Cheating Hangman
September 26, 2019 8:32 AM Subscribe
Cheating Hangman
I made a game of Hangman that "cheats" by changing the word it's thinking of in order to make it as hard as possible to guess. But while it cheats, it never lies: it remains consistent with what it's told you so far about the word length, matching letters, non-matching letters etc.
I was originally inspired by Nick Berry's excellent analysis of hangman strategy, many years ago. I reversed his procedure to instead come up with the hardest words to guess when playing hangman. But then just this week, a conversation with a work colleague got me thinking: what would a game of hangman be like if the hangman was allowed to cheat?
The game works by jumping in between you making a guess and it telling you whether or not the guess is correct. It considers all of the possible words that it could have been thinking about and "switches" to the one that maximises the number of guesses it expects you to have to make. I've written a blog post with more detail about its strategy and what it feels like to play it and I've open-sourced the whole thing.
It needs a pretty modern browser to work; sorry if your mobile phone isn't up to the challenge. Good luck!
I made a game of Hangman that "cheats" by changing the word it's thinking of in order to make it as hard as possible to guess. But while it cheats, it never lies: it remains consistent with what it's told you so far about the word length, matching letters, non-matching letters etc.
I was originally inspired by Nick Berry's excellent analysis of hangman strategy, many years ago. I reversed his procedure to instead come up with the hardest words to guess when playing hangman. But then just this week, a conversation with a work colleague got me thinking: what would a game of hangman be like if the hangman was allowed to cheat?
The game works by jumping in between you making a guess and it telling you whether or not the guess is correct. It considers all of the possible words that it could have been thinking about and "switches" to the one that maximises the number of guesses it expects you to have to make. I've written a blog post with more detail about its strategy and what it feels like to play it and I've open-sourced the whole thing.
It needs a pretty modern browser to work; sorry if your mobile phone isn't up to the challenge. Good luck!
Role: Programmer
This project was posted to MetaFilter by jacquilynne on September 30, 2019: Tsktsk!
Indeed, sugar and confetti, you're absolutely right (a fact that's very clear when playing Hangman against any kind of computer player): cognitive factors are absolutely something that really affects how "hard" a game of Hangman is for humans. Those aren't accurately reflected in the difficulty curve of Cheating Hangman or any other computerised hangman game I've seen.
Another bias I find, when I'm playing, is that it's easier to think of words-that-start-with something than words-that-end-with something. For example, looking at 6-letter words in Cheating Hangman's dictionary, there are 27 of them that end with "-YING". There are only 23 that begin with "BAS-". But I'll bet that you can think of more of the latter than the former given 10 seconds on each, right? For me, at least, it feels as though words might be stored somewhat like a linked list in my mind, with each syllable hanging-off the one that preceded it. Similarly, playing I Spy with items whose second and third letters are specified instead of their first introduces a significantly heavier cognitive load!
I think you're right about consonants. It's probably also true that letters that touch one another, especially if their presumed-sound matches the sound they make in the completed word, help too. I'm really interested in the linguistics of how this affects games in other languages (e.g. in highly phonetic ones like Welsh, Russian, or Lojban), but unfortunately I've neither the time nor talent to research it further.
posted by avapoet at 8:16 AM on September 27, 2019
Another bias I find, when I'm playing, is that it's easier to think of words-that-start-with something than words-that-end-with something. For example, looking at 6-letter words in Cheating Hangman's dictionary, there are 27 of them that end with "-YING". There are only 23 that begin with "BAS-". But I'll bet that you can think of more of the latter than the former given 10 seconds on each, right? For me, at least, it feels as though words might be stored somewhat like a linked list in my mind, with each syllable hanging-off the one that preceded it. Similarly, playing I Spy with items whose second and third letters are specified instead of their first introduces a significantly heavier cognitive load!
I think you're right about consonants. It's probably also true that letters that touch one another, especially if their presumed-sound matches the sound they make in the completed word, help too. I'm really interested in the linguistics of how this affects games in other languages (e.g. in highly phonetic ones like Welsh, Russian, or Lojban), but unfortunately I've neither the time nor talent to research it further.
posted by avapoet at 8:16 AM on September 27, 2019
This is brilliant. My eventual arrival at "tsktsked" was honestly wonderful.
posted by Nossidge at 9:09 AM on September 27, 2019 [3 favorites]
posted by Nossidge at 9:09 AM on September 27, 2019 [3 favorites]
I was just checking my Recent Activity, saw Nossidge's comment and went for another game: the result
posted by exogenous at 5:50 AM on September 28, 2019
posted by exogenous at 5:50 AM on September 28, 2019
That is evil and wrong and ridiculous and great.
posted by jacquilynne at 10:23 AM on September 30, 2019 [1 favorite]
posted by jacquilynne at 10:23 AM on September 30, 2019 [1 favorite]
« Older In The Terminals Of Minraud... | making/upcyling jewelry outta ... Newer »
posted by exogenous at 11:52 AM on September 26, 2019 [2 favorites]