Tic tac toe is a solved game - if you start and play optimally you always win. (you take 3 corners in 3 moves). If they start every board then they're unbeatable.
Note that this was created long before the recently posted article. So, its rules variations are totally correct, because it wasn't trying to match the article's definition of the game ;)
It originally used what it calls Rule 5a, as the article proposed, but that allows the first player to always win (see Perfect), so it now uses Rule 5b by default, but it can be toggled easily.
There are also a ton of other config options if you scroll down past the initial comments; most of the variants, like the Perfect one, just have different defaults for linking convenience.
There's one unfair thing in this though. If the board has already been won, sending them to the same board allows them to place a mark anywhere on the megaboard even if there are vacant spots available on the said board.
If they still have to go to the same board, there's a simple gambit described in a previous article that gives first player a big advantage. Open with the center square in the center board. Second player must play center board. You play the center square in the next board you're sent to. Continue, forcing the second player to play in the center board until it's completely filled and you have a marker in every surrounding board.
Using this gambit I managed to win against the variant posted by gjsriv.
Oh yeah, this is better. Fyi note that the AI is totally evil. Its first priority is to send you to a square that's already been won (does not matter who won) effectively rendering your move useless while you give it map mobility. It prioritizes that over making a move at winning a table even! Also, if you think that the AI will play to win the specific board and you can use that knowledge to your advantage... you're wrong, you're so wrong. The only way to make the AI do what you want is to pigeonhole it into a board where there's only one square left.
It looks like it's using the Monte-Carlo bot, "which plays out as many random games as it can within 5 seconds and plays which ever next position had the greatest percentage of wins". So, yup, not hardcoded :)
It doesnt follow the correct rules? or have i misunderstood them? when you send the player to a square already complete (with spaces left) they should have to go in that square, not get a free go.
5 (variation B - current default). If a player is sent to a mini-square that has already been won, or in which all the cells are already filled, then the player may next place his mark in any unfilled cell in any other mini-board.
The interplay of micro and macro reminds me of Go. I also like how offense and defense are so connected in a Yin Yang fashion.
Perhaps it should be called "TictIctaC TactActaC ToetOetoE". Another possibility: "Tic Frac Toe", giving a nod to the fractal quality. This game is great!
I won on my second try ... you should ignore winning games in favor of forcing the other player to drop you into the squares you want. None-the-less, it's a fun variation!
Maybe you only get to see the board they made their move in. You don't get to see the move unless you select that board to move in yourself.
Or maybe you don't even get that much information. You only know what moves have been made in the board you select and no further info. Do they control that side board yet? Are they trying to rig the center boards?
Taking this further, you are stuck with the board you select. Even if it has already been won. You get to update the world board, but otherwise lose your turn.
