Chess960 (or Fischer Random Chess) was invented and advocated by Bobby Fischer where the starting position of the pieces on the first and eighth ranks is randomized. There are 960 possible starting positions, hence the name Chess960.
The idea is that the random setup makes memorization of opening lines impracticable which forces the players to rely on their talent and creativity.
Sounds like sales hype for the latest opening book on some crappy gambit, but that's another story.
Prior to Fischer the random placement of pieces was known as Shuffle Chess, but Fischer's idea was to place restrictions such as retaining bishops of opposite colors and castling. Black's setup always mirrors White's. The ranks get shuffled by computer to create different starting positions for each game.
It's not very popular probably because people generally don't like change. The argument that it will spice up the game by eliminating memorization of openings deep into the middlegame sounds like a faux argument to me though because 99 percent of us can remember that much opening analysis and aren't very good players to begin with. Back in the pre-computer days when we all used Modern Chess Openings in postal chess and then got to the end of a column we were on our own and started playing chess like whatever class we were in...same in OTB games. Our opponent varies from our opening preparation at move three, or we forget it at move six and are on our own practically from the start of the game.
Hence, there's no compelling reason for most of us to abandon normal chess that I can see. Memorization may be killing chess at the highest levels which makes the games boring and so it's not nearly as much fun to play over a game by Carlsen as by Tahl.
Really good players have an excellent knowledge of the general principles, strategic themes, tactical patterns, endings etc...but let’s face it, most of us don't. So it seems to me the argument that Chess960 would eliminate all that is bogus and I don't see how it levels the playing field. No matter how you set up the pieces, I'm not going to be defeating Carlsen, Topalov, Anand, Kramnik and Nakmura. For that matter, I seriously doubt I could beat Meier, Gharamian, Nielsen, Mamedov or Motylev either. Who are they, you ask? They're at the bottom of FIDE's top 100 list and are all rated over 260.
Anyway, given the fantastic talent and memories Grandmasters have for chess and their ability to remember dozens of opening variations many moves deep, sooner or later there would be some kind of opening theory develop, especially with the help of engines.
Here is Starting Position 354 from my Fritz 12 program:
In this position playing a 5-minute game against Stockfish 6 after 10 moves we reached this position:
Stockfish evaluates this position at 0.11, or dead even. After 16 moves we were at this position:
My play has drifted downward a little bit and it's evaluated at a slight advantage for black (-0.30) My next move, 17.f4, turned out to be rather poor and the evaluation jumped to -0.70. I should have played either 17.h4 or 17.g3. Anyway, after 17.f4 I only managed to hold out another 17 moves before resigning. If Chess960 didn't level the playing field for me against Stockfish 6 and I doubt it would against anybody.
If anyone is interested in playing Chess960, Lechenicher SchachServer offers Chess960 tournaments. They are double round events, 5 players, 8 games and the time limit is 90 days for the whole game.
The server will select the starting position randomly and if castling is possible, the server will display a castling button so you really don't have to know any special rules. Registration is free and you use your real name and after registration you will receive an email with your User ID.
Of course LSS allows engine use, so why bother?
I'm sure there are other sites with Chess960 that don't allow engines, but, again, what's the point? I seriously doubt you will be a Carlsen, Topalov, Anand, Kramnik or a Nakmura in this form of chess either. Obviously, when it comes to Bobby's invention, I ain't a fan.