Chess960 vs standard chess: what is actually different
The two games share every rule but one, start with the same sixteen pawns on the same sixteen squares, and one of the 960 Chess960 arrangements is standard chess. The argument between them is narrower than most people think, and it is still worth having.
What is actually different between Chess960 and chess?
Two things, and it is worth being exact about them, because most descriptions are not.
The first is the array. Eight back-rank pieces are shuffled into one of 960 legal arrangements, subject to three constraints: the bishops sit on opposite colours, the king stands between the two rooks, and Black mirrors White file for file.
The second is the wording of the castling rule. Not the result. The result is identical to standard chess: castling short finishes with the king on g1 and the rook on f1, castling long with the king on c1 and the rook on d1. Only the starting squares differ, so the conditions have to be written in terms of the squares each piece travels over rather than as "the king moves two squares". The three prohibitions are unchanged. The king may not be in check, may not pass through an attacked square, and may not land in one, and neither piece may have moved before.
That produces two positions which look illegal and are not. Sometimes the king does not move at all: with the king already on g1 and a rook on h1, castling short is a rook stepping to f1. And castling long can move the king to the right, from b1 to c1, while the a1 rook jumps over it to d1. Both are ordinary castling. The full rule, and the cases that confuse people, are set out on the Chess960 rules page.
Everything else is untouched, and everything else is nearly all of chess. En passant,
promotion and underpromotion, stalemate, the fifty-move rule, threefold repetition,
insufficient material, touch-move, White moves first. Same board, same pieces, same
object. The one further change is in computer notation, where castling rights are written
as rook files, HAha instead of KQkq, because "kingside" means
nothing until you know where the rooks are.
The single most useful consequence is the one nobody mentions. All sixteen pawns start where they always start. Every pawn structure you have ever studied is reachable from every one of the 960 arrangements, and everything you know about it still holds.
Is standard chess one of the 960 positions?
Yes. Under the standard numbering the normal array is position 518, so ordinary chess is not the alternative to Chess960, it is a member of it, drawn with probability one in 960. You can step to it in the position generator and watch the familiar back rank appear.
RNBQKBNRNQBRKBRNThat fact does real work. Whatever you want to say about Chess960 as a game, that it rewards calculation over recall, that some arrangements produce lopsided early play, that it is perfectly playable, applies to position 518 as well, because 518 is one of them. The disagreement people call "Chess960 versus chess" is a preference between one arrangement that has been studied for a century and a half and 959 that have not.
Play Chess960 for long enough and 518 will come up. When it does you get an odd experience: standard chess with no preparation on either side, and no way to tell which of you owns a repertoire.
Is Chess960 better than chess?
The honest answer is that both sides of this are arguing about proportion, not principle.
The case for Chess960 is that above beginner level a share of every result is settled before the players sit down. Two club players reach move ten of a Ruy Lopez, one from a book and one from a video he watched twice, and the position in front of them was produced by neither. Higher up the share is larger, and a novelty prepared at home on move twenty-five can end a game that was never contested at the board.
The case against Chess960 deserves better than it usually gets, because preparation is not memorisation. Building a repertoire means picking structures that suit how you play, working through engine lines you do not understand at first until you do, learning the endgames those openings produce, and repairing the whole thing every time somebody finds a hole in it. That is study of the same kind as any other, done for years, and a player who out-prepares you has beaten you at chess rather than at recitation. Deleting it does not purify the game. It removes one of the game's skills, and it happens to be the skill that rewards patience.
Nobody actually believes a game should be all preparation, and nobody believes knowing the Ruy Lopez should count for nothing. Chess960 sets the proportion to zero. Its best defence is not that memory is illegitimate but that this is the only format where the other skills can be measured on their own.
Who is Chess960 better for?
A beginner is not punished for not knowing lines. A 1200 who loses on move nine to a trap he has never seen has learned nothing except that the trap exists. The same player losing on move nine in Chess960 has hung a piece, which is a lesson he can use tomorrow.
A strong classical player loses something he genuinely earned. If you have spent five hundred hours on the French Defence, Chess960 takes it and gives you nothing back, and being less enthusiastic about it is a reasonable position rather than a failure of imagination.
The clearest case is the player returning after fifteen years away. Tactics survive that gap almost intact. Opening knowledge rots, and worse, it rots invisibly, so you find out on move twelve. Chess960 is the format where that asymmetry costs nothing.
What do Chess960 players complain about?
The objections are real, and most of them have answers rather than dismissals.
Setting up is a nuisance. Somebody has to randomise, then both players place eight pieces, and Black has to mirror White file for file. The common club-level error is putting Black's bishops on the same colours as White's, when mirroring by file lands them on the opposite colours. It costs a minute a game and the occasional restart, which is much of why nearly all Chess960 is played online, where it is one click.
Castling stays confusing longer than it should, because the two strange cases turn up rarely enough that you forget them between sightings, and interfaces disagree about input as well. The habit that works everywhere is to castle by moving your king onto your own rook.
Study material is thin. There is no opening library that helps you, few annotated collections, and until recently no tactics sets, because every puzzle trainer was built from standard games. That gap is real and it is closing slowly. Engine-verified Chess960 tactics now exist as a free trainer and in book form, which took somebody mining the games for them.
Some arrangements feel unbalanced. They are, in the sense that White's first-move edge is larger in some positions than in others and a handful produce an immediate concrete threat. Two things answer it. The array is symmetric, so a weak square is weak for both players, and the tempo, not the structure, is what one side owns. And serious events play each drawn position twice with colours reversed, which is a fix anyone can copy in a casual game.
Why did Fischer invent Chess960?
Fischer announced Fischerandom in Buenos Aires in June 1996. His argument was that top-level chess had become a contest of preparation, that games were decided by home analysis and by seconds rather than at the board, and that creativity had gone out of the professional game. He kept the pawns, the piece moves, the board and castling deliberately. He was not proposing a new game. He wanted chess without the library.
He was also, in that period, publicly accusing leading players of prearranging games, and he never supported those accusations with anything. A correct diagnosis arrived mixed with a good deal that was not correct, and he was not the first to try it. Capablanca had worried decades earlier that theory would drain the game, though his answer was a larger board and two new pieces, and it went nowhere. Fischer's contribution was the constraint set: keep castling, keep the bishop pair on opposite colours, mirror the two sides. That is why his version is the one still being played.
Which should you play?
Both, for different reasons. Standard chess has the literature, the events and the opponents. Chess960 gives you the position nobody has seen, which is the only reliable way to find out what you can work out unaided. Deal yourself an arrangement with the generator, play it against the computer, and see how far the principles carry you. The reasoning that replaces theory is set out in the Chess960 strategy guide.