Eli Goldratt’s Theory of Constraints in running coaching: can we reliably create sustained athletic achievement in runners?

“Sustained athletic achievement” is a phrase seldom heard when talking about runners. By now, nobody needs to quote the staggering injury statistics in Western running populations: According to an epidemiological study, there are 2.5 to 12.1 injuries for every 1000 hours of running. 20 to 70% of those injuries are recurring, and 30 to 90% of those injuries result in a reduction in training.

Is this because running is inherently injurious? Probably not—and some would argue that we’re in no position to know: the rates of injury aren’t due to the fact that we’re running, but instead due to the fact that we’re running unprepared. In Movement, Gray Cook writes that “many times, the activity gets the blame when the blame should be placed on the poor foundation the innocent activity was placed upon.”

Let’s translate this: are our calves mobile and strong? Are our hips stable? Are our flexors and extensors working well together with our abductors and adductors? These are questions that runners typically only ask themselves after an injury or ten.

Whenever we train an athletic activity such as running, it’s important to figure out what might hold our training back, rather than just going out to hit the pavement and hope for the best. There is a theoretical framework that may provide us with a systematic way of finding solutions to these widespread problems: Eli Goldratt’s Theory of Constraints (TOC).

At the general level, the Theory of Constraints consists of 5 steps:

  1. Identify the system’s constraint.
  2. Decide how to exploit the system’s constraint.
  3. Subordinate everything else to the above decision.
  4. Elevate the system’s constraint.
  5. Find the new constraint.

In Critical Chain: the theory of constraints applied to project management, Graham K. Rand writes: “The system’s constraint is the part of the system that constrains the objective of the system.”

Overuse injuries in running are rarely generalized. In other words, it’s always something specific: either a bad knee, or shin splints, or plantar fasciitis is stopping us. In other words, that’s the constraint that doesn’t let us log more miles.

A lot of us are really good at doing the first two steps. We already identified the constraint (at least superficially speaking)—say it was a tight IT band. Then comes step two: deciding how to exploit the system’s constraint. We roll out our tight IT band, so that we can log as many miles as possible.

But a lot of us don’t get past step 2: we keep logging miles and more miles, until our IT band is so sore that we can’t run at all. Doing step 3 would mean figuring out how many miles we can run without injury. Here’s the problem: if we actually did an honest assessment, the answer would typically be “not many.” Certainly not enough to train for a marathon, probably just enough to train for a 10k.

Which brings up to step 4. We’re trying to train for a marathon—or train for a fast 5k—and this IT band doesn’t let us go far or fast. What do we need to do? Elevate the system’s constraint. Otherwise, that tight IT band won’t let us develop the speed or endurance we need for our event.

When you look at the problem of athletic development broadly, it doesn’t make much sense to spend time and effort developing endurance when a problematic knee or IT band isn’t letting you progress.

In Critical Chain, Eli Goldratt writes: “What property typifies the chain? It is the strength of the chain. If one link breaks, just one link, the chain is broken. The strength of the chain drops to zero.”

This is the tired story of overuse injuries and recurring injury in runners. We often sideline ourselves by running through injury. We break the chain, instead of strengthening it.  We try to increase our endurance, when ironically our present endurance may be greater than we know—but we can’t experience it, given that the system is constrained by a malfunctioning part.

We should always focus on the weak link. “Remember,” Goldratt writes. “You are not really interested in my link. You are interested in the chain. If I made my link stronger, how much did I improve the strength of your chain? Nothing. Absolutely nothing.”

In previous posts, I’ve alluded to the possibility that “the plateau” may be deeply related to the flawed thinking that Goldratt attempts to correct: perhaps the case is that we’re training endurance when the constraint of the system is strength, or hip stability. We don’t see gains in endurance because we don’t address the constraint, and we perceive that we “plateaued.”

What’s the problem? Why did we miss the constraint?

The problem, Goldratt proposes, may be in our ideas and in our personal culture. A typical assumption in project management is that “the only way to achieve good global performance (is through good local performance everywhere.” Although this idea seems to make sense at face value, Goldratt disagrees: “The fact that so many managers and almost all our systems are based on this assumption is regarded by TOC as the core problem…”

Project management and athletic training are not so far apart: the same problem is present in both. Look at your training plan.Most athletic programs look for good local performance everywhere. Chances are that your training plan is similar to many other training plans: do fartlek, strength training, endurance, cardio.  The mainstream philosophy is to hit every side of the problem, all at once. Of course this works, in the sense that the body develops, but does it work well?

