Notation: Complex Transitions 1

Scav and I have been trying to figure out how to denote complex transitions i.e. transitions between stances by means of other actions. In the discussion of my previous post ey noted the following problem:

I wonder about stance transitions that occur as part of an attack. In IAKSA kick-boxing (based loosely on Lau Gar Kung Fu), a front-leg kick usually coincides with a short hop forwards, landing in the same stance. A rear-leg kick can transition into the opposite stance or the leg can be brought back to land in the same stance.

The front-leg kick isn't much of a problem, but we don't yet have a way to cope with the issue raised by the rear-leg kick example. The performer needs to know what stance to assume following the execution of a kick. Experience tells me that, more often than not, the performer maintains the same stance after executing a kick. In which case we can keep the notation compact and efficient by adopting the following rule:

All changes in stance must be explicitly noted.

So, if I write:

    FP
    FBK
HMS 

This is to be read as:

1. From left half-moon stance
2. Execute a right front ball kick,
3. Return to left half-moon stance, and
4. Execute a right front punch.

Now, how about a stance transition? I think we have most of the notation we need already. Given that we render

1. From left half-moon stance
2. Half-moon forward
3. Into right half-moon stance.

as

    HMS
   ↑
HMS

It seems logical to record

1. From left half-moon stance
2. Execute a right front ball kick and
3. Land in right half-moon stance.

as

    HMS
   ↑FBK
HMS

Combining the symbol for forward translation of the center of mass (↑) on the same line as the symbol for front ball kick (FBK) seems a natural way to indicate that these things should happen simultaneously. This, in turn, suggests the following general rule:

Simultaneous actions are recorded side-by-side on the same line.

Aight, that's enough for now. Comments?

Notation: Basic Conventions

We've discussed the intent of the system under development at length; now lets try to write something down. I'd like to start with a simple example in order to get an understanding of fundamental issues which need to be resolved before moving on to something which approaches a practical application. Let's start with the following series of movements¹:

1. Start in horse stance.
2. Quarter-moon forward into left half-moon stance.
3. Half-moon forward into right half-moon stance.

What are the key pieces of information which need to be conveyed and what can be omitted to reduce redundancy? Let's look at it line-by-line²:

• Start in horse stance: This is the first line in a sequence of instructions, so "start" is implicit; it's nonsensical to start anywhere except at the beginning.
• Quarter-moon forward into left half-moon stance: Observations on this instruction:
  • It's possible to move both forward and backwards out of horse stance, so the direction of movement must necessarily be included.
  • You can move into any other stance from horse stance, so the desired final position must be indicated.
  • Both left and right variants of half-moon stance are available, so the left/right designation needs to be retained.
  • The "quarter moon" step is the only valid transition between horse and half-moon stances, so there's no need to state it explicitly.
• Half-moon forward into right half-moon stance: Does the left/right distinction have to be maintained, or is it implicit that you would transition from left half-moon to right half-moon in this case? Both left → left and left → right are valid transitions; the former would be accomplished by a shuffling step, the latter by a half-moon. Having specified right half-moon stance, however, the "half-moon" step then becomes implicit.

The sample instructions thus reduce to the following, minimum form:

1. Horse stance
2. Move forward into left half-moon stance.
3. Move forward into right half-moon stance.

In order to capture these instructions we'll need notation for the following concepts:

• Stance
• Relative motion
• left/right

Alright, now we're getting somewhere; we know what concepts we need to record, now we just need to decide how to get them down on paper. Turning once again to Choreo-Graphics, Ms. Guest categorizes notation systems into the following, broad categories³:

• Words and word abbreviations
• Track drawings
• Stick figure (visual) systems.
• Music note systems
• Abstract symbol systems

Can we make any statements at this point about which of these approaches, if any, is likely to suit our needs?

Words and word abbreviations are unlikely to be sufficient by themselves; as we've already seen it's very difficult to efficiently capture the complexity of movement in space and time using natrual language alone. I'm also inclined to eliminate visual systems at this point given Guest's comments on their relative disadvantages:

Visual systems are based on the idea that all dance is visual, that movements are designed to 'make pictures'. This may have been true of classical ballet with its vocabulary of selected, clearly defined positions, but not all movement has 'picture-making' as its purpose, and to try to describe such action in those terms is to force movement into a straight-jacket and thereby change its nature.⁴

MA performances are first and foremost about the use of the body to evade, redirect, and/or apply forces in combat. Rather than existing for its own sake, the external appearance of a form/technique arises as a by-product of this more fundamental concern⁵. This strongly suggests that a visual system is inappropriate for the types of concepts which we're trying to convey.

