Codenames is a party card game in which two “spymasters” give one-word clues to try to get their teammates to identify their team’s words ahead of the other team.
Teams alternate giving a combination of a single word (or proper noun, depending on house rules) and a number of tiles the clue relates to. On the board above, the red spymaster may give the clue “place” for “3,” hoping that their team will guess Park, Hospital, and Tokyo, and accepting the risk that they might hit Casino or Field (neutral cards).
I played this game over Zoom every day for the first few weeks of the COVID-19 quarantine in place policy, and pretty regularly since then. Over our daily Codenames sessions, my friends and I developed a table talk strategy to get an edge over our opponents. Every time we show that strategy to a new player, they tell us we’re cheating. In this post, I’ll explore if those strategies are actually useful, or are just blatant cheating.
Codenames Information Theory
As mentioned above, in a typical situation, a spymaster can give two pieces of information. The word (e.g. “place”) and the number of clues it relates to (e.g. “for three”).
The rules (rulebook) allow for the following:
You can guess a max of N+1 tiles, where N is the number given by the spymaster.
Your turn ends if your team selects a neutral (beige), opponent (e.g. blue if you’re the red team), or assassin (black) tile.
The spymaster can give ‘unlimited’ as a number, which allows for unlimited guesses (but does not convey how many tiles that specific clue relates to).
In the example above, if the team were to successfully guess Park, Hospital, and Tokyo on the “place for three” clue, they could guess a fourth clue. This comes in handy if they had remaining information from a previous clue due to an incorrect guess. For example, if they received the clue “name for two” and selected Goldilocks, ending their turn (because it was a neutral), they could now guess Penny with their extra guess.
When a team is within striking distance of winning the game, the options for delivering information expand. If you have three clues left, the spymaster now has the options to say either:
X for 2 (+ the extra guess)
X for 3
X for unlimited
This means the spymaster has three different ways to deliver the information “X,” and therefore can embed more information into the remaining variable—the number. While spymasters can’t talk with their team to strategize, and cannot react to team table talk (from the rules: “the spymaster is expected to keep a straight face”), they can listen to the table talk.
This means if we’re getting to the end of the game, the team has an opportunity to ask the spymaster to embed certain information in their clue. When the team is approaching their hopefully-final turn, they could give directions of the form:
If Scuba Diver is our word, give an unlimited clue.
If Day is our word, give a 3 clue.
If neither is correct, give a 2 clue.
This allows the spymaster to give the team more information.
The reason this strategy only makes sense at the end of the game is that is the only time where it doesn’t hinder the spymaster’s ability to give helpful clues. On the last turn, whether the clue is a 2, 3, or unlimited all give the team the three guesses necessary to win.
For all other turns, having your team embed assumptions into the number limits the spymaster’s options as a clue giver.
If on the first turn, the team said:
If Angel is our word, give a 1 clue.
If Ski is our word, give a 2 clue.
If Captain is our word, give a 3 clue.
etc.
That would be an inefficient way to get a single word, and would hinder the spymaster’s ability to give better clue. If the red spymaster had already thought of “place for 3,” they would now no longer be able to give that three clue, as the team would assume the three was indicating that Captain was correct, which it is not.
The Case Against Table Talk
There are four rules (rulebook) that make me think our “house rules” table talk strategy is questionable.
“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 of a good clue, you say it. You also say one number, which tells your teammates how many codenames are related to your clue.”—This reads as if the number the spymaster shares must be the number of tiles (or “codenames”) that relate to the clue, and does not leave room for changing that number for other reasons. While this could be hard to adjudicate and impossible to referee (e.g. the spymaster could make a plausible cause for relatedness or non-relatedness of tiles as necessary), it seems the number should primarily be about how many tiles are related and not a free variable for strategic use.
“The number you say after your clue can’t be used as a clue. Citrus: 8 is not a valid clue for LEMON and OCTOPUS.”—The strategy above does employ using the number to convey information, which again seems to be against the rules. In the table talk strategy, the number isn’t a clue in the sense that the number itself describes a tile in the same way 8 relates to octopus, though it is a clue in that it points to a certain tile explicitly.
“Sometimes you may have multiple unguessed words related to clues from the previous rounds. If you want your team to guess more than one of them, you may say unlimited instead of a number.”—This framing of the rule mixes what is allowed mechanically with an expectation of how that mechanic should be used. Given that the game designers included this rule as a way to catch-up for previous missed clues, it does seem they intend for this to be the only way to catch-up, and not by incorrectly using a “for 3” when the clue actually relates to one tile and two catch-up tiles.
“When your information is strictly limited to what can be conveyed with one word and one number, you are playing in the spirit of the game.”—Technically, all the information is still being conveyed in one word and one number, the table talk strategy just loads more information into that number than might be originally intended.
The Verdict
Based on the above, I think it’s clear that advising the spymaster on what clues they can say does go against the spirit of the game.
It’s unclear to me whether the table talk strategy mechanically breaks a particular rule, as the rules are explained in a way that mixes mechanics and expectations around usage, though I think the closest breakages with the table talk strategy are:
The number provided using this strategy may not relate to the number of codenames related to the clue, when the rules specify that it should.
The number is being used as a clue.
I’d rule that the table talk strategy does in fact break the rules of Codenames.
Gavel gavel 👨⚖️