By the best standards, probably not. And if you keep getting sidelined by injury, certainly not.

I hope to have shown that the principles provided by the Theory of Constraints can be easily adapted to create a system for athletes and coaches, by which they can jointly achieve two objectives that are typically at odds with each other: injury prevention/management and athletic development. Applying the Theory of Constraints to athletic coaching may allow us to define athletic development in such a way that these two objectives cease to be in conflict. I believe that on a deep level, this conflict of interests is the likeliest culprit of the staggering running injury statistics. Settling it will benefit athletes, coaches, and the running culture in general.

I’ll devote my next post to fleshing out the details of this conflict of interest (and how to resolve it).

Descriptive vs. prescriptive in running.

When I read articles about running, I often come across phrases like “no single foot-strike pattern is representative of the entire running population.” True enough, but it doesn’t really help runners: all it does is describe the present state of affairs of the running population.

The problem that I see with this is that many people—many scientists, even—take this descriptive observation about the world and turn it into a prescriptive one. Within their statement is a hidden interpretation (shown in italics):

 “No single foot-strike pattern is representative of the entire running population. Therefore, no single foot-strike pattern should be adopted as a baseline gait for a human population.”

Why is this problematic? Let me give you another example—one that we’re all comfortable with.

Let’s suppose that we did the very same research, only about how people lift heavy objects. Statistically, our findings would be similar to running; as researchers, we’d be prompted to say: “no single lifting pattern is representative of the entire human population.” In other words, we’d make our analysis, and see that some people lift objects by bending at the waist, and others lift objects by bending at the hips.


The difference, of course, between running and lifting heavy objects is that we have a clinical standard for lifting. We know that bending from the waist is a bad idea for virtually any human out there. There are only three options for lifting objects, and when you really think about it, there’s only one:

  1. Bend from the waist to lift a heavy object and get injured
  2. Bend from the waist and only pick up light objects without injury
  3. Bend from the hips (correctly) to lift heavy/light objects without injury

In light of this knowledge, let’s review the following statement: “no single lifting pattern is representative of the entire human population. Therefore, no single lifting pattern should be habitually adopted as a baseline lifting pattern for a human population.” This statement seems ridiculous, and kind of insistently missing the point.

But we should keep in mind that the reason it seems ridiculous is because we have a clinical standard for lifting heavy objects, namely, to minimize trunk flexion throughout the lifting action.

This, of course, doesn’t mean that midfoot/forefoot striking is better than rearfoot-striking (although it certainly sets me up to make that argument).  What it does mean is that descriptive observations about a population’s habits tell us very little about what that population should be doing. They only tell us much about what it is doing. And what we do know is that given the stratospheric injury rates for runners, the running population is doing something wrong.

We need a clinical standard for running. In order to get one, the first step is to stop interpreting descriptive statements as if they were prescriptive ones.

UPDATE: Here are a couple of good articles on how foot-strike could be a function of running speed. This all adds to the question: what should the clinical standard be—which part of our foot lands first? Probably not. But we need a standard. There are a few ideas out there, but I’ll leave that for another post.

How do we measure athletic performance?

In Critical Chain, Eliyahu Goldratt captures a statement that has been immortalized in the arena of business management:

 “Tell me how you’ll measure me, and I’ll tell you how I’ll behave.”

When the criteria that we measure are wrong, or incomplete, we can create disastrous consequences.

Goldratt tells the story of a steel company, in which the standard for measurement for the production division was tons per hour. This is great, right? Measure how much the production division is putting out, and everything should go well! Well, not so much: this form of measurement wasn’t taking into account the needs of the company as a whole.

As Jamshid Gharajedaghi writes in Systems Thinking: Managing Chaos, “Major organizational theories have implicitly assumed that perfectly rational micro-decisions would automatically produce perfectly rational macro-conditions.”

That was the assumption in this case: measure tons-per-hour (which is a rational measurement for the production division) and the macro-conditions would be rational too: the company would just make more money!

As you probably can intuit, that wasn’t the case: the imperative to maximize tons per hour created a situation where materials that were easy and quick to process were processed first (regardless of delivery promises), and specifications that the buyers asked for were ignored, in order to reduce production time: if the buyers wanted say, a ton of 12” steel beams and a ton of 10” steel beams, the production division would just produce two tons of 12” steel beams (cutting down set-up time by half) and then complain to the supply division that they only received a single ton of steel to work with.

A very similar thing happens in athletic training: we tend to create all the wrong results because here too we measure the wrong things.

