Connections review: the puzzle that turned “I guessed too early” into a national pastime

NYT Connections, in case you have somehow avoided it for the last 18 months, works like this: every day, the puzzle gives you 16 words arranged in a 4×4 grid. You have to sort them into four hidden categories of four words each. You get four guesses. If you guess wrong four times, you lose. There are no clues. The only information is the words themselves and the unspoken promise that the categories are clever.

The first time I played it, I solved it in 90 seconds and decided it was a fine little puzzle. The second time, I lost in 45 seconds and decided it was actively malicious. By day 30, I was hooked. By day 100, I had started a group chat with three friends to compare daily results. As of this morning, day 547 of my daily play, I have a strong opinion: this is the most carefully designed daily puzzle the New York Times has ever published, and the closest thing the form has to a Wordle-equivalent breakthrough.

What’s actually happening on screen

The pleasure of Connections is not the categories. The pleasure is the trap.

Every Connections puzzle is built around at least one “overlap word” — a word that obviously belongs to one category but actually belongs to a different one. Yesterday’s puzzle, for instance, included PALM, BEACH, BANK, and SAVINGS. The obvious category is “things to do with money.” The actual category is “things that are also Florida cities.” Two of those words are Florida cities. The trap is that, having committed to the money category, you put SAVINGS in your guess instead of, say, NAPLES.

This kind of misdirection is not new — cryptic crosswords have done it for 90 years — but it has never been delivered in a 90-second daily-puzzle format before. Connections does it every single day. Some days the trap is gentle. Some days it is gleefully cruel. The puzzle that broke me hardest was the day in March 2025 when the categories included “words that follow CROSS” and “words that follow WORD” and three of the four candidates worked equally well in both.

The four-tier difficulty system is the unsung hero

Connections classifies its four daily categories by difficulty: yellow (easiest), green, blue, and purple (hardest). The colors are revealed only after you successfully solve a category. Until then, you don’t know which is which.

This is, on reflection, an extremely clever piece of design. The yellow category is almost always the obvious-looking one — the trap. The purple category is almost always the one that, in retrospect, you should have spotted in the first 30 seconds and didn’t. The green and blue are the meat. Solving in the wrong order — picking yellow first, blue second, green third, purple fourth — is the marker of a beginner. Solving purple first is the marker of a player who has stopped trying to be clever.

I now solve purple first 60% of the time. My win rate is, accordingly, 89%. Eighteen months ago it was 62%.

What it does well

1. The category writing is reliably brilliant. Connections is constructed by a small editorial team led by Wyna Liu, and the categories show evidence of careful editorial selection. They are not random. They are not algorithmically generated. They are written.

This shows up in the puzzles. A great Connections category will have one obvious red herring (the trap), one delightful piece of wordplay (the smile-and-nod moment), and a tight thematic frame (every word in the category genuinely belongs). Mediocre Connections puzzles — you get one or two a month — miss on one of these. Great ones nail all three.

2. The pace is correct. Connections takes 90 seconds on a good day and four minutes on a bad one. That is the right amount of time. Wordle is shorter, the full crossword is longer, the Mini is in between, and Connections sits at exactly the level of cognitive engagement that fits between waking up and pouring coffee.

3. The streak system is sane. Unlike Wordle, Connections does not break your streak if you lose — it tracks your current win streak, your longest, and your overall win rate, but it does not display them as the punishing single-number streak that Wordle does. The result is that I feel free to take risks on hard categories, knowing that a wrong guess is not the end of a 547-day investment.

What it doesn’t do well

1. The mobile UI is slightly worse than the desktop. The 4×4 grid is the right shape for a desktop browser. On a phone, the categories at the bottom of the screen sometimes get covered by the keyboard or by the iOS home bar. NYT has tweaked this twice in the last year and it’s better than it was, but still not perfect.

2. Some categories are too clever for their own good. Connections has, on rare occasions, run categories that depend on a piece of cultural knowledge so specific that solving them feels less like a puzzle and more like a trivia question. “Fantasy authors named C.S.” (one such category last fall) is technically valid. It is also a category that excludes the 90% of solvers who do not happen to read fantasy novels written between 1940 and 1965.

3. There is no archive on the free tier. If you miss a day, you miss a day. Yesterday’s puzzle is gone. (Subscribers to NYT Games can play the archive going back six months. Free users can’t. This is a small but persistent annoyance.) Some solvers also use third-party trackers like nytimescrossword.net to see what they missed the day after.

Who Connections is for

Almost everyone, but especially: people who like Wordle but have started to find it repetitive, people who would like to play a daily puzzle but find the full crossword intimidating, and anyone who has ever been told they are “too confident” in a way that has never quite been confirmed by their performance under pressure.

If that last description fits you, congratulations: Connections is the puzzle that will, on a regular basis, gently and decisively tell you you were wrong. This is a feature, not a bug. Our strategy guide goes into the four-pile method I now use every day.

The verdict

Connections is, after 547 days, the puzzle I open second every morning, immediately after the Mini and before the full Crossword. It is the cleverest daily-puzzle launch the NYT has done since Wordle, the only major daily where the puzzle’s primary skill is anti-overconfidence, and the rare app I would unhesitatingly recommend to a non-puzzle person.

I have lost more games of Connections than I want to admit. I have liked every single loss. The trap, when it works, is the whole point.