There is a fad circulating online, for a little online word-game called “Wordle.” It’s okay, I guess. Just a little word-guessing game, and perhaps part of what draws people to it is that you’re only allowed to play once a day, which creates a kind of artificial scarcity.
Frankly, there’s a variation on Wordle called Absurdle that I like better. Unlike Wordle, you’re allowed to play as much as you want. But it’s much, much more frustrating. That’s because instead of the puzzle choosing a random word and you having to guess it, this version makes the puzzle “hostile” – if you guess the word the computer has chosen, but other options are available, the computer will change its mind, and move to a different word. So you’re trying to guess at a moving target. It’s exactly like playing 20 questions with a 6 year old, actually.
And then I found Semantle. This game is, perhaps, superficially a bit like Wordle or Absurdle. But instead of just guessing at spelling out a word that the computer has chosen, instead you’re trying to guess a word based on a kind of “hot/cold”, described relative to some rather complex semantic maps of word use. These are the same sorts of mega-dimensional semantic vectors (co-occurrence matrices, I think) that are used in AI-styled language translators, such as e.g. google translate. Anyway, this last is the game I find most addictive, as I try to think about how the semantic fields play out in a large corpus of sample texts.
