Fascinating Rook Endgame from a Daily Game

In a recent Daily game, I got into this 4 vs 3 rook-pair endgame (White to move, and I was white; technically we reached this endgame a ply ago, since there was a queen trade on d5 and then Black played a6-a5).  I think it’s a fascinating position, since even though it’s theoretically a win for White (and I did end up winning), Black's outside passer is not to be underestimated, and even though White has a connected passed pawn on e3 it’s not so simple to promote, especially if Black starts trading rooks.

Because I’m a good boy, I already went through this game with a fine-toothed comb after it ended, but I think this endgame in particular is worth considering.  If you take my word and the engine's that White is winning, how do they actually win it?  At some point they’ll need to make a luft to avoid back rank weaknesses; which pawn should they push and how far?  When is it okay to trade the rooks, and when does that blunder a draw?  For Black, how do you hold?  Should you try to hold onto the pawn, or use it as a distraction while you attack White’s 4-on-2?  When should Black trade rooks?  In what lines do they have winning chances, if any?

The game continued Ra1 a4 Ra3 Reb8 g3 Rb3 Rxb3 axb3.  I eventually had to sacrifice my remaining rook for the pawn, but I’d managed to get my king and passer so far up the board that my opponent had to sacrifice back (although in the meantime I had blundered into a theoretical draw).  The computer liked Rxb3, even calling it a great move, but I wonder if it would have been more practical to keep the tension with a move like Rd3 (the other candidate move I’d been considering), since I didn’t have a perfect grasp on the resulting position and briefly wandered into dead draw territory anyway.

Just some musings about the position. I was wondering if White could double rooks on the 7th and win yet another pawn while still keeping an eye on Black's passer.

My variation may be wrong, since Black can play 33...a2! instead of Rxa7

That was something I was considering, but I wasn’t sure if it was okay to let the pawn advance any further.  Stuff like 33...a2 is exactly the reason why.

You two are discussing a fantasy. 33. .... a2 was not possible in the actual game unless the analysis next to the 2nd diagram was a lie.

They're talking about the subline.  Rook Handler is right that that's Black's way to save it in that line.

Interesting endgame.  I'd have thought that two rooks on the 7th would win it, but apparently not, at least not in that manner.

Hmm, not to particularly interesting tbh

Double the rooks on the 7th, go for mate or grab pawns, then go stop the a pawn.

But how, exactly?  Rook Handler tried that, and it doesn't quite seem to work.  I'd love to see an improvement.

This is the trap everyone’s falling into.  You need to start with Ra1 Ra3 to blockade the outside passer and tie down the opponent’s rook.  Then all three pieces on that file are frozen, and it’s basically a one-rook-per-side endgame with a four-on-three.  Black going after the blockading rook with an eventual …Rb3 isn’t scary as long as you can get your other rook behind the pawn after Rxb3 axb3.  Just make sure that your own king stays active, without getting used as a target — especially if one pair of rooks gets traded and you can’t blockade the pawn anymore.

I was thinking if there is any variation for us to win the a-pawn early and still keep a winning position in the game, so I have tried a few ideas. It seems like it will never be straightforward after all.

hello

This is the way.

The problem with that strategy is that if you don’t blockade the pawn ASAP, it can advance pretty far shockingly quickly, and every step that pawn takes forward is one step further away from the position being a win.  The reason outside passed pawns work as a decoy in the first place is that you really do need to pay attention to them.  Black can give away all their other pawns if it lets them monopolize that file.

Two rooks on the seventh rank can’t stop a pawn on the other side of the board if Black stays passive but solid, and multiple people have found that attempting to win the game like this is risky at best.  You can win some pawns, sure, but 8th-rank defense stops checkmate, and your own pawns are never going to beat Black’s without a blockade.  Maybe it’s a cleaner conversion with precise play, but you can’t just handwave the pawn by talking about “keeping an eye on” it.

The one thing that might tip the balance here is if you can win the pawn with your king, but if the black rooks are placed on a8 and b8, I don’t see that working either, since if you try to match with rooks on a7 and b7 to fight for the b file and let your king pass, Rxb7 Rxb7 lets Black push the pawn again, and if it’s already gotten to the third rank and your king isn’t right there, that position is a dead loss.

Dismissing the advice from higher ranked players. This is the way.

No, his advice seemed valid.