NXT, if you had 100 attempts at a 10 foot putt, how many do you hit so hard that even a dead center putt doesn't go in? I don't think your distribution is very accurate. I think if you plotted some sort of density over your red rectangle, you'd find that a vast majority of that distribution is going to be centered around the hole (not to mention with both plotted on the same graph, it's not properly centered for the breaking putts, making it look like there are waaaay more putts that are hit too hard).
I also have to admit that I don't really know what I'm looking at with the blue and red distributions you'd drawn up. I get the red part, but the blue isn't clicking with me. I don't think you can just plot them together like that in the manner you did. If that is supposed to be an image of what you had boxed up in the red rectangle (and rotated 90deg counterclockwise), it's not at all an accurate portrayal of distribution because the graph you pulled it from isn't distribution of putts...it's just the combinations of speed and direction that work from a given distance, which we've already agreed does not equal the actual probability of making putts because the distribution of putts is not completely random over that area.
You need something like the following:
So that shows the window of an approximate
AND ARBITRARY THAT I HAVE JUST NOW MADE UP distribution of speeds for an attempted 10 foot putt. You admit that if the speed is correct, the window for missing left/right is so wide that a golfer would be almost physically incapable of missing their line so bad that they're missing out on being able to use the whole window (which logically follows that it means if they get the speed right, it's a guaranteed make). So the probability of making the breaking putt is determined almost solely on the probability of getting the speed right on a known line that works.
We would need to see the distribution of launch angles as well. Then we can combine the appropriate probabilities together to get the ACTUAL probability of the overall putt, for each type of putt.
Now, the actual distribution from data (which I admittedly don't have) might be a much flatter bell curve where the misses are more evenly spread out across the entire range. Maybe only 2% of the putts end up close to the target speed...this along the lines of the the same sort of thing you guys were talking about earlier when explaining to ship the "most likely" stuff...I know you know this NXT, I'm just stating it all to be clear to everyone.
Hell, it might not even be an even distribution. It might be skewed to the left, to be a heavier distribution toward the target speed, where there are very few misses that are short and most of the misses that are very long are far and few between. If you looked at this distribution solely from a minimum and maximum view, it would not accurately reflect the distribution and it would in fact be very misleading. Just stating the mins and maxes don't solve this problem.
Here's a quick look at what I mean:
So anyway, as straight putts get longer, the left to right window becomes smaller and smaller, while the speeds maintain most of their "make windows" (granting that flat putts have a much wider speed window than breaking putts in general). The problem is not knowing the distribution of launch angles OR speeds for putts. Again, my intuition tells me that the combination of getting speed right on breaking putts (which we've agreed has a very high probability of "being on the correct line" as the nature of "having the right speed") is greater than the combination of hitting the speed window AND the direction window of a flat putt of the same length. Obviously, I don't know at what length the breaking putt would take the lead, but I really think it's going to once I get the correct data to be able to crunch some numbers.
The real point is, we don't know. Where can I find stats? Do you think Pelz would send me some basic data if I emailed him?