For example, look at how catastrophic it has been for a majority of people to focus on weightloss when they want to become “fit.” People have been pushed into disregarding a majority of the body’s systems to focus on only one variable: the amount of visible body fat. That’s what we measure, and people’s behavior typically homes in on body fat (or on any one thing), disregarding anything else.

As Goldratt cautions us, “Most of the local improvements do not contribute to the global.” In other words, by homing in on body fat we might be creating huge problems for us further down the line (see: thyroid problems).

A similar thing happens when we try and measure “miles.” That we measure this is ridiculous, given the well-known saying “quality over quantity.” Well, what is a “quality mile”? Do you know? I certainly don’t—or at least, it hasn’t been defined for me. How many runners out there hobble out their miles, fighting through knee pain and achilles tendonitis, popping kidney-destroying doses of ibuprofen? A lot.

We’re measuring the wrong things, we’re seeing the majority of people behave the wrong way, and we’re achieving the wrong results.

The successful ones are those who staunchly ignore the ravings of modern exercise culture—“fight through the pain,” “pain is inevitable”—and concentrate, against all odds and against the prevailing cultural current, on quality, precision, and correct function.

Even though we’d like to think that these problems are intractable, and solutions are difficult to find, they’re actually out there. Solutions to these problems have been diligently developed in the realm of business management, and have yet to reach the domain of athleticism.

The fact of the matter is that by measuring the wrong things, we’ve created these negative, pathological systems for ourselves. It is because we measure the wrong things, that we create the wrong behaviors. Injury, overtraining, and hypergymnasia don’t exist in a vacuum. They exist in a context of social pressures and cultural expectations—cultural measurements of people. And they’re the wrong ones.

So what are the right ones?

That’s the million dollar question.

The beauty of the thing is that tens of millions of dollars have been already spent in answering this question—and the answer has already saved smart businesses hundreds of millions of dollars. We have ten-million dollar answers to our million dollar question.

And the answer, generally speaking, is to make measurements that reflect the whole, instead of a particular faculty, property, or variable. Generally, this means changing a quantitative measurement (position of the heel upon foot-strike) with a qualitative one (gait harmony).

Instead of measuring foot-strike, let’s measure correct movement. Instead of measuring miles, we can measure distance covered without injury. Or, for example, if we must insist on measuring power, let’s make sure it’s not muscle power (of one or two muscles or isolated systems) but rather athletic power (of various systems in various domains across the whole body).

Always, always, measure the whole. That’s what coach Alberto Salazar does with his Athletes at Nike: he learned the lessons of overtraining—what he calls “extreme athletic excess”—the hard way. And he knows that the only way to get quantity—lower times for his athletes (or perhaps, more medals), is by addressing the qualitative aspects of training, and by being better at it than maybe anyone else in the world.

Most of us fitness enthusiasts know how to periodize training plans so well that we can’t get any better at it. Most of us would benefit little from broadening an already encyclopaedic knowledge of training exercises. How we implement training programs is where most of us fail. And our behavior, the execution of the “how,” is tied above all to the way that we measure ourselves, and to the measurements that we conceive for others.

Let’s make damn sure that they’re the right ones.

A bit of running advice.

During the swing phase, lead with your knee, not with your foot. By “floating” the knee in front of you as your leg swings up, maintaining knee bend as long as possible, you will:

  1. Increase full-body forward lean.
  2. Allow a complete contraction of the extensors (gluteus maximus, hamstrings, and calf muscles).
  3. Allow a complete contraction of the flexors (Sartorius, iliopsoas, frontal calf muscles).
  4. Increase your speed by increasing thigh spread (the distance between your swing thigh and your pushoff thigh).

Galen Mo

Look at Galen Rupp and Mo Farah in the picture above: their swing thighs are in a very similar angle to the ground despite Galen being in late stance phase and Mo in early pushoff. (The main difference is the angle between their thigh and their calf, not the thigh and the ground). You can see that their swing hip is completely rolled forward, meaning that their feet (off camera) can easily manage full pronation.

Floating the knee should feel like you’re falling, similar to what you feel during a lunge, except that your foot ends up coming down under the center of gravity. You should feel your swing hip hike up and your pushoff hip press down. Master this by skipping, focusing on bringing your knees far, far in front of your body with your thighs in a straight line.

Bonus points: Look at their body geometry. At the height of the swing phase, you can draw a straight line from the top of the knee to the bottom of the elbow in both athletes, smoothly connecting thigh and forearm. Elegance always holds the key to speed.

