed to make CLA fail, that would be a trick. Instead we use the exact bar every other method in this series was held to.

I want to be clear about what testing actually means here, because the word gets thrown around loosely, and I mean it in the strict sense. The scientific method is not complicated. You start with a hypothesis, a clear claim that could turn out to be right or wrong. You define, in advance, what evidence would confirm it and what evidence would refute it. You run the test against fixed criteria you did not invent for the occasion. Then you follow the result wherever it goes, even when it is not the result you were hoping for. Change any of those steps and you no longer have a test, you have a verdict wearing a lab coat. That is the difference between science and argument, and most jiu-jitsu debates are argument dressed up as science, two people who each decided the answer first and went looking for support. I want to do the opposite here. So here is the hypothesis, stated plainly enough to be wrong: CLA is a curriculum, or CLA is a system. Those are the two claims people implicitly make when they treat it as a replacement for either. We are going to test both against the definitions this series already built, define what a pass would look like before we look, and report what happens. If CLA passes, it passes.

And one thing has to be said before we start, because it changes how you should read what follows. In science, failing a test is not failure in the way most people hear that word. It is information, and often the most valuable kind. A failed hypothesis does two things. First, it settles a question that was otherwise going to be argued forever. Instead of two camps circling the same debate for another decade, you have a fact, and the field gets to move forward from it. That is progress, and progress usually looks like ruling something out. Second, a result you did not expect often points somewhere. If a thing turns out not to be what everyone assumed, the very next question, well then what is it, tends to be more useful than the one you started with. So whatever the tests below return, read them that way. Not as strikes for or against CLA, but as information, the kind that could finally tell us where it belongs and let the argument move forward instead of looping. I am not going to tell you the results here. We are going to run the tests and find out together.

A curriculum, from Parts One and Two, answers what you teach and in what order, and it takes one of four shapes, linear, rotating, flow, or positional. A system, from Part Three, is the connective layer above the sequence, the set of principles and beliefs about what is true across positions, the understanding that generates an answer even in a position you never drilled. Those are the two definitions.

Here is what a pass would require. To pass as a curriculum, CLA would have to tell you what to teach and in what order, to fit, however loosely, one of the four shapes. To pass as a system, it would have to supply a set of principles about what is true across positions, a point of view that generates answers in places you never drilled. Those are the two rulers, set before CLA entered the room. Now we measure.

Test one: is CLA a curriculum?

Here is the curriculum standard, taken straight from the earlier pieces. To be a curriculum, a thing has to do three jobs.

Now hold CLA against those three. CLA does not tell you what to teach. It does not tell you whether to cover the guard before the back, or mount before side control, or anything else about content. It does not tell you the order, and it does not commit to any of the four shapes. You can run a constraints-led class inside a linear curriculum, inside a positional one, inside any of them. CLA is completely silent on all three jobs, not because it is a poor curriculum, but because it is not trying to answer the question a curriculum answers.

Curriculum test: FAIL. CLA defines neither content nor sequence, and fits none of the four types. It is silent on every criterion a curriculum exists to satisfy.

That failure is clean and total, and it is worth pausing on, because it already rules out one of the two things people claim CLA is. Whatever CLA is, it is not a curriculum. It cannot be the thing you build your what-and-when around, because it says nothing about either.

Test two: is CLA a system?

This test is more interesting, because CLA looks closer to a system. It is principle-flavored. It talks about adaptability, problem-solving, and skill that survives live resistance, which is exactly the language Part Three used for a system. So we have to be careful and fair here. Here is the system standard from Part Three.

Now measure. The third criterion, CLA seems to hit, its entire purpose is adaptable skill that works in novel situations. But the first two are where it fails, and the failure is precise. CLA does not supply the principles. It is a method for helping a student arrive at solutions through designed constraints, but it is deliberately agnostic about which principles, and in its purest form it avoids stating them at all. It does not hold a point of view about the art, because it is not about the art, it is about how learning happens. You could use CLA to teach a pressure-passing point of view or a movement-based one or any other. The method does not care, and that neutrality is exactly the problem for this test. A system is your content, your convictions. CLA is a delivery mechanism you could bolt onto anyone system, or, in the dogmatic version, use to avoid having one.

System test: FAIL, but a near miss on one of three. CLA produces adaptable skill, but it supplies no principles and holds no point of view about the art. It is the how of arriving at understanding, not the understanding itself.

So it fails the second test too. Not by as much, and the closeness is informative, but it fails, because a system is the what you believe, and CLA is studiously silent on the what by design.

What two failures reveal: the third axis

A method that fails the curriculum test and fails the system test is not a weak curriculum or a weak system. It is a third thing, and the two failures are how we locate it. Part Two was about sequence, what order you teach. Part Three was about connection, what ties it together. CLA answers neither, because it answers a different question entirely: how do you run the mat. Do you show a technique and have people repeat it, or do you design a constrained game and let the solution emerge from live problem-solving. That is the delivery question, and it is a genuine third axis, independent of the other two.

