...time for a function, which returns the mutable segments.
At the moment, I do it this way that I define those in a table, making a mutable check needless for a single puzzle.

Must have been somet impatient people, which abandoned the mutable test, if I look at the average impact.
If you do mutating with recent best, worst case is an eqal result like before.
A vote again for changing the way how impact is calculated.
http://fold.it/portal/node/988385

I'm not sure it matters but the snap routine is using the index into the mutable segments rather than the segment pointed to. I didn't catch this for quite some time.

for iSidechains=1, foldit.GetSidechainSnapCount(i) do
foldit.SidechainSnap(i, iSidechains)
fsl.WiggleAll(iterations)
end

Should be

for iSidechains=1, foldit.GetSidechainSnapCount(segment) do
foldit.SidechainSnap(segment, iSidechains)
fsl.WiggleAll(iterations)
end

or so I believe.

I made major changes for version 3.02 fixing the bug Gary mentioned, which actually uncovered a few other problems. Should work better now.