I finally found the post I was looking for:

<Aug-29-12
PinnedPiece: Just curious...how bad a move is h5? Does that put black's hackles up? or easily defended?

click for larger view

Seems to me it causes black some serious decision -making. ....Rh8 is not much of an answer, I don't think. ...Qb7 seems to be ignoring our intent. >

After the train wreck, I did a good job publicly of not freaking out, and privately of not throwing my man-cave appliances into one another. So the rebuilding process began with my computers still intact.

<26…h5 27. ♗xg4 hxg4 28. ♔g2> There was no other way for White to untie the position. 27. Kg2 would have transposed at best.

Black had a laundry list of computer options here, but they were all pointless or led to Qxe5 anyway, so it wasn’t long before we were planning exclusively for:

<28…♕xe5 29. ♖hf1> At long last! Black regains the gambit pawn and White activates the last piece.

Here was another greatly annoying list of options for Black. To me, that will always be the scariest thing in these Challenges. The most obvious move was again best, though.

Probably the most interesting sideline here was 29…b6 (or even 28…b6, which transposed). I found that White would probably win with 30. Rxf7 & 31. Rd1, and it looks like Team Black knew this as well.

<29…♕e6> Opting to return the queen to the best square rather than invite trouble.

After that relatively tunnelistic series of recent moves, the game opened up here again somewhat, and this is where Team White had to find the win.

click for larger view

It was a dark and stormy night….

Team White worked frantically, as mentioned, for this move that was White’s last, best shot to dictate the course of the game. Ranking Team White’s most dramatic votes, this was number one because of a last-minute switch from Rc2 to Rf4. If nothing else, it was encouraging that the lead analysts were able to pull the switch—influencing enough silent voters—on very short notice.

The obvious moves are, indeed, Rf4 and Rc2, but frustration kept driving us back to try weirder things. Qf5 and a4 were two of them, but they didn’t work.

In this position, both sides are limited to scant few productive moves, so I tried to use that to our advantage with the idea 30. Rf2. Black was ready to reposition the knight with b6 and Nb7, followed by Nc5 or Nd6, but if White could safely give away a tempo, Black would have had to choose the wrong one. I believe this idea worked for 31…b6—temporary elation, but after more analysis I saw that 31…c5 was a problem. I managed to produce an ending where White had a small but disappointing advantage; this was rejected because in other lines we had more such endings to choose from.

30. Rc2 was looking like our best shot for a while on the strength of (at least) these two endings, both starting 30…c5 31. Qf5 b6 32. Qxe6+ fxe6 33. Rf4 e5:

1) 34. Rxg4 Nb7 35. Rc1 Nd6 36. Rxg8 Rxg8 37. g4 Kd7
2) 34. Re4 Nb7 35. Rxe5 Nd6 36. a4

White is up a pawn in both, but in the time allotted we could not find wins.

Still, it was leading the vote against the most promising Rf4 line, discovered independently by me and <galumph>. That line was longer but also looked drawn until, once again, <karpkasp> swooped in and extended it to another win! We knew that the variations had not been checked very well, but, being good sports, we again chose to strike forth into the unknown. Besides, Rf4 just feels right, doesn’t it?

<30. ♖f4 b6> Moves that cut to the chase in a big hurry.

Black had worthy contenders in Rde8 and c5, but they didn’t get the knight involved soon enough.

In the nick of time, Team White had established a new main line on which to ride to victory, but there would be one more surprise….

On move 31 there’s another batch of tempting options, but we had already shown them to fail. Qf5 quickly draws; a4 and Nf1 provided no way to maintain the advantage in the computer’s eyes.

31. Rc2 was the most interesting. After 31…Nb7, the computers like 32. Qf5 with a very good score, but here again was an ending we couldn’t solve after the queen trade. Our cleverest idea was instead 32. Rd2, where the computer favorite …Nd6 gives White new life. But the simple …Na5 invites a repetition or a return to the 32. Qf5 line. The only escape route was 33. Re4, which we examined thoroughly, but we found a difficult ending yet again after 33…Qxc4 34. Rxg4 Rxg4 35. Qf5+ Kb8 36. Nxg4 Qb3 37. Rd3 Qb2 38. Qf6 Qxe2+ 39. Nf2.

That left only one plan that seemed to keep the game alive.

<31. ♖e4 ♕d7 32. ♖d1>

White’s move 32 options Re5 and e3 were quickly tossed out.

Here it seems Team Black always had their eyes on 32…c5, but 32…Nb7 was plausible too, or maybe we just hoped that it was; it certainly sucked up lots of our time. Black definitely made the right call, though. 32…Nb7 33. Qxd4 Nd6 34. Rxg4 Qe6 35. Rxg8 Rxg8 36. e4 Nxc4, then some choices, such as 37. a4 Kb8 38. Rf1 Nd6 39. Rf4 f5 40. e5 Ne4 41. Rxe4 fxe4 42. Qxe4 made for a too-interesting game from Black’s perspective. White had several earlier choices besides, and therefore we came to expect:

<32…c5 33. bxc6 ♘xc6>

We looked at 33…Qxc6, just in case; White gets a sizeable advantage back in several lines. Even if it had been a good move, Team Black probably couldn’t have tolerated leaving the knight on the rim any longer.

<34. ♖f4>

34...?

