This coprime stategy for getting a "random" segment-to-work-on order is interesting.

Why does it make a difference if a sidechain is frozen?

As moving of the sidechains can change the number of snap positions, I reccomend to change the the line:

do_sidechain_snap(idx,r)

to

if r<=get_sidechain_snap_count(idx) then do_sidechain_snap(idx,r) end

to prevent an error.

Thanks for the suggestion Crashguard but I believe this issue is already addressed through the use of the boolean variable improvement_made.

(And if you didn't freeze the sidechain you'd always return to the starting position)

...the "bad argument" for snapping already doesn't occur anymore?

I do see it occasionally on this script and have no idea why: suspect a possible bug in foldIt.

Thanks Crashguard303, your suggestion worked. "r" is being used without being correctly bounds checked.

I don't know why anyone would intentionally introduce a vulnerability into their code.