I recently attended a jujutsu seminar led by a senior instructor ("Shihan"); let's call him 'T'. I had a good time, and learned a lot; Shihan T is an excellent martial artist, and a very good teacher. However, on several occasions, I found my mind wandering over to the subject of science – how it's done, and what purpose it serves.
You see, Shihan T is one of those people who likes to explain things in terms of ki (also known as The Force) and meridians. As a geek (the science/maths kind, not the chicken-guillotine kind), I am neither willing nor able to accept explanations of that sort. And yet, there's no doubt that Shihan T knows his stuff. He was throwing us around the room like a giant bear throwing rag dolls (it's OK – it's a friendly sort of bear, and the rag dolls know how to breakfall). He could break down any of his techniques and explain them in his own terms. Even if his explanations had little or no scientific basis, you could test many of his statements on a human guinea pig; by and large, their reactions were consistent with his claims.
The way I see it, the purpose of science is to build models that let you make predictions about phenomena in a system (typically, the universe). Science never asks the question 'why', and only asks 'how' relative to the model you're working in. Newton observed that things falling towards the ground tend to accelerate at 9.8ish m/s2. Shihan T observed that if you point such-and-such joint towards such-and-such meridian, your guinea pig will begin to wriggle about on the mat and scream strange insults at your ancestors.
There's no straightforward method to decide how best to model a system. (This is analogous to the feature selection problem in machine learning.) Even if some characteristic of the system seems to be relevant, it might just be a random pattern that emerges in your observations ("training data"), or have some correlative (rather than causative) link with the phenomenon you are trying to predict. Thus, with a bad model, you could thus come up with a hypothesis that matches your training data perfectly, but does an awful job predicting new observations ("test data") that you make after you formulate your hypothesis.
It is possible (and, I think, very likely) that this is how Shihan T's explanations work: over the course of several hundred years, people observed many of the ways human beings can move (or be physically manipulated), and came up with a model that quite accurately matches their training data. People rarely discovered new ways to manipulate the human body, so there was no real test data to validate the hypothesis with. On the rare occasions that test data was found ("Hey, I didn't know the wrist bends that way!"), the models were simply revised – a meridian moved a couple of inches this way, a pressure point added over there. A scientist or machine learning expert might call this cheating, since a model derived in this manner will not perform well on new test data – but for the martial arts, there really isn't much more test data. People in the foreseeable future will have the same joints they have today.
This could also be why certain non-traditional medicine systems (or, rather, super-traditional medicine systems) work better than they might be expected to. Any system that's slightly more effective than random chance has a small chance of catching on.
I do think that the models currently in use in the scientific community are far more effective than ones used in the past. We've had far more training data to learn from, and far more test data to validate with. Most verifiable, falsifiable predictions that can be explained in terms of ki can also be explained in terms of anatomy and physics. "To project your energy," Shihan T told us while demonstrating a throw, "the first thing you have to do is look at your partner." And then he turned to do so. In doing so, he turned his head in the direction his torso was facing, and his spine straightened out; this gave him the structural alignment he needed to perform the technique. (Every try lifting dumbbells with a bent spine? Don't. It'll be hard, and you'll hurt yourself.)
My point isn't that 'ki' is legitimate science, or can explain things that legitimate science can't. My point is that even unscientific investigation can sometimes (through years of Darwinian trial and error) provide usable models. Of course – not all models are equal. Many are blatantly self contradictory, and many others simply don't make falsifiable statements. Ki escapes both these traps to some extent by being an extremely nebulous concept – it's hard to even make a clear assertion about it, let alone find two assertions that contradict each other. It means entirely different things to different people. Obi-Wan Kenobi described it as "an energy field created by all living things. It surrounds us, penetrates us, and binds the galaxy together." I've heard other people simply define it as 'structure'.
You see, Shihan T is one of those people who likes to explain things in terms of ki (also known as The Force) and meridians. As a geek (the science/maths kind, not the chicken-guillotine kind), I am neither willing nor able to accept explanations of that sort. And yet, there's no doubt that Shihan T knows his stuff. He was throwing us around the room like a giant bear throwing rag dolls (it's OK – it's a friendly sort of bear, and the rag dolls know how to breakfall). He could break down any of his techniques and explain them in his own terms. Even if his explanations had little or no scientific basis, you could test many of his statements on a human guinea pig; by and large, their reactions were consistent with his claims.
The way I see it, the purpose of science is to build models that let you make predictions about phenomena in a system (typically, the universe). Science never asks the question 'why', and only asks 'how' relative to the model you're working in. Newton observed that things falling towards the ground tend to accelerate at 9.8ish m/s2. Shihan T observed that if you point such-and-such joint towards such-and-such meridian, your guinea pig will begin to wriggle about on the mat and scream strange insults at your ancestors.
There's no straightforward method to decide how best to model a system. (This is analogous to the feature selection problem in machine learning.) Even if some characteristic of the system seems to be relevant, it might just be a random pattern that emerges in your observations ("training data"), or have some correlative (rather than causative) link with the phenomenon you are trying to predict. Thus, with a bad model, you could thus come up with a hypothesis that matches your training data perfectly, but does an awful job predicting new observations ("test data") that you make after you formulate your hypothesis.
It is possible (and, I think, very likely) that this is how Shihan T's explanations work: over the course of several hundred years, people observed many of the ways human beings can move (or be physically manipulated), and came up with a model that quite accurately matches their training data. People rarely discovered new ways to manipulate the human body, so there was no real test data to validate the hypothesis with. On the rare occasions that test data was found ("Hey, I didn't know the wrist bends that way!"), the models were simply revised – a meridian moved a couple of inches this way, a pressure point added over there. A scientist or machine learning expert might call this cheating, since a model derived in this manner will not perform well on new test data – but for the martial arts, there really isn't much more test data. People in the foreseeable future will have the same joints they have today.
This could also be why certain non-traditional medicine systems (or, rather, super-traditional medicine systems) work better than they might be expected to. Any system that's slightly more effective than random chance has a small chance of catching on.
I do think that the models currently in use in the scientific community are far more effective than ones used in the past. We've had far more training data to learn from, and far more test data to validate with. Most verifiable, falsifiable predictions that can be explained in terms of ki can also be explained in terms of anatomy and physics. "To project your energy," Shihan T told us while demonstrating a throw, "the first thing you have to do is look at your partner." And then he turned to do so. In doing so, he turned his head in the direction his torso was facing, and his spine straightened out; this gave him the structural alignment he needed to perform the technique. (Every try lifting dumbbells with a bent spine? Don't. It'll be hard, and you'll hurt yourself.)
My point isn't that 'ki' is legitimate science, or can explain things that legitimate science can't. My point is that even unscientific investigation can sometimes (through years of Darwinian trial and error) provide usable models. Of course – not all models are equal. Many are blatantly self contradictory, and many others simply don't make falsifiable statements. Ki escapes both these traps to some extent by being an extremely nebulous concept – it's hard to even make a clear assertion about it, let alone find two assertions that contradict each other. It means entirely different things to different people. Obi-Wan Kenobi described it as "an energy field created by all living things. It surrounds us, penetrates us, and binds the galaxy together." I've heard other people simply define it as 'structure'.