Lastly, there doesn't seem to be be much reason to consider using a music note system. These types of systems arise in contexts where the intended audience is already familiar with the use of musical notation to convey timing⁶. There's no reason to assume that your average MA practitioners has such a background, which negates any benefits that such a system might have vs. a purely abstract system.

That leaves us with track drawings and abstract symbol systems, both of which show promise. I'm drawn to two systems in particular, Feuillet notation and the vertical staff variant⁷ of Laban Movement Analysis. Feuillet notation indicates left and right by placing symbols on either side of a center line. Given our relatively simple requirements at this point it seems like a reasonable way to make such distinctions.

Laban uses a vertical, multi-line staff which is read bottom to top, left to right. The staff is overkill for our purposes, but I think that reading symbols vertically has the potential to make the system much more intuitive. Let's suppose that we adopt the Feuillet convention for left/right and the Laban convention of reading from bottom to top. Further, let's stipulate that the top of the page always represents "front"/"forward"/"ahead" relative to the performer. Under such conditions symbols on the left of the (printed or implied) center line will always correspond to the reader's left, symbols on the right will always correspond to the reader's right, and reading from one symbol to the next will have an implicit association with forward movement⁸. Looking into the future a bit symbols for rotation, shifts of attention, and the direction of strikes/blocks can use the top of the page (i.e. the readers/performer's front) as the main point of reference, making it easy to render and interpret them. Having provisionally identified a field upon which to record the details of a performance, and a convention for left/right, the next step is to develop a means for conveying the two remaining concepts in our list, stance and relative motion.

Let's start by figuring out how to convey stance. I originally considered using an existing, ideographic language such as Chinese for rendering the names of stances and strikes on account of the fact that it would be more compact than using plain English. On reflection, however, this seems like a bad idea. Most MA practitioners are hobbyists and have neither the time nor the inclination to learn a new alphabet; if we want this system to be accessible to the average student we'll need to find a better approach than rendering actions in Chinese. Can we render actions using symbols which are either already known to the student or at least easier to learn than Chinese?

If that's what we're looking for, a compact, accessible means of assigning unique identifiers to stances, why not just use acronyms or abbreviations? The sample sequence that we want to record contains two stances, horse stance and half-moon stance; I see no reason at present why we can't abbreviate these to "HS" and "HMS". These are relatively compact representations and should be fairly easy to remember; if need be a brief key/glossary can be included with the transcription to help jog the student's memory. This approach need not be limited to stances either; it should work just as well for blocks and strikes.

So let's see what we've got given the conventions we've adopted so far:

Reading from bottom to top this reads: "horse stance", "left half-moon stance", "right half-moon stance". I've place the abbreviation for horse stance directly on the (implied) center line since it doesn't have left and right variants. So far so good, now we need to convey "move forward", which brings us to the consideration of transitions.

I expect that there's going to be a lot to say about transitions eventually, but right now I'm just going to concentrate on the basics. Before we get into the introduction of specific symbols lets think for a minute about how such symbols should be placed relative to the abbreviations denoting stance. Two schemes immediately present themselves; transitions can be denoted via

• "Decorations" on stance (or other) symbols.
• Dedicated symbols.

I believe that the latter is probably more appropriate in the long run. If transitions are made their own, stand-alone symbols it becomes easier to avoid ambiguity w.r.t. sequence; it's harder to intuitively understand the sequence of actions if transitions are associated with specific symbols. This also seems appropriate from a theoretical standpoint; moving correctly from one stance to the next is just as important as standing correctly. Indicating transitions by means of ancillary notation on other actions makes them "second class citizens".

If transitions are going to get their own symbols then it logically follows that, given our concern with intuitively conveying sequence, such symbols should be placed vertically between stances.

Now let's think about indicating movement. In the martial arts movement is fundamentally about displacing your center of mass/dan tien in a controlled fashion. This makes it somewhat different from dance which, per Choreo-Graphics, is concerned not only with movement from point A to point B, but also the style in which such movement is accomplished. The existing notation is intended to describe specific modes of movement (walking, jumping, gliding, etc.), which makes it overly complex for our present needs. We need to come up with some simple notation for conveying "move thataway".