Rde8, Qe6, Rg7, Rg6, or Kb8, if we can limit it to five possibilities. The top move could be anything, depending on which engine you ask and at which search ply. Or which person.

The two Teams approached this move from completely different angles. Perhaps it was a matter of Rybka vs. Houdini; perhaps it was engines vs. intuitions or some other combination of factors. But somehow, Team Black had been mostly planning for Kb8 since the dawn of time, and the other moves only got attention as the day approached. Team White had been mostly planning for Rde8 and only started looked hard at alternatives when the move was staring us in the face.

Well. That’s not such a big deal, in retrospect. We didn’t have a forced win anymore in Rde8 or any other line anyway. After 34…Rde8, there was a pretty good line for White that at least would have extended the game to torturous lengths where we might have been able to eventually find a good ending: 35. Rd2 Rg6 36. Qf5 Rge6 37. Nxg4 Rxe2+ 38. Rxe2 Rxe2+ 39. Kf1 Re7 40. Qd3 Qe6, etc. 37. e3 may also have provided something.

34…Qe6 offers White several options and is too complex to address here. (It seems <Golden Executive> is the person to ask about this.) I for one thought it was as good as anything else.

The resilience of 34…Kb8 also became known to us reasonably beforehand, so by the time Black’s 34th rolled in, we knew to expect anything.

<34…♔b8>

And yet, this was still something of a surprise. For one thing, I thought this gave White too many good options, too many ways to find a good continuation. Apparently, Team Black had had time to prepare for all of them… or had they? Okay, don’t panic.

Of course, it was well known before Kb8 showed up that 35. Qf5 leads to a simple draw. Sadly, the line I’m referring to is what actually happened in the game. The only branch worthy of search, in my opinion, was 37. Rc1, but we had already taken care of that too. I saw that Team Black was prancing about their “magical” 40…Nc2 in the kibitzing, but we hit upon this right away. Not that we needed to; most of us felt it was hopelessly equal after 39…Nxa3 and a string of apathetic et ceteras.

Well, no problem. I was planning a nifty line with 35. c5!

I’ll get back to c5, but meanwhile, other analysis was taking place, rather diligently. Rd2 and e3, the third reasonable option, surfaced in short order. We also did some work on Kg1 and—as a last-second desperation—Qc2, but I’ll summarize them with “didn’t pan out” and get on to the main variations.

Team White’s most frantic search in the entire game was going on now. For me and most others, it was also the last.

35. Rd2 seems hopeful at first, or at least complicated, after 35…Qe6 36. Qh7 f6, long lines thence. However (duh!) 35…Ne5! forces the same dismal ending that happens after 35. Qf5, and that’s that. I remember that some Rd2 votes stubbornly remained after this was discovered.

35. e3 was reasonable, and not…quite…dead…after 35…Qb7 36. exd4 Ne5+ 37. Qe4 Qxe4+ 38. Rxe4 Nxc4 39. d5 Nxa3 40. Rxg4 a5 41. Rf4.

35. c5 has a story of its own—the game’s final subplot.

Before Kb8, I was thinking 35. c5 Ne5. At first, Black seems to be okay with 36. Qe4 Ng6 37. Rf5, etc., but the hard move to spot is 37. Rxg4!!, which doesn’t register in the engines. Long line, then White ends up one healthy pawn ahead and sitting pretty. Glory and ululation!

But there was still 35…Qd5+ 36. Qe4, etc. and the line looks fairly dead. I think this is as far as Team Black ever got, which increases the pseudo-drama of the story!

But there’s more. 35. c5 Qd5+ 36. Kf2! is the next improvement. 36…Qxc5 37. Nxg4 Rg6 38. Kf1 is best, and several options emerge for Black. 38…Qd6 is most prominent (most others don’t work), and I discovered after 39. h5 Rg5 40. Nf6 Qe6 41. g4 Re5, White is back with a vengeance. This is the line I was trying to sell the team on. Still playing for a win, I said.

Then, who should show up out of the blue but <karpkasp>, who swooped in once again to say that I had missed 38…f5, which is very drawish, but that he would vote for 35. c5 and hope we could find a way around the problem—he wouldn’t be back before the vote closed.

It didn’t take long for me to find that there was no way around it. We had a draw in all variations, all moves, everywhere. Our best chance—slim though it may be—was, and still is, officially 35. e3, but it was too late—most votes were Qf5 or c5, and the former was still leading, against all reason.

We all knew that Team Black would see the proper way to handle c5 in short order, if they hadn’t seen it already. I appealed to the team to vote for c5 for poetic reasons and offer a draw after 35…Qd5+ (and because we could probably retain the dignity of a pawn-up but unwinnable rook ending), since it was too late for e3. At the last minute, <isemeria> switched to c5, probably feeling sorry for me, to create a tie between the top two moves.

<35. ♕f5>

The winner of the meaningless coin-flip, and the end of the story.

I’m proud of what we were able to accomplish with our limited resources on Team White. We were ready to pounce on any number of Black mistakes, but with lots of hard work and a little bit of luck they always managed to find the right path.

No doubt the move of the game was 5…g6. I hope the professionals take note! OTB this would cause White serious headaches!

In the Biggest What-If department, I wonder if we could have defeated 5…Nge7. Our analysis was pointing towards yes.