In terms of Physics there are only certain valid options to choose. I base all of my techniques against scientific training and testing and re-training and more testing. In the end I take the things that work the highest percentage of time and follow the training methods that achieve the most productive and efficient results.
To determine the "best" punch in my world you take ALL the punches and put them in a test. The testing is done against what you want to strike. Since I focus on self defense only, I want to strike the human body. With anatomy we know there are good spots to hit with a fist and then there are not so good spots. Now that we have the striking areas defined we strike with all of the punches and measure a) damage we receive by punching and b) damage we deliver. When you take the results you have one or two that stand out. Out of the one or two I choose the most versatile punch and start training it. While this is not the most perfect process it does deliver the highest yielding percentage shots - which is what I bet on.
If you had to run out into the woods and you could only grab three things you would choose the tools that you can use in the most situations and especially the most critical need areas such as survival.
There may very well be 40 different ways and types of punches, but I really only need to train one to an above average level to be more effective than most. If you are going to shoot aim for the body.
Now you're talking about something entirely different from what you originally proposed. If you want to design a curriculum based on what you think is the most effective way to punch, kick, or whatever then you're just creating your own art. No problems there. Other people have done the same for centuries. There just isn't anything "generic" about it. For some reason, whenever different people attempt to design the "most effective" curriculum, they come up with different results.
The testing and design process is likely to be more tricky than you seem to realize. Suppose we do a bunch of tests and determine that punching method A delivers the most impact. Great! Only next we discover that the best way to punch without being counter-punched requires punching method B. Uh oh, it turns out that the best way to punch without being taken down requires punching method C. The best way to punch and follow up with an immediate kick requires punching method D, and so on. At some point you have to decide what your priorities are and how to weigh the trade-offs. Those decisions are the beginnings of a style.
You'll also start to realize that you can't just test each technique in isolation. Techniques arise out of underlying body structure and mechanics. Suppose your tests show you that the best techniques for evading an attack come from Bujinkan taijutsu while the best techniques for landing a punch come from Wing Chun. Now you've got a problem. You can't apply effective Wing Chun punching from a taijutsu stance and you can't apply effective taijustsu evasion methods from a Wing Chun stance. The body mechanics just won't work. Whichever techniques you end up deciding to include in your curriculum, they need to work together in terms of the underlying body structure, body mechanics, tactical doctrine, etc. When you decide how to do that, you're creating a system which is specific and not generic.
You should also remember that, contrary to what many people like to claim about the design of their systems, it isn't really practical to be all that scientific in the process of testing your decisions. How do you "test all the punches"? Just try them all? What if you're just more skilled at one punch than another? You'd have to take a bunch of beginners and train each person in a different punching method for the same number of hours before testing them. Of course, you can't have just one person per sample group. Maybe one person would be naturally stronger or more talented and so the punching method you assigned to him would seem to be the best. You'd need to randomly assign a fair number of individuals into each experimental condition and carefully control the training so that all other aspects of training (intensity, number of hours trained, supplemental exercises, etc) were exactly the same. Then you'd have to design your actual tests. How do you determine the "damage received" by punching? Do you match all your test subjects up in repeated bare-knuckle fights and observe the results? What rules do you enforce in these fights and how might those affect the outcomes?
In the end, unless you are a ruthless dictator with the power to draft thousands of your subjects into training and fighting gladiatorial matches for your amusement, you aren't likely to find a "scientific" answer as to the best techniques for a martial art. (That's assuming there are "best techniques" as opposed to "best techniques for a given individual in a given situation at a given time".)