I vote for using arrows for this purpose, since they're easy to draw and easy to understand. That would give use the following transcription of our sample:

This reads:

1. Start in horse stance.
2. Move forward into left half-moon stance.
3. Move forward into right half-moon stance.

I think that's good for now. Let's recap the rules that we've developed so far.

• Symbols are read from bottom to top, left to right.
• The top of the page always represents the performer's front.
• Symbols are grouped around/on an implicit center line.
• Symbols to the left of the center line indicate left-hand variants, symbols to the right of the center line indicate right-hand variants, and symbols placed on the center line indicate an action with only one form.
• Specific stances, strikes, blocks, etc. are indicated using two- and three-letter acronyms.
• Symbols for transitions are placed in between symbols for actions.
• Displacement of the performer's center of mass is indicated via arrows.

That's it for this edition. Interested parties should comment at will.

---

1 "Horse stance" is common across a wide range of karate/kung-fu styles, but the rest of the phrases may need an explanation. Briefly:

• Half-moon stance: One foot forward, one foot back, forming a stable base. Hips face forward, 50/50 weight distribution, hands on guard.
• Half-moon forward: In half-moon stance the rear foot arcs inward, touches load-bearing foot, and arcs out again at a 45° angle to land in the forward position. The path traced by the moving foot looks like a half-moon/half-circle.
• Quarter-moon forward: Half of "half-moon forward". In horse stance one foot comes in, touches the load-bearing food, and arcs out at a 45° angle to land in the forward position.

2 This analysis is tied to a specific style, but makes use of the general assumption I wrote about in my previous post that there's usually one "correct" way to execute a particular action. Note also that this assumption is probably only valid in the context of basic MA pedagogy; in advanced studies and/or the real world there may be multiple "good" ways to do something.
3 Pp. v - vii
4 P. 64
5 To a first approximation. The actual extent to which forms/techniques are driven by basic applications vs. appearance can vary widely from school to school.
6 P. 96
7 P. 121
8 I believe that the association between reading successive symbols and forward motion is beneficial since, in my experience, forward motion relative to the performer is far more prevalent than backwards or sideways motion.

Nunchucks (Provisionally) Protected By The 2nd Amendment

The The Volokh Conspiracy reports that a judge in Placer County, CA, has ruled that nunchucks are covered by the 2nd amendment. The memorandom for dismissal and the defendant's reply to the DA's arguments in opposition make for interesting reading. The argument is as follows:

1. The 2nd Amendment applies to the states.
2. This includes the possession of arms in one's residence.
3. The 2nd Amendment is not limited to firearms.
4. Nunchucks are a type of arm covered by the 2nd Amendment.

Based on the defendant's reply it looks like the DA countered that the 2nd amendment is not unbounded and thus that the CA statute barring possession of nunchucks doesn't run afoul of the right to bear arms. The reply itself states that the legislature might, for example, ban possession of arms by felons, but that the categorical ban on possession by all persons outside of a self-defense school was overly broad.

The judge apparently bought that line of reasoning and ruled from the bench rather than issuing an opinion. Eugene Volokh assumes that the state will appeal, hence the "provisionally" in the title.

My $0.02: Felon or not, its ridiculous that California made simple possession of a set of 'chucks illegal. You gotta wonder how the legislature came up with its list of dangerous weapons; its a motley assortment of items, common and uncommon, which don't seem to be tied together by a common theme. I mean, really, are "air gauge knives" such a threat that they need to be called out for special treatment?

I wasn't able to find any material relating to the legislative history of the Danger Weapons Control Law¹, but the mere mention of something as obscure as an air gauge knife makes it look to me like the list was probably assembled retrospectively over time in response to particular incidents. Which is just an idiotic way to do things from a design standpoint; every time you arbitrarily ban a particular weapon you just encourage people to find ways to work around the law. For example, here's how "shuriken" is defined:

(11) As used in this section, a "shuriken" means any instrument, without handles, consisting of a metal plate having three or more radiating points with one or more sharp edges and designed in the shape of a polygon, trefoil, cross, star, diamond, or other geometric shape for use as a weapon for throwing.

Does that mean its OK to possess needle type shuriken, since they only have two points? Or what about two double-ended needles annealed together in the center to form a cross, since they really can't be characterized as a "metal plate". Wouldn't it just be easier to focus on illegal actions like stabbing people rather than quibble about what does and does not count as a dangerous weapon?