The two failures, and the one axis CLA does answer. Silent on what to teach, a near miss on connection, and fully present on delivery.

This is why the arguments about CLA never resolve. People are trying to place it on the sequence axis or the connection axis, and it does not sit on either, so the debate spins. Put it on its own axis and everything clicks. You can deliver any of the four curriculum types through prescriptive instruction or through constraints. You can deliver a system either way too. CLA lives on the delivery axis, and it is one of the two answers to the delivery question, the other being direct, prescriptive teaching. Failing both earlier tests is not a demotion. It is a location. It tells us precisely where CLA belongs and, just as importantly, where it does not.

What CLA actually is, given its due

Now that we know where it sits, it deserves a fair and full description, because it is a real, research-backed framework and not a slogan. CLA comes out of a field called ecological dynamics, and its core idea is that skill is not installed by repeating an ideal technique in isolation. Skill emerges from the interaction between the learner, the task, and the environment. The coach manipulates constraints, the rules of a game, the space, the grips allowed, the goal, and under those constraints the learner self-organizes a solution that fits their own body and the live problem in front of them.

Two ideas inside it are genuinely valuable. The first is representative learning design, the principle that practice should resemble the real thing, so skill built in practice actually shows up under live resistance instead of falling apart. The second is that solutions are individual, the best answer for one body and one style is not identical to the answer for another, so a method that lets solutions emerge tends to produce skill that actually belongs to the student. These are not small points. They are among the most useful things modern coaching has to offer, and dismissing CLA wholesale means throwing them away.

Why it fits everything this series argues

Here is what should be obvious by now: CLA is not an enemy of the positional, principle-based, resistance-tested approach this series keeps pointing toward. It is its natural delivery partner. Representative design is just another way of saying train against resistance, the same standard the whole series holds up. Individual solutions is the same truth as everyone executes differently. Skill that emerges from solving live problems is the schema idea from Part Two, understanding built around positions and problems rather than a pile of memorized moves. CLA, used well, is one of the best tools available for delivering exactly the kind of curriculum and system this series argues for. That is precisely why getting its category right matters so much.

The real cost of misunderstanding it

This is the heart of the article, and it is why the test was worth running. If you do not understand what CLA is, you will misuse it, and you can miss in either direction. Both errors come from the same root, a wrong idea of the category.

Overuse it, and you make the dogmatic mistake, the never teach technique, only run games position. If CLA is purely the how, then running CLA with no curriculum and no system beneath it is a delivery mechanism with nothing to deliver. It is a way of teaching that refuses to say what is being taught. Students move and sweat and solve little puzzles, but with no sequence guiding what problems they meet and no principles naming what they are learning, they can drill in circles for years, motion without progress. That is not a failure of constraints. It is the predictable result of asking a delivery method to also be the content and the point of view, which are the two tests it just failed. A how with no what is not a program.

Underuse it, and you make the opposite mistake, dismissing it as a gimmick and clinging to nothing but cooperative drilling. That throws away the single best tool available for making skill transfer to live rolling. The coach who never constrains, who only shows and has students repeat in compliance, builds the brittle skill Part Two warned about, technique that looks clean in the drill and vanishes against a resisting stranger. That coach is leaving the most powerful part of skill-building on the table out of a misunderstanding of what CLA is.

Both extremes are comprehension failures. Understand that CLA is a delivery method, powerful but empty of content on its own, and you dose it correctly. You use it constantly, inside a curriculum that gives it sequence and a system that gives it meaning. Miss what it is, from either side, and you break the whole thing.

It is not new, and that is fine

One more thing worth saying plainly, because the hype depends on hiding it, and because being precise here is the fair thing to do. The core idea is not a discovery. Good coaches have always used constraints. Positional rounds, start in a bad spot and work out, only pass on this side, win by getting to the knee line and nothing else, that is constraint design, and it predates the vocabulary by generations. In the nineties we were doing positional drilling constantly, and that is exactly what this is at its root.

But I want to give credit where it is due, because there is something genuinely newer here, and it is worth naming precisely. It is not the constraint itself. It is the resolution. CLA takes positional drilling and zooms it way in, to a much finer grain. Old-school positional work constrained a whole position, escape side control, pass the guard. CLA constrains a single narrow problem inside that, one specific point A to point B, and drills only that. That micro-specialization is the newer part, and it is a real benefit, because working at that resolution lets a student understand one small piece at a depth that broad positional rounds never quite reached. So the honest framing is this: the concept is old, the granularity is new. It is a micro version of something we have always done, and the micro is where the value is.

What ecological dynamics added on top of that is a research-backed framework and a language for doing on purpose and precisely what good coaches did by feel. That is a real contribution, and it is an evolution, not a revolution. Treating it as a brand-new gospel that overturns everything is exactly the mindset that produces the throw-out-technique overreach. It was always a tool. It is a sharper, better-understood tool now. That is the honest size of the thing.

