Yeah. But let's say we go back to the lachlan guiles's leg lock idea.
If you are training live. And there are people who are good.
Then they are going to be the ones developing the activity.
It doesn't have to evolve but people may find themselves dominated by people who do evolve.
Absolutely, but the additional considerations have to do with the nature of the art we are practicing.
We want to develop the skills used by historical sword fighters. But we can't practically/legally/ethically fight each other for real with real swords. We can only practice a relatively safe approximation in the hopes that it is close enough so that the skills would carry over in the event that we were magically transported back into the past.
That's where these other aspects of the art's development come into play. We don't want to end up just getting better and better at competing in a sport which is completely disconnected from the reality of actual historical combat. So we have these additional reality checks (not replacing competition performance, but supplementing them):
Did the people who actually fought with swords for real ever use a particular technique or tactic? If not, there's likely some good reason. Did they only use that technique or tactic in a particular situation or culture or time period? Then there's probably a reason for that too, which we might not figure out just from our tournament results.
Can we actually cut solid objects effectively using the same techniques and body mechanics that we use in sparring? This can take a lot of practice to get good at and it can be difficult to reliably judge in a tournament setting.
Are the sparring rules that we use to approximate actual fighting the best rules we can use for that purpose? It's not hard to end up with a situation where someone gets really good at winning tournaments with tactics which would likely get them killed in a real sword fight*. So we research and experiment in order to get the best approximation(s) we can for developing a skill that we will probably, hopefully never have a chance to test in real life.
*(Alternately, there were also historical fencing competitions which were conducted with non-lethal intent. So another way to run tournaments is to research how those were carried out and match them as closely as we can within the parameters of modern safety concerns.)