---

1 Though this article draws a connection between the 1974 NY ban and the release of Enter The Dragon.

Notation: Palingenesis

After prolonged thought on the subject I'm inclined to give up what I've developed so far in the way of MA notation and start anew. The music-staff-like system that I've been contemplating has a number of shortcomings: its limited in what it can capture, the lines of the "staff" make it difficult to read the overlaid symbols, and much of the staff is often completely empty. This latter fact in particular strongly suggests to me that there may be a more compact, efficient, and readable way of conveying the same information.

Rather than reinvent the wheel this time it seems prudent to survey the field and see what's been done in the past. There's bugger-all in the way of notation specifically devoted to the martial arts, which is the primary reason that I took up this project in the first place, but there's been a surprising amount of work done in the related field of dance notation. In particular Ann Hutchinson Guest has written a book, Choreo-Graphics, which compares a number of historic and contemporary systems, thus saving me the hassle of trying to do that myself.

Intent

Early on in the book Ms. Guest makes the following observation:

What in one dance notation system is seen by some as an advantage is seen by others as a disadvantage. Is the aim of the system to be simple, providing a memory-aid for those who know the style of movement? If so, it is likely to be found efficient by those who need scientific accuracy. But if great care is taken to analyze and record movement precisely, many people may find such a system too complex, requiring that too much attention be paid to the nuances of movement and consequently too great a need to analyze. For an objective evaluation of any system, one must know the purpose that system was intended to serve. Different systems have different aims.¹

There is an unavoidable trade-off between completeness and complexity: a simple system may fail to capture all the necessary information, while a system which captures every detail may prove unwieldy. The appropriate balance between the two is determined by the ultimate aim of the system; what, exactly, is it trying to accomplish? I've got to answer that question before proceeding.

Rereading my original post on the subject of notation I believe that the intent of this hypothetical system that I'm devising can best be summarized by the phrase "be better than English". The MA books in my possession describe forms and techniques in English, occasionally augmented by pictures, but these descriptions suffer from various shortcomings that I lay out in the post. I'm trying to build a system which can do a better job conveying these forms/techniques to the same audience with an equal (or greater, if possible) level of detail.

Let's think about the audience for a second... who are they? The intended audience is a function of the topic under discussion. Manuals associated with a particular school/studio are directed at that school's students/staff. Books concerned with a particular style outside the context of a studio usually assume a reader with a generic MA background. The same can also be said for materials which deal with general MA theory.

What's useful to note here is that, while the assumptions regarding the audience's background may change, the level of detail with which techniques are presented tends to remain fairly constant. Few, if any, of the materials in my collection attempt to assemble a scholarly, scientific record which accurately captures every detail of a performance. Rather, they tend to follow a practical, "how to" format that assumes the reader understands how to execute specific actions, and focus instead on documenting the sequence in which the actions are to be performed.

It follows from there that, since the ultimate goal is to provide a superior replacement for these types of materials, I need not worry about documenting the subtle nuances of a performance. The system under development should assume that the audience understands the basic building blocks of a form or technique and should instead seeks to convey in as much detail as is feasible the means by which these actions are strung together. This approach, while seemingly optimal, necessarily hinges on an accurate understanding of the audience's background, which raises another question: What can the audience be assumed to know?

The Common Body Of Knowledge

A community of practitioners, of any discipline, shares a set of conventions, techniques, definitions, best practices, and so on. This common body of knowledge provides the informational background against which the community's activities take place; all members of the community can simply be assumed to know certain things. Ms. Guest notes how, in the field of dance, this fact can be leveraged to produce more compact and readable forms of notation:

An obvious device for anyone wishing to jot down rapid notes is use of a letter (or letters) for the name of each step. If the steps themselves and their manner of performance are widely known, such abbreviations suffice. All that needs to be recorded is the sequence in which the steps occur in the dance,¹

It seems to me that this holds true for the martial arts as well. Within a particular school or style phrases such as "front punch" or "roundhouse kick" represent well-defined actions with an ideal form of execution; they need no further explanation³. An MA notation system can be made more efficient by omitting such redundant details provided that the nature and composition of the shared body of knowledge is well-understood. So I ask again: What can we assume the audience already knows?

In my post on functional requirements I hypothesized that any form/technique can be broken down into a sequence of postures, strikes, and acts of attention⁴. I believe an MA system can safely assume that, within the confines of a particular school/style, each of the following types of performance elements have a preferred, "correct" method of execution:

• stances
• blocks
• punches, kicks, and other strikes

Having set up this fundamental assumption we'll need to be vigilant for places where it may not hold, providing some mechanism for dealing with deviations from the rule as they arise.

