There is a rare, but interesting pattern that I illustrated in puzzle 50 on my puzzle site, and it occurred not once, but twice in this game. And both players missed it.

First, Steven could have put the game away early with |20.s11.

Later, it looks like technolion could have come out victorious with |37.h7. (That one isn't as clear because of all the gaps and the distance to the top, but it looks like it works.)

Yes there are two main branches:

37.h7 38.j10 39.g9
-- 40.i12 41.h14
-- 40.h11 41.g13

37.h7

Hm, this looks like a bug. I click on move 36 in the move list, then this second instance of move 37 in this post, and I get a bizarre branch 39h11 40g13 41h7

38.j6 39.f8 40.h5 41.e6 42.e2 43.e4 44.d4 45.c5 46.f4 47.b3

I should probably string moves together on spaces only, and not carriage returns. Also, I should put up a warning if the first move number in a sequence is too high to make any sense in context. I'm still not sure what to do if the move numbering isn't sequential; there are a lot of ways one could make a typo and I'd like to be graceful about it, and it's a tricky thing to detect using regular expressions. I'd have to change a lot of code. But I am sure that no matter what I do, Twixter will still stumble upon yet another way for it to violate his expectations... he's darn good at that.

Anyway, Twixter, it sounds like you're agreeing that 37.h7 seems to work for White here.

Yes allow me to reiterate the second branch:

37.h7 38.j6 39.f8 40.h5 41.e6 42.e2 43.e4 44.d4 45.c5 46.f4 47.b3

and white wins.

So, If I had placed some text between the two main branches in my previous post, it would have behaved correctly?

"Correctly" may be a loaded term, but yes, it would have behaved much better. I'll treat it as a bug, though, and fix it so that carriage returns are treated as intervening characters, and then maybe edit this conversation so that it doesn't cause confusion.