I wish NYT's Connections Felt... More Joyful
New York Time's new game is good, but plays it safe in a way that feels disappointing
"Wordle" represented a pretty big shift in the way that we think about phone and puzzle games. Previously, games like Sudoku, Ken Ken, and Crossword Puzzles were our dominant forms of puzzling experiences. They required a lot of effort to play. Even games like the Mini Crossword were more involved. And none of them had a cute little representation of your experience playing the game. But Wordle proved that puzzle games could be short, fun, and bursting with joy.
NYT's new-ish game Connections seems pretty clearly to be playing in that space. In it players are given a 4x4 grid of words that they're asked to group into 4 related groupings, by selecting 4 words at a time and guessing that they are connected.
It's short enough, it uses colorful squares for communication, and it sits right next to Wordle in the NYT Games App. It's not bad, and the puzzle setting is often quite clever, but the game as a whole bums me out a bit. It’s not that the game is bad, or that you shouldn’t play it. It feels like the game was created to fit the mold of a New York Times game, and frankly, for a company that prides itself on innovating and pushing the boundaries of journalism, it feels out of step to create games that feel safe, small, and limited.
To be fair, I'm also the guy who asked if the wildly successful Marvel Snap was any good. So knowing my long history of takes this probably means Connections is destined for success. It's not that Connections is bad, or that you shouldn't play it. It just feels safe and disappointing, but also like it leans heavily on the icons of what made Codenames and Wordle so fun, but without actually using them to good effect.
In order to do that I want to talk about the magic of Wordle, Codenames, and why puzzle setting is just as important as a good set of rules.
Wordle Is a Teaching Game
Wordle is freaking brilliant. I think it was Frank Lantz who talked about how some games feel like you're capturing a small part of the universe in an idea. And Wordle fits that.
I think Wordle made three innovations in puzzle games.
It shrank down the puzzle solving experience that you get out of something like a Crossword and encapsulated that "AH HA" moment of finding a word, and made the entire game about it.
It's a word puzzle that teaches you how to win while you play.
It figured out how to reinforce the players unique experience playing the game in a way that was quick, fun, and shareable.
It's the second innovation that's just... it's just so smart. A lot of the discussion around modern games is exactly how much effort a game should put into futzing the numbers to make them feel smart or if it should expose them to the full challenge of the rules and let them figure it out on their own. Wordle somehow squares that circle, but teaching the player as it goes, making it increasingly likely they'll guess the word, but without making it inevitable.
Wordle taps into players' knowledge about the probabilities of combinations of letters and reinforces their understanding of the construction of the English language without making it onerous or unfun. And the output. The five-by-something grid of grey, yellow, and green squares becomes a tapestry that communicates real information about how that player experienced their journey of the game.
When Wordle tells you what letters are accurate, what your brain starts doing is figuring out what other letters are likely and unlikely to be part of the remaining word shape. And it's absolutely wild and joyous to me that Wordle manages to compress so much information about what people already know about word probabilities without being stuffy.
Connections Also Has Squares
Connections also has squares, but they feel tacked on. Kind of like the New York Times realized that having these emoji that you could use to share the game would increase the "virality" of the game, but without spending all that much time thinking about what they mean. I cannot for the life of me tell you what different color combinations are supposed to communicate. And it's kind of rough to share a 4x7 grid with your family when they all have 4x4 grids. It doesn't seem to communicate much outside of how much you struggled. It's not joyous.
But this is all kind of surface level. Let's dive into the conceptual distance between definitions of words.
Codenames, The Party Game About Being Good at Words and Concepts
Codenames is a wildly successful board game by Vlaada Chvátil published in 2015 by Czech Games Edition. It's played on a 5x5 grid of words where two teams compete to guess the correct words (their "spies") before the other team. One player from each team is the clue giver, and can see the answers. Their goal is to give one word clues that relate to the words on the table their teammates are supposed to guess.
For example, let's say you see the words "Spider" and "Millipede" on the table and you know your team is supposed to guess them, but you fail to see the word "Baby", so when you give the clue "Crawlers" to your team, you have to sit and wait and see if they can suss out which ones you intended.
It's hilarious because the combination of randomness of the words, the layout, and the relationship between players creates this wild kind of alchemy that feels you could play forever. It's fun because it taps into all sorts of social connections about shared meaning, inferences, and how well we know our friends.
Connections is Also About Word Association
Unlike with Codenames, Connections has the challenge where the players aren't given clues about the groupings of words, or the category they fit into. Players have to infer the relationship between the words and the categories all in one go. So you see these sorts of repeated group types like ___ word or categories of stuff. And this is where even Polygon noted that:
Connections hasn’t become a viral sensation like Wordle, perhaps because it’s just a dash snobbier, landing somewhere between the guessing game’s accessible “I know words” on-ramp and the elitist Crossword.