Summary

In conclusion, this system which I've set about devising is intended to provide a practical record of forms/techniques targeted at the practitioners of a particular school/style. What I hope to develop is a framework that works well for any style which the practitioners thereof can then customize as they see fit.

---

1 P. xv
2 P. 1
3 Or, if they do, the directions for a form aren't the place to do it.
4 At least in the case of styles/arts centered on atemi waza. That's what I have the most experience with; I can't say whether my observation generalizes to arts which place more emphasis on locks/throws.

Notation: Conveying Posture

In my previous post I asserted that a notation system must be able to describe the posture of the performer. It follows from there that we need to have a better idea about what we mean when we say "posture".

Most martial arts systems have a set of canonical stances, the existence of which I'm going to take as a given. We need not concern ourselves with the question of what, exactly, this set contains; it is sufficient to know that each stance has a label with which it can be uniquely identified. Stances, like strikes, are assumed to be atomic; the statement "performer in X stance" is assumed to need no further explanation.

"Posture", however, entails more than just providing information about what stance the performer is in. There's also the question of orientation i.e. what direction is the performer facing? Which brings us quite quickly to the question of coordinate systems and absolute vs. relative positioning.

Let's step back a moment and consider the fundamental purpose of this exercise, namely "describe forms better than English". Flipping through some of my manuals I see two themes when it comes to describing the orientation of the practitioner. Most of the time the orientation of the practitioner is given implicitly using relative coordinates e.g. "turn left", "turn right", "turn 180°". Less frequently there are instructions using absolute directions such as "face front" or "facing X o'clock",

It would seem, then, that there is a need for both absolute and relative positioning. It is necessary at the beginning of the form to provide some sort of absolute direction but, once an absolute position is established, all future orientations can be described using offsets from this initial position. This suggests the following features:

• The initial orientation of the performer is specified using absolute coordinates. The specific absolute coordinates are up to the user; they need only be sufficiently understood/defined within the context of any particular school.
• Transitions between postures are specified using relative offsets.

In order to specify relative coordinates we're going to need to adopt a specific coordinate system. W.r.t. atemi waza it seems like cylindrical coordinates make the most sense¹ since typical changes in relative orientation can be described using a single number.

Stance and orientation are sufficient to describe a static posture, so the next item to consider is how to describe transitions between postures. Here are some examples taken from one of my manuals:

• Draw left foot into cat stance facing left.
• Right foot half-moons forward.
• Right foot steps into horse stance facing 9 o'clock.
• Right foot steps back pivot clockwise 180 degrees and draw into right cat stance facing 3 o'clock.
• Hop forward toward 1:30 o'clock on left leg, with right leg held in flamingo position.
• Step out with left foot, jump and pivot 180 degrees, landing in low right twist stance facing 3 o'clock.

My heart is filled with dread and despair... those last couple look devilishly difficult. I'm not particularly worried abou the jumping and turning; those things, I believe, can be described fairly succinctly. What's going to be hard to deal with are all the fiddly bits about foot/leg positioning. Might it be the case that written English is the best medium for conveying such instructions?

Let's look at the first item in detail. Given what we've already discussed I believe we can readily handle concepts such as "cat stance" and "facing left", but we don't as of yet have any way of designating the left foot or conveying the action "draw into". We definitely need a way to designate limbs in some fashion, but I'm doubtful at this point whether its necessary to be able to convey a concept as complicated as "draw into". It would be sufficient, at least for a first cut, if we could render something equivalent to "turn counter-clockwise 90° into left cat stance by moving the left foot".

Based on the above the notation system will need some means of conveying the following concepts:

• turn
• left/right (hand, foot, version of stance)
• clockwise/counterclockwise
• (Change in) position of dan tien (low/lower, high/raise, very high/jump, move forward/backward/left/right)
• Foot movement/involvement in transition between stances.

---

1 Whereas if we were describing throws spherical coordinates might be more apropriate.

Notation: Statement Of Functional Requirements

Someone is paying attention after all, which means I should take up the subject of notation from where its been sitting, collecting dust, for the better part of a year and a half¹. My last activity in this regard was to start thinking about Jesse Crouch's comment regarding the need to be specific about the target of a strike. This led to a series of posts (1, 2, 3, 4) investigating potential targets on the human body, also incomplete due to time having its way with me. A secondary aim was to see whether there was any evidence supporting the existence of critical/pressure/vital points i.e. targets which have effects above and beyond what can be explained by simple mechanics/physiology (short answer: "no").

