Why was my move a mistake? Please explain the engine's logic here

I played ...Rad8 and the computer says it was a mistake! It actually said I went from a winning position to a drawn position after making that move, and suggested this instead:

To me, it looks like a fairly quiet position where you can't really go wrong with any move... and yet, the computer seems to think that some moves are +2 pawns better than others! Why is that?

Right off the top, I would say its a "mistake" (though i don't like the choice of words) because your space advantage lies on the queenside, and you moved the queenside rook, instead of the kingside rook.

So first off, the difference between the two moves on my PC with SF10 is only 0.8 not 2.0

When there's a big difference in eval without any tactics, it means there are big strategic goals that are move order sensitive (perhaps something like a pawn break) but also it could be that the opponent is temporarily unorganized / passive and if given a few moves they'll be able to get their pieces out. That seems the most likely case here, as white's minors are all bunched up together.

---

I happened to get the moves right, here was my thinking.

First off I look at the pawn structure, and that tells me to try and target some weaknesses on white's kingside, or make use of my space on the queenside.

Looking at the pieces, it's pretty clear the kingside is a no go. White has all 4 of his minor pieces in that quadrant.

Next I considered d5, just a small calculation. 1...d5 e5. Ok this hangs c5, but in terms of piece activity is this a goal for me? Well I closed off my g7 bishop's diagonal and now both of white's bishop's diagonals look dangerous. Ok so I don't think d5 is a goal. (This combined with d6 not being a weakness means Rad8 burns two tempi).

Just for the sake of being complete lets briefly consider ...c4 as a goal to undermine e4. But white wont take. So it will be something like ...c4 [white move] ...cxd and I've opened the d file against my backward pawn. Hmm, maybe that's just equal, not good or bad in terms of being a general goal.

I also habitually look at my minor pieces. What are their jobs and which want to move? My bishops are fine, but the knights probably want to move. Nf6 is blocking my bishop and Nc6 is hobbled by c3 and also tender due to Bg2.

So I don't like the kingside, I don't really like the center, but I do like the idea of the queenside because that's where white lacks pieces and I have space. Lets see what we're working with.

First thought is I want to play a4 and a3 because that increases the g7 bishop's value by undermining c3. It also opens up a square for the c6 knight which I said I wanted to move. So I do a little calculation. 1...a4 2.a3 (to stop black from playing a3) and now I notice 2...Na4 which is both a fork on b3 and a common strategic pattern (diagram below). So I know that 1...a4 2.a3 cxb must be answered with 3.bxc (even if black didn't have a discovery after 3.Qxc3) 3...Na5 and I've gained the b3 square, and c3 and a3 are targets, maybe best pressured by a rook. So maybe even 1...a4 2.a3 cxb 3.bxc Rb1 (instead of Na5)

So your engine's choice of 18.cxb is a dubious move, if you take away nothing else from my post, take this common and useful strategic pattern (below)

In this structure b2 is often an attackable weakness that can be a liability for a long, long time. Well into the endgame (especially if white has already played a3). In GM games you'll sometimes see a player use a lot of tempo just moving their rook pawn down the board just to go for a structure like this. And if a piece (like a knight or rook) can go to a hole on b3, then it also attacks the center/kingside.

By the way, I said Rad8 lost two tempi. Classically a tempi is worth 1/3rd of a pawn. So two would be 0.66. My engine evaluates the difference between Rad8 and the best move as ~0.8 so just burning tempi accounts for pretty much all of it.

(Going from a good square, to a just-as-good square only loses 1 tempo, but from good to bad loses 2 tempi because it comes with the added obligation of moving it back to a good square).

I will admit stiggling that I didn't fully understand your post. Some of these concepts are a bit advanced for me, but I am intrigued:

So in the structure you mentioned above, are we trying to play a3 ourselves, or are we trying to induce our opponent to play a3 so that we can get a nice knight or rook outpost on b3?

In my diagram (imagining there are other things going on that I hid just to highlight the queenside pawns) probably black is not wanting to play a3. Black is probably wanting to focus on b3 and b2. If black can prevent b3 then b2 is effectively a backward pawn black can try to pile up on.

In your game, black is wanting to play a3 because it undermines c3 and the dark squares in general. Your g7 bishop likes this very much, and the position gives your queenside play a natural bonus (due to your extra space there and white's lack of pieces there).

I don't mean to make it advanced. I hope you're familiar with the idea that the position dictates which area to play (kingside, center, queenside) and in general how to do it (pieces or pawns). There are basically two elements: 1) Space advantage (meaning your pawns in that area are more advanced than the opponent's meaning more safe squares to maneuver for you) and 2) Piece advantage (more pieces on that side of the board than the opponent). That will often tell you where to play. How to play isn't as easy it has to do with experience and study.

Anyway, so in the diagram in the OP the position is telling black to play on the queenside so Rad8 doesn't make sense. If you were worried about the g2 bishop I'd be fine with Rb7 or Rc8 or even Ra7 or Ra6, but Rd8 gets a frowny face

Anyway, maybe I should have started there, and not mentioned the rook pawn push thing.

It's still a bit counterintuitive to me because I was always told to control the centre of the board, or go for tactics against the king. I've never really thought about ignoring both the centre and the king, and just going for a queenside attack.

I think I need to study this general strategy a lot more. I feel like my #1 weakness right now is overall strategy, particularly in quiet positions.

One nice thing about a queenside attack, lets say you're white and you infiltrate with Rc7 and you start menacing black's pawns on his 7th rank. Lets say you have a knight on d5 helping you out (or a knight on a central square).

Well, that can easily pivot into a mating attack. You use the queenside squares as outposts to attack all the surrounding squares and pieces.

Also you can create a passed pawn which the opponent may need to sacrifice a piece to stop (or at least assign a passive blockader to the pawn, which can be as good as being down a piece).

So those are some ideas with a queenside attack / infiltration that maybe will make it feel more concrete or menacing

But for example in the French advance variation, black is often focused on the queenside / center.

In a classical king's indian defense, white is focused on trying to make his queenside play generate threats before black's natural kingside attack gets out of control.

In both cases the players are seeking play where their main pawn chain "points." Sometimes you hear this as general advice i.e. to "attack where the pawns point" which just means where you have more space than your opponent.

Rad8 loses two tempi.. It runs into Bh4 and you must move the rook again, since its now pinned.. If you ignore the pin then e5 is coming and you'll lose an exchange..