Deconstructing the Plateau: Part 2

When our athletic ability plateaus, and we no longer see the gains in speed, strength, or endurance that we used to see before, we tend to increase our training volume: more hill repeats, more squats, more miles.

Training like this is rarely the right answer. The human body is a phenomenally complicated system—those of us who have been chronically overtrained and injured know that for a fact.

Obvious, straightforward approaches aren’t enough for a system like this. Sure, there are parts that are plainly related to particular abilities: fast-twitch muscle fibers, sugar and ATP to speed, slow-twitch muscle fibers, lungs, fat, and mitochondria to endurance, and muscle size and maximal effort to strength. But ultimately, we need to appreciate the behavior of the system as a whole, and tailor our training to the system as a whole.

If we want to achieve this, there is no idea more important to understand than the systems thinking notion known as emergence.

Emergence addresses the fact that a whole is larger than the sum of its parts: while the parts of a particular system, whether they be atoms, muscles, cars, or people, have properties of their own—atoms are vibrating at certain rates, muscles can be strong or weak, cars can be fast or slow, and people can be skeptical or not—when you put these parts together into a system, you get properties that apply only to the system, and not to the parts.

In other words, these properties emerge from the interaction of the parts, and therefore, of their organization into a system.

These are called emergent properties.

Solidity, for example, is an emergent property. A liquid becomes a solid when the molecules that compose it get colder (and therefore closer together) and move beyond a certain temperature threshold. Nothing happened to the molecules themselves. But the changing nature of their interaction changed a property that expressed itself in the system of molecules: that system went from being a liquid to being a solid.

In the same fashion, speed, strength, and endurance are emergent properties of the human body in the athletic domain. How so? Even if one muscle is strong, fast, and possesses good endurance, it can’t express that speed, strength, or endurance unless the rest of the body’s faculties—opposing muscles and circulatory, endocrine, and nervous system, to name a few—are also functioning properly and interacting correctly with each other.

What does this tell us?

That the particular capabilities of particular muscles or internal bodily systems don’t matter as much as the proper interaction between those systems.

To develop greater endurance, it is not enough, for example, to train simply the aerobic engine (even if you think of the “aerobic engine” as comprising the lungs, blood vessels and capillaries, diaphragm, all the way down to the mitochondria). Bone, tendon, fasciae, and even the fluid sacs around the joint must be developed enough to withstand the added use made possible by a more powerful aerobic system.

Without developing these systems—and others—together with each other, and ensuring that they are equally balanced and capable of interacting with each other at the highest level of performance, we’ll see our increases in athletic ability slowly grind to a halt.


In a previous post I mentioned how different variables—say, the power of different bodily systems—go from being apparently unrelated to being frustratingly interrelated as we develop the system’s capabilities:

“The perceived set of independent variables changes to a formidable set of interdependent variables. Improvement in one variable would only come at the expense of the others.”

-Jamshid Gharajedaghi

That is essentially what is happening here: as we develop the cardiovascular system, the musculoskeletal system, or the nervous system, we find that further increases in our cardiac output, muscle power, or ability to concentrate lead us down a problematic path. If we develop too much capability in one of these domains, without training others, we’ll end up creating conditions—like running too many miles on untrained calves—that will end up destroying our athletic ability.

But there’s more to this than just training each given component, and more to it than even training them to match each other’s capabilities. As I mentioned above, systems aren’t really built from parts; they’re built from interactions. So we must train the ability of the different parts to interact with each other.

To name one common example of what happens when we don’t, there is the Valsalva manuever, which consists of holding our breath when we exercise. We do this because of a dysfunction of the deep muscles of the hip and the spine, and their inability to work together with the diaphragm. The Valsalva manuever can raise thoracic blood pressure to dangerous levels, and put the athlete’s life at risk.

It is typical to see runners with hip/spine dysfunctions hold their breath every few steps, or time their breath to the landing of a particular leg. This has the potential to exacerbate a whole bunch of gait problems, not to mention the loss of speed, power, and endurance, and the effort implied in having to overwork the lungs to make up for the dysfunction at the hip and the spine.

This, or some form of it, is typically why as runners and athletes we plateau. We can’t move forward with our development: the components themselves are preventing each other’s ability to evolve.

The underlying problems must be resolved, but above all, the functioning of these systems must be synchronized. They must interact; their functioning must assume that the other system is also at play.

When we achieve this with more and more of the body’s components we will observe dramatic increases in the emergent properties of the athletic body: speed, strength, and endurance.