This is from Yusupov's "Build up your chess" volume 1, chapter 11 (pins).

``````[Event "Munich"]
[Date "1994"]
[Round "Diagram 11.4"]
[White "A.Yusupov"]
[Black "P.Nikolic"]
[Result "1-0"]
[FEN "8/2r2kpp/2N1p3/5p2/b3B3/4P2P/1R3PP1/6K1 w - - 1 2"]

2.Bf3
( 2.Rb6 fxe4 {Not actually the best defence according to Stockfish} 3.Nd8+ )
2...Bxc6 {In the actual game, black played ...Rxc6 here} 3.Rc2 1-0
``````
• Option 1 (Yusupov's recommendation): Bf3 and white is a piece up.
• Option 2: Rb6 and white can win a pawn.

Who wouldn't prefer being a whole piece up? Stockfish! It looks like both options win the game (approx +5 for white), but Stockfish seems to have a mild preference for option 2.

Is there a valuable lesson in here, or is it a "computer strategy" that humans shouldn't try to imitate?

It might be not even as simple as the other answer suggests (regarding only objective evaluation), as "both lines are (roughly) equal" is probably already from a certain perspective. Let's view it specifically from the computer and the human eye and that we want to win the game. I outline only the central points.

• Computer, Variant 1: "I'm an exchange up, Black will put up a hell of a resistance, I have to penetrate with my king through four pawns, might be annoying."
• Computer, Variant 2: "I mop up the two e-pawns and then comfortably waltz my majority to promotion. Whatchamean, human patzer, my Nd8 is endangered? Can't calculate 20 moves in advance?"
• Human, Variant 1: "Black hasn't the least counterplay. It takes time to crack the fortress but I can play by my buttocks."
• Human, Variant 2: "My Nd8! His figures molesting my hinterland! No thanks!"

Since Yusupow probably didn't write his book for use by a computer, I tend to follow his advice.

• +1 I was thinking of mentioning that line 1 is much easier for a human to play, and you convinced me to edit it into my answer. Jun 12 at 10:17

It's not so simple. 1. Bf3 leads to a position where White is up an exchange. 1. Rb6 leads to a position where White is up two pawns. Both are worth roughly the same in terms of material. Look at the lines carefully:

• After 1. Bf3, Black does not have to play Bxc6. 1...Rxc6 2. Bxc6 Bxc6 -- White is up an exchange.
• After 1. Rb6 fxe4 2. Nd8+ K-wherever 3. Rxe6 followed by 4. Rxe4 leaves White up two pawns.

You might think in the second line Black can play 2...Ke7, but then after 3. Nxe6 Black still has too many weak pawns (e4, g7) to defend.

Of course as a human the exchange-up position after 1. Bf3 is much easier to play than the two-pawn-up position after 1. Rb6. Black has completely no counterplay, White is up a safe amount of material in a simple position, and can almost play on autopilot. But still: objectively, both lines are equally good.

Stockfish just tells you that both moves win. The exact evaluation is irrelevant; if you increase the depth, I'm sure the evaluation will go up.

There is really no need to overthink this. For example, if you're up two pieces, it might be a good idea to give up one but make sure the opponent has no counterplay, as opposed to keeping both and allow a few checks. Stockfish might not do it, but it will show that both moves win, just like here.