How to actually use it

So here is the practical version, the part you can run this week. A constraint is a tool you drop into a lesson, not a replacement for the lesson. The workflow is simple. Teach the position and the why, then constrain it to build the skill against resistance. You are not choosing between showing technique and running games. You do both, in order.

Take a concrete case. You are teaching a guard pass. First you teach it, the position, the grips, the mechanics, and the reason it works. Then, instead of feeding cooperative reps, you break the pass into sections and constrain each one. Round one, the only goal is to win the knee line, nothing else counts, and the guard player resists live. That is it. One focus, full resistance, a clear win condition. Then the next section gets its own constrained round. The student is now solving the real problem live, in a piece small enough to actually improve at, which is where skill is built. You showed them the answer, then you designed the game that forces them to make it work for real.

Where it meets the system and the curriculum

There is one more use that is the most powerful of all, and it resolves a tension Part Three left open. Part Three said a great teacher makes the connections and the principles explicit, teaches the why out loud. A strict reading of CLA says the opposite, do not tell them, design a constraint so they discover it. Constraints let you have both. You can use a constraint to teach a principle, by designing a game where the principle is the only way to win. If you want a student to learn that controlling the hands breaks posture, build a round where they can only score by controlling the hands first. They will discover the principle under live resistance, and then you name it. Discovery and instruction stop being enemies. The constraint delivers the principle your system already believes in, and the naming makes it stick. That is CLA and a system working together, each doing the job it is actually built for.

The test does not end the argument, it moves it

Here is what running this test actually accomplishes, and it is not what you might expect. It does not end the debate about CLA. Debates like this rarely die, they relocate, and a relocated debate on better ground is most of what progress ever looks like. The old argument was should you teach only CLA, or should you integrate it, with people shouting past each other about a teaching methodology as if that were the whole question. Once you see that CLA is a methodology and nothing more, that argument dissolves and a sharper one takes its place: can you run a school on a teaching methodology alone, with no curriculum and no system underneath it?

That is a better question, and notice it is no longer about CLA specifically. It applies to everyone. And the first honest answer is yes, you obviously can, because most academies already do. The majority of school owners I have met have no curriculum and no system. What they have is a teaching methodology, usually the traditional technique-based model, and a stack of lesson plans they pick more or less by what feels right that week. That is running on methodology alone. So a school that teaches only CLA, with nothing under it, is not some radical outlier. It is doing structurally the exact same thing as the technique-only school down the road. Different method, identical foundation, which is to say none.

So can you teach on a methodology alone? Clearly. The real question is whether you should, whether it is any good, and here I do not care which camp you are in. I am not arguing CLA against traditional instruction, or either one against a healthy blend of the two. The methodology is not the argument. The argument is this, and it is hard to debate: a school with a curriculum and a system, delivered by any competent methodology, will beat a school with no curriculum and no system, no matter how good its methodology is. Even the best blend of methods, run with nothing underneath it, still produces randomized outcomes.

That is the whole point of building a curriculum and a system in the first place. You build them because you have decided what outcome you are trying to create and you are engineering the path to it. Strip them away and it does not matter how you deliver, CLA, technique, or the finest mix of both, you are teaching without a destination. You can still say your goal is outcome X. You just have no equation to get there, so you will not, except by accident. Teach beautifully and arrive nowhere on purpose.

So the debate was never really about CLA, or about which methodology wins. It was about the layer almost nobody is arguing over. Win the methodology argument completely, pick the perfect way to deliver, and you can still lose, because delivery sits on top of a curriculum and a system, and without those two you are gambling on results. That is where the real conversation should have been the whole time.

The bottom line

CLA failed the curriculum test and failed the system test, and that double failure is the finding, not a knock. It is neither of those things because it is the third axis, delivery, and it is one of the most valuable tools on that axis when you use it for what it is. Run it against the standards this series set and the verdict is consistent and clear: not a curriculum, not a system, a way of running the mat that can teach either one better than cooperative drilling ever will. Understand that, and you use it in every class, inside a real curriculum and a real system, and it makes your students better faster. Misunderstand it, in either direction, and you either build motion without progress or leave your best tool in the bag. The science does not diminish CLA. It puts it exactly where it belongs, which is the only place it works.

Coming next

Step back and here is what this piece was really for. It was not to have the final word on CLA. It was to open a new section of the series, the teaching methodologies, the third axis, and CLA is a good door to walk through because it is the one everyone is already arguing about. Answering what is CLA was the way in. From here the series stays on this axis and widens out. We will come back to CLA, but alongside the other methodologies, prescriptive instruction, games-based teaching, guided discovery, differential learning, and the rest, and look at how each one actually works, what it is good and bad at, and how it can be integrated into a real curriculum and system rather than floated on its own. That is where the useful work is: not picking a methodology to worship, but understanding all of them well enough to build the best plan for what you are trying to create. The goal of the whole series remains the same, a complete, honest map of how to build and teach Jiu-Jitsu, from what you teach, to what ties it together, to how you deliver it. This piece just opened the last of those three.

Open Minds. Open Mats. Open Source.

Back to the Free Library