So, what does that mean for the development of a notation system? My main takeaway from that line of inquiry is that its important to be able to indicate the target of a strike, but that specific targets are going to be a matter of debate/preference. So, rather than prescribing a specific list of targets, a useful system of MA notation will provide the means for unambiguously specifiying a target while allowing the user to fill in definitions for emself.

I have the feeling that my previous attempts to date have fatal shortcomings, so now is a good time to review the requirements that have been unearthed to date:

• The fundamental problem which needs to be solved is how to represent a MA performance in 3-space in a compact, readable format.
• The focus (for the time being) should be strictly on atemi waza; locks and throws are different enough that they might need another system entirely.
• A notation system should be art-agnostic. There seems to be enough commonality among the various arts (at least with respect to atemi waza) that a single notation system can accommodate the needs of all without getting bogged down in specifics.
• Individual strikes should be considered atomic; it's proven too cumbersome to try to describe the mechanics of a single strike in detail. If such description proves necessary there are pre-existing systems (notably Eshkol-Wachman Movement Notation) which can be pressed into service.
• A performance can (tentatively) be broken down into a sequence of postures, strikes, and acts of attention. Thus a notation system needs to answer:
  • What is the posture of the body?
  • What strikes are delivered?
  • What are the targets?
  • What is the relative timing of the strikes?
  • Where is the performer's attention directed?
  • How is the transition made from one posture to the next?

The focus, I believe, should be on answering those questions subject to the constraints outlined above. While it make take a little bit of doing to come up with a workable system I see no red flags at this point to make me think that such an endeavor is fundamentally impossible.

---

1 Funny how having a kid can really suck up all your spare cycles.

A Brief Meditation On Pain And Injury

I watched Ninja Assassin last night. It was entertaining in an overwrought and hyper-violet fashion, but Rain gives one of the least convincing MA performances I've seen in awhile. All the money they spent on CG-enhanced fight scenes couldn't hide the fact that he handles a sword poorly; the numerous, nameless extras they got to portray the assassins-in-training were far more convincing.

But that's all secondary; what I talk about is the role of pain and injury in training. The movie's flashback sequences depict the routine infliction of non-lethal injury as a standard part of the assassins' curriculum. Much of what is shown would be, I suspect, counter-productive in real life (e.g. severely lacerating the soles of a student's feet), but it got me thinking about pain and injury in a more general fashion.

People, as a general rule, tend to avoid pain. This is true of martial artists as well; we may be more accepting of pain as a consequence of training, but we don't seek it out either. The question that arose in my mind, after viewing the film, is how much we avoid pain for its own sake and how much we avoid it as a harbinger of injury?

The typical pain of training is transitory; you'll feel fine half an hour after practice. Knowing this makes such pain easier to ignore, embrace, and/or endure since the pain is, in some non-trivial fashion, "all in your head". But the pain that accompanies real injury is more than that; it signals that you've damaged yourself to some degree and may not be fine when you get home. Injury has real, long-term consequences, so it's generally wise to pay attention to such signals.

The complication, from the perspective of training, is that there isn't a bright dividing line between pain that should be tolerated and pain that should be heeded. Instead there is a vast, grey swath of land that varies considerably from individual to individual, one which we tend to navigate, with good reason, conservatively. I suspect that most people never develop a sense of the amount of pain they can, or should, tolerate in practice. They don't know how much of a beating they can take (or deliver) and still come out OK on the other side.

As a consequence they may unduly restrain themselves, forgoing actions which might have been beneficial to them. Or, less commonly, they may overdo it and unintentionally injure themselves. Neither outcome is desirable, which demonstrates the value in being able to accurately judge the limits of endurance.

Which brings us back to the role of pain and injury in training. A comprehensive curriculum will assist the student in navigating the "vast grey swath" that I referred to above. For those who consent to it this could quite legitimately include the infliction of various degrees of physical injury. Such a practice would serve two functions:

• Acclimatize the student to pain in general.
• Assist the student in establishing the boundary between minor and serious injury.

It's not terribly difficult to see how this might be achieved in practice. The training regimen might, for example, include the use of makiwara, decreasing the amount of padding as the student progresses. Sparring could be made more realistic by reducing/eliminating padding and/or relaxing rules about contact. And so on, all with the idea of gradually increasing the amount of superficial injury to which the student is exposed.