Question about the Sicilian

I was playing as white during this game- en.lichess.org/GDIKntbT/white#63

I eventually won on time in a pretty even game, but my question is about the opening. Starting at move 5, Stockfish is suggesting for me Nxc6, which I don't understand. I have a centralized and defended knight, and it is suggesting that I just give it up in exchange for a less developed knight. I don't see the upside to it. But this is the suggested move for quite a few moves. I'm going to go out on a limb and hypothesize that Stockfish is potentially better at this game than me, so presumably it is seeing something that I don't. If anybody can help me to see what that advantage is, I'd sure appreciate it.

Oh, and I've never tried to import a game into lichess text before, so sorry in advance if the formatting is messed up. If I do it wrong, a little help please?

There are multiple playable moves and stockfish is suggesting one of them, which does not mean that Nxc6 is superior to other moves. Also, playing Nxc6 at move 5 or so does not give you more of an advantage (if it does no one will play the sicilian anymore).
Theory is often more reliable than engines in openings.
I hope this helps.

Stockfish is weak in openings when it has no access to its opening book. 5 Nxc6 is a strategic error as it trades a better developed piece Nd4 for a less developed piece Nc6 and as it allows black to get even more pawns in the centre and as it opens up the b-file, which is more dangerous than the c-file.
This being said, your 5 c3 is not the best move either. You play a move that does not develop any piece and that blocks the natural square c3 for your knight on b1. Good moves are 5 Nc3, 5 Be3, 5 c4, 5 Bc4

Fully agree with #3

If opp allows it, play the Maroczy setup in the Sicilian (5.c4), except in some cases where he can pin a Nc3 with Bb4.

I want to add some thoughts to #4.
5 c4 in this and similar positions was invented by the Hungarian grandmaster Geza Maroczy (pronounced Marutchi). It does not develop any piece, but it controls square d5 and thus hinders black to play ...d5 and thus creates a so called "bind", the "Maroczy bind", that is a strategic advantage that lasts until the endgame.
Once this became known, black players tended to avoid the Maroczy bind in the above example with 4...Nf6 instead of 4...Nc6, thus first forcing 5 Nc3 and thus avoiding 5 c4.
Later ways have been found to successfully fight the Maroczy bind. For example Gary Kasparov has demonstrated in World Championship matches against Anatoly Karpov that he could allow the Maroczy bind and then play ...d5 anyway, even at the cost of a pawn and get tactical chances as a compensation.
When black was no longer afraid of the Maroczy bind, white players tended to avoid it themselves and just play the natural developing move Nc3. For example Robert Fischer has played many games as white where Samuel Reshevsky as black allowed the Maroczy bind. In some of these games Fischer went for c4, in other for Nc3 so he must have been undecided if the bind was worth going for or not.
The exception Linnemann mentions is for example 1 e4 c5 2 Nf3 e6 3 d4 cxd4 4 Nxd4 a6 5 c4 Nf6 6 Nc3 Bb4. When black plays in this variation 6...Qc7, then white avoids the pin 7... Bb4 with 7 a3, but then black has the pseudo sacrifice 7...Nxe4.

Here is the famous Karpov-Kasparov game with the pawn-sac @tpr referred to:

de.lichess.org/bWMp94dX#16

It was the sixteenth game of their WCH Match 1985 in Moscow. Interesting is that it was not the first time that Kasparov played 8...d5, they already had it on board in game number 12 until move 10...Nb4 (in which Karpov enforced a draw as it was new for him). So Karpov was prepared. Still Kasparov scored a nice win after some inaccuracies by Karpov.

So Kasparov showed another way to equalize against 1.e4 c5 2.Nf3 e6 3.d4 cxd4 4.Nxd4 Nc6 5.Nb5!? d6 6.c4.

fourteen years earlier Fischer experimented with 6.Bf4!? e5 7.Be3 Nf6 8.Bg5!? and got a perfect score with it, and later Leko used it a few times, but this is not good for an advantage in my opinion. White has a full minus tempo compared to the Sveshnikov line 1.e4 c5 2.Nf3 Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 e5 6.Nb5 d6 7.Bg5. The idea is that white can move back his b5 Knight to c3 instead of a3, but in newer games white achieved nothing. He can enforce the draw, not more. Still this is a good line on patzer level. Black has to know the theory.

after 5.c4 Nf6 6.Nc3 Bb4 white is more or less forced to play Nxc6 and i do think that black has a few ways to equalize here. (@tpr i was referring to this line, when i said 'except in some cases where a Nc3 can be pinned')

Today virtually everybody in the 2700 club plays 5.Nc3 which is good for a slight advantage in my opinion. From a philosophical point of view this is logical, 5.Nb5 moves a developed piece a second time. There must be some clear drawing lines.

in general, after 2...e6, if black does not play ..d6 and therefore keeps the option to play Bb4 then c4 seems not good to be good for an advantage.

an exception is Kan, 4...a6?! (i consider this to be a suboptimal move), where c4 becomes playable again. Carlsen showed in his WCH match with Anand 2014 how to play if black pins the Knight:

de.lichess.org/jIIOdiKf#12

Still my personal preference against Kan is 5.Bd3 (instead of 5.c4). It keeps the option to play Nd2-c4 which throws an eye on b6 and d6.

