Relative to its forerunner Jujitsu, Judo demonstrated that the ability to practice techniques against resisting opponents is far more important than the inherent "potency" of a technique. But there's a feedback cycle between these abstract, "Platonic" techniques the humans who employ them:

  1. "Live" training is required in order to reliably apply non-trivial techniques, especially sequences of techniques, against resisting opponents.
  2. The greater the injury potential of training a technique (not the technique itself) against a resisting opponent, the fewer training partners will be available. This generalizes to whole systems, and varies over time with technology.
  3. When few people train a technique against active resistance, less total brain-power is dedicated to understanding and perfecting the technique. This is the corollary to Linus' Law in software, "given enough eyeballs, all bugs are shallow."

In the context of martial arts techniques, the more safely and sustainably you can train a given technique, the better both the technique itself and your implementation of it are likely to be. As I said to above, whole systems that include only techniques that can be safely trained against a high level of resistance will have two features missing in other systems:

  1. Practicioners are more likely to successfully employ the system, on-demand.
  2. Techniques are more likely to be "debugged" and refined through many repetitions by people of various sizes, shapes, and backgrounds.

So here's the stronger claim: Pressure-tested techniques yield effective practicioners. Pressure-tested practicioners yield effective techniques.

Part 2: How to Break the Cycle

This is a cycle that should be embraced, cultivated, and amplified at every opportunity. Of course, there are several efficient ways to sabotage it.

History over Function

Dogmatism may prevent "fixing" or improving inherited techniques. My understanding is there are at least a few Aikido schools very strictly oriented toward preserving the art as Morihei Ueshiba taught it. In this case, even if it's present, pressure testing isn't allowed to refine techniques.

My first formal education in defensive handgun shooting was in 2001, but despite the growing dominance of the Isosceles Stance, I first learned the Chapman Stance. I'm sure there are still some instructors, either too lazy to learn a better technique or too sure that Jeff Cooper had everything figured out, that teach it today.

However, it seems that the practice and outcome of pressure testing tends to discourage historical orientation, when it conflicts with efficacy.

Partial Pressure-Testing

Imagine a system where only some of the techniques are pressure tested, and the rest are only rehearsed with a compliant partner. This describes the sparring I did a long time ago in Kempo, where closed-fist punches and kicks were common, but other strikes, throws, joint locks, and chokes were absent.

Some who prefer this approach claim that the deadliness of the techniques is more important than the pressure testing, and that practice with a compliant partner is sufficient. It's impossible to make any controlled test of this claim, because those techniques must only be used at full-force and full-speed against a criminal assailant. In a more haphazard way, security camera evidence is slowly piling up and allowing at least a stochastic, hit-or-miss accumulation of evidence.

Artificially Restricted Pressure-Testing

When the rules of pressure testing prohibit or discourage techniques that can be trained safely and have self-defense value (for whatever reason) they artificially limit the feedback cycle. Stephan Kesting relates one example from BJJ (I lack firsthand knowledge of both safety and street-efficacy of the mentioned techniques):

...the IBJJF -- the International Brazilian Jiu-Jitsu Federation -- has really clamped down on what’s legal in their BJJ competitions... Essentially they have banned the most effective leglock (the heel hook) and one of the most effective leg control positions (the reap)...

A similar example can be taken from IDPA regarding reloads, which, although perhaps well-intentioned, significant evidence suggests lacks any parallel in actual civilian self-defense.

3.4.6 Shooters may not perform a reload which results in a magazine (or loose rounds) being left behind after there was an unfired cartridge in the chamber, magazine, or cylinder at the time the reload was initiated. When done intentionally, this is commonly known as a "speed reload", but doing this unintentionally is still illegal and will result in a Procedural Error penalty being issued.

Because banned techniques will be practiced by competitors less frequently (or not at all), those folks are less likely to be able to successfully employ "bad" techniques on-demand, and the "bad" techniques are less likely to be refined and improved over time.