Somehow I expected AI to give clues that combine 4-5-6 words at a time. It's not at all impressive to me. And I'm not a serious player at all
The Mechanical Turk is super impressive if you don't know how it works.
OpenAI's o1 Playing Codenames
91–100 of 138 posts
Re: OpenAI's o1 Playing Codenames
#92Earlier quoted context omitted.
If you really want to get good, your goal is not so much to get as many tiles as possible, but rather to get the tiles that are semantically distinct from your opponent’s. A single mistake that triggers your opponent’s tile is generally enough to lose the game. And even if they don’t do it, having them uncover the tiles from their side that are semantically similar to your own team is also useful. If you want to get…
> you learn to abuse the fact that the tile layouts follow rules and that you can rule out certain tiles without considering the words. Can you clarify? Isn't the card placement random?
It’s randomish. There are facts about the possible layouts you can memorize.
I don’t know most of these rules but there’s never 5 in a row; or even 4 in a row if you’re the team with one fewer (second team to play).
Edit: because the game layout is determined by choosing one of a few dozen possible layout cards and randomly rotating it
Re: OpenAI's o1 Playing Codenames
#93Dude I looooooooooved that game. How long did u spend building it?
Re: OpenAI's o1 Playing Codenames
#94Earlier quoted context omitted.
The S-tier tactic: When that high-number clue is cut short by a turn-ending mistake, the guessers tell their clue giver to inflate the number given during the totally unrelated next clue by however many remained from the truncated turn for which they don't need additional information to locate (and therefore it would be wasteful for a future clue to re-group those) so the stated number of that next clue must allow fo…
I don't think this sort of communication from guessers to clue giver is in the spirit of the game (at least in my play group). However, inflating later clues is a reasonable approach! It's just that I don't think you're allowed to communicate the amount of inflation. Guessers must determine whether people 5 has slack to allow additional guesses on previous clues.
The communication is only necessary/important if people haven't set this as a convention in the first place. I'll say that prior to ever looking at my clues: "I will give you higher numbers than what I said if you miss by more than 1. THe number I pick will always be high enough as to allow you to, with the +1 guess you get for free, make guesses on all the words I was hinting at.
There's also all kinds of not necessarily intended communicaton from the guessers in the fact that you can listen to which words they were considering and didn't pick. Nothing in the game attempt to say that you should not consider, say, whether they were going in the right or wrong direction in their guessing, but it sure can make a difference in how to approach later clues. If they were being very wrong, there might be a need to double up on words that you intended, and that your guessers missed.
In the same fashion, nothing in the game saying that I cannot listen to those guesses as a member of the other team, whether guesser or spymaster, and then change behaviors to make sure we don't hit words they considered as candidate words without very good reasons. Let them double dip on mistakes, or not make their difficult decisions easier. It's not as if the game demands that everyone that isn't currenly guessing should wear headphones to be sure they disregard what the other team says or does.
Re: OpenAI's o1 Playing Codenames
#95I don’t find this “super good”. It’s mostly giving 2 clues which is the most basic level of competence. The paper 4 clue is reasonable but a bit lucky (eg Jack is also a good guess). I also don’t see it actually using tactics properly, which I would consider part of being “super good”. The game isn’t just about picking a good clue each round! Now obviously it’s still pretty decent at finding the clues. Probably bette…
Can you elaborate on some of the more advanced tactics? When I play, it's mostly about getting a good 2 clue each time. Then if you can opportunistically get a 3 or 4, that's awesome. Some tactics come in for choosing the right pairs of 2's so you don't end up mismatched, or leaving clues that might be ambiguous with your opponent's... But that's mostly it. It'll be fun for multiplayer! Just like how in other online…
experienced players who know their teammates well can reliably get 3-4s. if you only go for safe 2s against these opponents you will lose every time.
Re: OpenAI's o1 Playing Codenames
#96Earlier quoted context omitted.
I don't think this sort of communication from guessers to clue giver is in the spirit of the game (at least in my play group). However, inflating later clues is a reasonable approach! It's just that I don't think you're allowed to communicate the amount of inflation. Guessers must determine whether people 5 has slack to allow additional guesses on previous clues.
The communication is only necessary/important if people haven't set this as a convention in the first place. I'll say that prior to ever looking at my clues: "I will give you higher numbers than what I said if you miss by more than 1. THe number I pick will always be high enough as to allow you to, with the +1 guess you get for free, make guesses on all the words I was hinting at. There's also all kinds of not necess…
You can of course play however you want (and I certainly think this is clever), but imo this is likely against the spirit, and perhaps letter, of the rules.
The rule on giving clues is:
"If you are the spymaster, you are trying to think of a one-word clue that relates to some of the words your team is trying to guess. When you think you have a good clue, you say it. You also say one number, which tells your teammates how many codenames are related to your clue." (emphasis mine).
The rule states that the number should be the number of words related to the clue. There is later provisions allowing you to use zero and infinity, but outside of these carve-outs (and imo the "allowed" language is telling here, since it implies any other number not equal to the number of words is not allowed) I don't think this is legal.
Re: OpenAI's o1 Playing Codenames
#97A small weakness in this test is that one of the keys to strategic Codenames play is understanding your partner. You're not just trying to connect the words, you're trying to connect them in a way that will be obvious to your partner. As a computing analogy: you're trying to serialize a few cards in a way that will be deserializable by the other player. This test pairs o1 with itself, which means the serializer is th…
Ehhh I don’t think that’s accurate. The problem is not linking 4 words. It’s linking 4 words without accidentally triggering other, semantically adjacent words. This task could probably be solved nearly just as well with old school word 2 vec embeddings
> This task could probably be solved nearly just as well with old school word 2 vec embeddings
I've tried. This approach is well beyond awful.
Re: OpenAI's o1 Playing Codenames
#98I don’t find this “super good”. It’s mostly giving 2 clues which is the most basic level of competence. The paper 4 clue is reasonable but a bit lucky (eg Jack is also a good guess). I also don’t see it actually using tactics properly, which I would consider part of being “super good”. The game isn’t just about picking a good clue each round! Now obviously it’s still pretty decent at finding the clues. Probably bette…
Ok, we've taken supergoodness out of the title now. Presumably the post is still interesting!
(Submitted title was "I got OpenAI o1 to play the boardgame Codenames and it's super good".)
Re: OpenAI's o1 Playing Codenames
#99Earlier quoted context omitted.
If you really want to get good, your goal is not so much to get as many tiles as possible, but rather to get the tiles that are semantically distinct from your opponent’s. A single mistake that triggers your opponent’s tile is generally enough to lose the game. And even if they don’t do it, having them uncover the tiles from their side that are semantically similar to your own team is also useful. If you want to get…
> you learn to abuse the fact that the tile layouts follow rules and that you can rule out certain tiles without considering the words. Can you clarify? Isn't the card placement random?
There are 40 setup cards with 4 possible rotations that specify agent placements, so it's theoretically possible to do some kind of memorization.
Personally I'd find that kind of play style very unfun, and would rather switch to fully randomized boards if I played enough that it became a problem.
Re: OpenAI's o1 Playing Codenames
#100Earlier quoted context omitted.
Of course! If you’ve played Codenames and introspected on how you play you can see this in action. You pick a few words that feel similar and then try to justify them. Post-hoc rationalization in action.
Except you also examine the rationalization as part of deciding whether to act on the impulse or not.
Yes and you may search for other words that fit the rationalization to decide whether or not it's a good one. You can go even further if your teammates are people you know fairly well by bringing in your own knowledge of these people and how they might interpret the clues. There's a lot of strategy in Codenames and knowledge of vocabulary and related words is only part of it.