"fourteen years earlier Fischer experimented with 6.Bf4!? e5 7.Be3 Nf6 8.Bg5!? and got a perfect score with it, and later Leko used it a few times, but this is not good for an advantage in my opinion. White has a full minus tempo compared to the Sveshnikov line 1.e4 c5 2.Nf3 Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 e5 6.Nb5 d6 7.Bg5. The idea is that white can move back his b5 Knight to c3 instead of a3, but in newer games white achieved nothing. He can enforce the draw, not more."
Indeed, this line stemming from Bronstein is the Sveshnikov with one tempo less. Nevertheless Fischer destroyed Taimanov with it, despite that Taimanov had especially prepared for it as he knew Fischer had adopted the line. I have always thought that this indicated that the Sveshnikov must be bad as white has one more tempo than Fischer. Also former USSR champion Iossif Dorfman and former second of Kasparov called the related Kalashnikov "shocking". There is another related line: 1 e4 c5 2 Nf3 e6 3 d4 cxd4 4 Nxd4 Nf6 5 Nc3 Nc6 6 Ndb5 d6 7 Bf4 e5 8 Bg5 which leads to the Sveshnikov with one more move, i.e. no tempo won or lost. Also here white plays Nbd5, i.e. playing the same piece a second time, but eyeing the weak spot d6.

Do you maybe mean starting at move 6, not 5?

At move 5 for white, its top moves are Nc3 and c4, which are the top 2 human moves.

5.Nxc6 is something like its 8th move.

It does start to like Nxc6 after the insertion of 5.c3 a6, but then Nxc6 makes a lot more sense (a6 just doesn't make sense in the resulting structure).

Further, at that point, there's no conflict between SF and accepted theory, because there is no theory of that position :)

Apologetics for SF aside, it is still certainly true that there are positions where the established human opening theory is superior to SF's judgment, but in 2017 there are a lot fewer of them than many people suppose.

That is likely from fossilized opinions from several years ago; as just one indication of the danger of relying on generalizations from years ago, I'll just point out that SF has improved by nearly 400 points in the last 4 years at bullet time controls (just to head one objection off at the pass, this is from testing on the same hardware, so this is just the gain from software changes).

The increase is somewhat less than that at slower time controls, because of the compression effect on ratings from more draws at slower time controls, but that doesn't really change the point; it's hundreds of points stronger whichever way you dice it.

This is a case in point: where there is established theory, on move 5 for white, SF's top 2 moves are also the top 2 human moves.

EDIT: I'd also hesitate to make the claim @tpr makes, that SF is "weak" in openings without access to an opening book. As stated above, there definitely are some positions in which its play is inferior to established human theory, but the number of such positions is quite small in May of 2017 (especially, I'd wager, compared to the number of positions where it improves on human theory).

In 2010, sure, top engines in the opening left much to be desired. Now, if you work really hard at it, you'll find some of those positions where their play is inferior, but it'll take some work :)

Is it weakER without a good opening book than it is with one? Sure, but so is every entity, human or computer (as a fun note, lacking opening knowledge actually seems to hurt humans more than it does computers, as indicated by results in 960 games between humans and engines, where the results are far worse for humans than in classical chess). Is it weak, full stop? No, not even close :)

@tpr #7 i may have been a bit harsh with the judgement of the Fischer variation. I have over 200 games in my database with both players above 2300, it is a good setup. And Fischer made awesome use of it. Two wins against Taimanov, one against Petrosian and one against Najdorf, thats quite a success. I like the first against Taimanov most.

de.lichess.org/ztqwtyPh#11

This game should have been a draw, as Taimanov blundered in a completely drawn endgame. Taimanov and his team then thought 8...Qa5+ was a mistake and in their second game changed to 8...Be6 and lost again. Former World Champion Petrosian and his team still thought this variation must be good for black and played 8...Be6 again and got beaten again. Bear in mind that this was before the Sveshnikov got popular. I guess if Fischer had not retired when the Svechnikov came up, it would never have gotten popular at all.

What is also interesting, that in the revanche match Fischer-Spassky of 1992 both players refrained from open Sicilians and played it closed with Nc3 and Nge2. The years of nonplaying must have led Fischer to the belief that the open Sicilian is no good. Bent Larsen has called the open Sicilian a "one trick pony".