Timeline for Build a perfect AI for the game 15
Current License: CC BY-SA 3.0
24 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 14, 2014 at 6:20 | history | edited | Joe Z. | edited tags | |
| Nov 24, 2013 at 7:02 | comment | added | boothby | Hah. I just learned about this game, and the first thing when I got home was to post it here. Alas, I've been beaten to it. | |
| Aug 7, 2013 at 17:18 | answer | added | Peter Taylor | timeline score: 1 | |
| Aug 7, 2013 at 2:12 | comment | added | Joe Z. | If the AI is given a garbage input like [6 7 9][1 2 4 8] (in which case 3 and 5 is a double trap) without having made any moves itself (but just having [6 7 9][1 2 4 8] as a starting position), that's considered an "initial losing position". Then it's allowed to make a losing move because it has no other choice. But from an empty board, the AI should never allow [6 7 9][1 2 4 8] to be reached. | |
| Aug 7, 2013 at 2:11 | history | edited | Joe Z. | CC BY-SA 3.0 | added 59 characters in body |
| Aug 6, 2013 at 14:44 | comment | added | John Dvorak | what is the intention of the word "initially" in that statement? I will update my answer, however | |
| Aug 6, 2013 at 14:36 | comment | added | Joe Z. | I think I misunderstood your question then. What I meant is, if the AI is presented initially with a position that's "winnable" from the other side, it's allowed to lose because the other side has a winning strategy and there's absolutely nothing the AI can do. Otherwise the AI must prevent a loss if possible, including taking full advantage of the other player screwing up a winning position by making a wrong move. | |
| Aug 6, 2013 at 4:52 | comment | added | John Dvorak | "Do I understand correctly we are allowed to lose if we get into a winnable position that our AI couldn't have got into?" – Jan Dvorak yesterday "Yes, that is correct. ..." – Joe Z. yesterday | |
| Aug 6, 2013 at 4:35 | history | edited | Joe Z. | CC BY-SA 3.0 | added 97 characters in body |
| Aug 6, 2013 at 4:30 | comment | added | Joe Z. | I should clarify: "the AI should never allow the game to get into said losing state from an empty board" or from any other non-losing position. | |
| Aug 6, 2013 at 4:27 | history | edited | Joe Z. | CC BY-SA 3.0 | Whoops. |
| Aug 6, 2013 at 4:25 | vote | accept | Joe Z. | ||
| Aug 6, 2013 at 12:01 | |||||
| Aug 5, 2013 at 15:35 | history | edited | Joe Z. | CC BY-SA 3.0 | added 98 characters in body |
| Aug 5, 2013 at 15:34 | vote | accept | Joe Z. | ||
| Aug 6, 2013 at 4:24 | |||||
| Aug 5, 2013 at 14:30 | answer | added | Peter Taylor | timeline score: 7 | |
| Aug 5, 2013 at 13:57 | comment | added | John Dvorak | well, mine does :-) | |
| Aug 5, 2013 at 13:50 | comment | added | Joe Z. | Well, I guess it should win if given the opportunity. | |
| Aug 5, 2013 at 10:18 | answer | added | John Dvorak | timeline score: 4 | |
| Aug 5, 2013 at 6:34 | comment | added | John Dvorak | hmm... it seems we are allowed to draw even if we can force a win. Is that true? | |
| Aug 5, 2013 at 4:12 | comment | added | Joe Z. | No, the AI can play second as well, but either way the game is solved (with perfect play, the game will always end in a tie). | |
| Aug 5, 2013 at 4:07 | comment | added | John Dvorak | do I assume correctly the AI is always the first to play? | |
| Aug 5, 2013 at 3:17 | comment | added | Joe Z. | Yes, that is correct. If the AI is presented with a game state that it cannot win or tie (e.g. a double trap), then it is allowed to lose. However, the AI should never allow the game to get into said losing state from an empty board. | |
| Aug 5, 2013 at 1:50 | comment | added | John Dvorak | Do I understand correctly we are allowed to lose if we get into a winnable position that our AI couldn't have got into? | |
| Aug 4, 2013 at 23:19 | history | asked | Joe Z. | CC BY-SA 3.0 |