Algodoo

As far as I know, the runners are exactly the same but run at varying speeds. This is a puzzle that I haven't figured out. If anybody can figure this out and explain it in a comment below, then I will rate the scene of their choice a 10.

Let me think:
Are you asking why deterministic program give a different results? I guess it is non-deterministic because of smallest differences like rounding a float number, for example 1/3 to 0.3...34 in memory.
Maybe try searching for a "double pendulum" or "butter fly effect", it may have some connection with it
Did you tried different solvers of joints and test if you get more deterministic results? Or maybe try slowly walking instead of jumping, set simulation frequency high etc. to make collision more precise
I'll subscribe to this scene, I'm interested what you find out!

Q: Are you asking why deterministic program give a different results?
A: Yes.

Q: I guess it is non-deterministic because of smallest differences like rounding a float number, for example 1/3 to 0.3...34 in memory.
A: You may be right, but the differences in running speeds seem too large for rounding errors.

Q: Maybe try searching for a "double pendulum" or "butter fly effect", it may have some connection with it:bonk:
A: It could be something like that, but each man has his movements tied to a global master angle.
A1: I was wrong, each man calculates scene.my.angle in his head. I changed that to have one guy do the calculation. No improvement.

Q: Did you tried different solvers of joints and test if you get more deterministic results?
A: No, but it sounds like a good idea.
A1: Tried that, no improvement.

Q: Or maybe try slowly walking instead of jumping?
A: No, didn't try that.
A1: Walking slowly improved it quite a bit, so that makes the timing of the angle calculation suspect.

Q: Set simulation frequency high etc. to make collision more precise.
A: Yes, I had tried that early on, but did not see a big improvement.
A1: Tried it again. No improvement.

Q: Maybe the men are sliding. I will try increasing foot friction.
A1: Set friction to +inf, no improvement.

I think the random changes in velocity are mostly caused by bounce and slip between their feet and the ground. Even if you make Restitution of the plane and all of the lower legs of all the men zero, and make friction for all that infinite, there is still some bounce and slip that occurs. I was able to reduce the bounce and slip to near zero by setting the Y value of velocity damping for all men to 1.0 (maximum damping). I did that by selecting all body parts of all the men and then changing Y velocityDamping to 1.0. Then when I ran the scene from the beginning, the men stayed in fairly close "lock step" with each other. When I set the Vel damping back to zero, the mean deviation of their velocities began to increase.

Try it and see if you agree.

EDIT: The more I thought about this the more I think the issue is caused by bounce alone. With friction set to infinity, I doubt that there will be any slip at all. But when random bounces occur, the time that each foot is not in contact with the ground will vary considerably with each bounce. Since each bounce duration is somewhat random the velocity of each man will be randomly changing.

It looks like bounce is the biggest factor, followed by sim.frequency, and then friction. Since you solved the puzzle, I will rate the scene of your choice a 10. Thanks.

Actually, I get more pleasure from solving a puzzle than I do from receiving the reward, and I have no concern for ratings, but thanks for the offer.

alternative race hmm

Orange got broken
Blue got chopped arm and leg
Green got chopped leg
Yellow got damaged
Magenta hopped and got hurt
Red got tired