And I think part of the reason that Connections occasionally feels barren is because you don't get the sort of wild bank shots that you do in Codenames where you're counting on the specific word you picked to categorize your clues makes it clear it's NOT something.
In order for a game of Connections to be really successful, a player has to jump through three specific sets of hoops to properly group words:
The Concept of the Group They're Sorting on
The Words That Belong in that Group
How Well Those Words fit in that group
And for the game to be fun, it needs the clues to conceptually overlap, but not so much that it’s impossible to differentiate why one word would belong in one group or the other. And, if the clue giver relies too much on the process of elimination, it could be an unfun experience for the player, because it's just a wing and a prayer that they see the thing they want correctly, first.
And so the challenge here, and the thing I'm very empathetic towards for the puzzle of setters is that what you really want is to create this sort of experience where there are multiple words that could potentially fit in multiple groups, but that upon further reflection it's quite obvious.
And that's why I don't think that's a problem of the clue setters so much as it is the structure of the game. So let's talk about setting clues.
What Good Sudoku Taught Me About Programmatic Puzzle Making
It's easy to fall into this belief that the system of a game is so good, that the puzzles that get created are an afterthought. Like with Sudoku, how could it be possible to mess up. But on both the Good Sudoku episode of the Eggplant podcast, and also in the book Thinking Inside the Box it becomes clear that puzzle setting is a core part of making a good puzzle game. It's easy to make a Crossword or a Sudoku that are technically solvable but unfunny to play.
What’s challenging about puzzle setting in Codenames is that you want the conceptual distance between the words to be close enough that it’s overlapping, but not that it’s so overlapped that there’s no way to differentiate between the various concepts.
I think this also illuminates why Codenames is so fun as a board game. The people you’re most likely to be sitting next to are also the people who are best able to navigate that terrain of words that have super similar meanings, but are far enough apart that you can differentiate. And Connections doesn’t have this luxury.
But that’s where I think there’s a more complex version of the game that would be more compelling because it uses progression through time as opposed to process of elimination to make the connections transparent. In its current version, unlike in Wordle or even Codenames players don't get more information over time about what how the clue giver is thinking about the words in front of them, all you receive is a yes or no answer about if you got the clue right (and if you've guessed 3 correctly but 1 incorrect you'll get a "one away"). So there's no AHA moment when you see a word taken off the board that helps you get a deeper understanding of the conceptual boundaries of the category, there's an elimination and knowledge that grouping things together is slightly easier.
And because the game flow doesn't educate the player while they play, it's all on the puzzle setter to try to make this clear. And often, you can't. Monday, October 16th’s clue is such a good example of this, as it had brilliant and challenged puzzle settings.
One of the clues was, "Days of the Week", and the roughly applicable clues were
Tuesday, Wednesday, Friday, Saturday, Sunday
Another clue was "Things going Bad"
Spoil, Rot, Fester, Turn, Sour
And the third group was "Addams Family"
Lurch, Thing, Fester, Wednesday
So I actually think "Fester" clue is brilliant here because it could technically fit in both "Things Going Bad" and "Addams Family", but festering specifically relates to the cuts and wounds on a human body, and the rest of the words do not. It fits into that sort of fun Codenames “bank shot” where if you were grouping things together you could understanding how it didn’t quite fit, but might not be able to put your finger on it.
On the other hand we have "Wednesday", which fits perfectly in both, and there's no obvious reason you would choose one over the other unless you’re able to understand the other groupings. In a game of Codenames (in this poor, tortured, metaphor) a good clue giver might specifically guide the player first towards "Addams Family" so they don't actually trip themselves up, but there's no sense of progression in the game here, so if you can't see "Addams Family" it's going to be close to impossible to figure out which 4 days of the week fit together.
When you step back and look at the whole view of the various words in the game, a kind of interesting set of relationships shows up about how the different words are grouped together conceptually. And this is something Codenames does really well as a board game, too. The progression through time helps you reveal relationships that might not have been apparent previously.
Creating Simple Puzzle Games is Hard
It’s not that Connections is a bad game, but as a game design it seems uninterested in challenging its players in a way that might lead to something interesting or joyful. It uses the words and concepts of Wordle and Codenames, but doesn’t appear to dig deeper into what would make this game truly sing. Not everything has to fit within a Wordle grid.
What do you think of Connections? I’m curious if others have found it quite as perplexing as I have.