Difference between revisions of "Losing play"
(→Strategies for losing play: added section) |
(→References: copy-edit) |
||
| (One intermediate revision by the same user not shown) | |||
| Line 26: | Line 26: | ||
=== Aiming for a free move === | === Aiming for a free move === | ||
| − | If your opponent is [[intrusion|intruding]] into one of your [[virtual connection]]s, you can defend it in a way that makes it more complicated, in the hope that your opponent ends up playing a move which does not even ''threaten'' to break that connection. Your opponent doing so basically gives you a free move elsewhere. | + | If your opponent is [[intrusion|intruding]] into one of your [[virtual connection]]s, you can defend it in a way that makes it more complicated, in the hope that your opponent ends up playing a move which [[Intrusion#Invalid_intrusion|does not even ''threaten'']] to break that connection. Your opponent doing so basically gives you a free move elsewhere. |
This is mainly when the opponent does not realize the [[virtual connection]] exists, so the opponent is likely to continue playing nearby, due to (if the opponent is Blue) thinking "stop Red from connecting here" rather than "gain from Red defending Red's connection here". | This is mainly when the opponent does not realize the [[virtual connection]] exists, so the opponent is likely to continue playing nearby, due to (if the opponent is Blue) thinking "stop Red from connecting here" rather than "gain from Red defending Red's connection here". | ||
| Line 33: | Line 33: | ||
Most [[computer Hex]] algorithms are not good at losing play. Both neural network algorithms and the more traditional alpha-beta-search algorithms are optimized to find the most promising move from a number of possibilities. However, if all available options are clearly losing, these algorithms do not have a notion of which moves are "more" losing than others. When presented with such a position, most algorithms just make random moves. | Most [[computer Hex]] algorithms are not good at losing play. Both neural network algorithms and the more traditional alpha-beta-search algorithms are optimized to find the most promising move from a number of possibilities. However, if all available options are clearly losing, these algorithms do not have a notion of which moves are "more" losing than others. When presented with such a position, most algorithms just make random moves. | ||
| + | |||
| + | == References == | ||
| + | |||
| + | * Michael H. Albert, Richard J. Nowakowski, and David Wolfe. "Lessons in Play: An Introduction to Combinatorial Game Theory". A K Peters, 2007. This book discusses strategies for losing play (in any game, not just Hex), and contains the phrase "give them enough rope". | ||
[[category: Advanced Strategy]] | [[category: Advanced Strategy]] | ||
[[category: Computer Hex]] | [[category: Computer Hex]] | ||
Latest revision as of 02:58, 2 October 2023
Losing play is when a player knows that they are theoretically losing, but continues the game in the hopes that the opponent will make a mistake. This typically happens when the winning player is less skilled than the losing player and may not be able to see the winning strategy.
One example of losing play is a handicap game, where the stronger player is losing at the start of the game, but hopes that the opponent will make enough mistakes to tilt the game in the stronger player's favor.
Contents
Strategies for losing play
Losing play feels very different from regular play, because the player already knows that there is no winning move and therefore has no viable options to choose from. The player might be tempted to resign. The objective of losing play is to induce the opponent to make a mistake.
Giving them enough rope
A common strategy for losing play is to make the situation as complicated as possible, so that the opponent will not be able to analyze it in the amount of time they have. This strategy has the gruesome name "giving them enough rope". The name is derived from the proverb "Give someone enough rope, and they will hang themselves". In Hex, this can take several forms:
- Do not simplify. Don't play any moves that would unnecessarily simplify the board position and clarify the situation for your opponent. For example, if your opponent has a complicated virtual connection that uses an intricate web of double threats, don't intrude in the double threats, as it would simplify your opponent's connection. Another example is fast forwarding a ladder, rather than playing it out.
- Give your opponent many potential responses. If you play a forcing move, your opponent typically only has one possible response, and you are effectively giving them no choice but to play a winning move. Instead, play a move where the opponent must choose from many potential responses. Hopefully they will pick the wrong one.
- Play an unexpected move. It often helps to play a move that your opponent did not anticipate. The opponent is forced to analyze your move from scratch, and may not have enough time to complete the analysis. Effectively you are giving them a puzzle that they may fail to solve.
Playing a fishing move
A fishing move is a move that the opponent could foil, but they may not know how to do so. If the game is still undecided or you are winning, playing a fishing move is a bad idea, because when the opponent foils, it worsens your position. But in losing play you literally have nothing else to lose, so you may as well try a fishing move. If you do so, try to play it as early as possible, hopefully before your opponent can see through what you are doing.
Feinting
A feint is a fake threat, designed to get your opponent to overreact. It can be a move that looks like it might threaten your opponent's connection, without actually doing so. The hope is to goad the opponent into defending the connection anyway, wasting a move. Feinting is a form of bluffing.
A feint is different from a fishing move, because a fishing move actually does require a response, but the opponent might respond in a way that is disadvantageous for them. On the other hand, a feint is a pure bluff that the opponent could just ignore.
Aiming for a free move
If your opponent is intruding into one of your virtual connections, you can defend it in a way that makes it more complicated, in the hope that your opponent ends up playing a move which does not even threaten to break that connection. Your opponent doing so basically gives you a free move elsewhere.
This is mainly when the opponent does not realize the virtual connection exists, so the opponent is likely to continue playing nearby, due to (if the opponent is Blue) thinking "stop Red from connecting here" rather than "gain from Red defending Red's connection here".
Losing play in computer Hex
Most computer Hex algorithms are not good at losing play. Both neural network algorithms and the more traditional alpha-beta-search algorithms are optimized to find the most promising move from a number of possibilities. However, if all available options are clearly losing, these algorithms do not have a notion of which moves are "more" losing than others. When presented with such a position, most algorithms just make random moves.
References
- Michael H. Albert, Richard J. Nowakowski, and David Wolfe. "Lessons in Play: An Introduction to Combinatorial Game Theory". A K Peters, 2007. This book discusses strategies for losing play (in any game, not just Hex), and contains the phrase "give them enough rope".
