Why your second guess matters more than your first

Two players open with the same word. One of them is done in three guesses. The other runs out of rows. It feels like luck, and a little of it is, but play those two boards back and the fork between them is almost always the second guess.

Here's the setup they shared. The opener is CRANE, and it comes back like this: C grey, R yellow, A grey, N grey, E green. So the answer ends in E, it has an R in it somewhere that isn't position two, and it has no C, no A, no N. That's a decent haul from one guess. What happens next is the whole game.

The player who solved it

The first player treats guess two as a measuring tool. She picks STORE. It isn't a hunch about the answer. It's built on purpose: the R moves to position four, the E stays at position five where it's already proven, and S, T, and O are three fresh letters that between them sit in a large share of English words.

STORE comes back with the T green and the O yellow. Now the board reads: T at position two, R somewhere that isn't two or four, O somewhere that isn't three, E at five, and no C, A, N, or S. The shape has nearly closed. One more guess, chosen from the handful of words that fit, and she's done in three.

Notice what guess two did. It answered three separate questions at once: where does the R live, is there a T, is there an O. It didn't try to be the answer. It tried to make the answer obvious, and then guess three got to be the answer.

The player who lost

The second player treats guess two as his first real attempt at the word itself. He plays RINSE. It fits what he knows, sort of: R at position one, E at position five. But look at the cost. The N was already grey. That slot tested a letter the board had already cleared, so it taught him nothing. And I and S are the only new information in the whole guess.

RINSE comes back mostly grey. He's used two of six rows and he still doesn't know about T, or O, or U, or a single other common consonant. From there he starts chasing words that rhyme with his guesses, because those are the words he can see, and each chase burns a row on letters he half tested already. Six rows go fast when two of them were spent going over old ground.

Same opener. Same feedback. The difference is entirely in what guess two was for.

Information or answer: pick one on purpose

Every guess after the first sits somewhere on a line. At one end is the pure probe, a word chosen to test the most new letters, even when it can't be the answer because it breaks a proven position. At the other end is the pure attempt, a word that fits everything known and tries to end the game.

Early, the probe wins. With five rows left, shrinking the field is worth more than a one-in-thirty shot at glory. Late, the attempt wins, because a probe that hands you certainty on the last row hands it to you too late.

The mistake isn't picking either end. The mistake is drifting into the middle without deciding: a word that reuses a cleared letter and keeps a yellow letter in its proven-wrong spot, which is neither a clean test nor a real attempt. RINSE wasn't a bad word. It was an undecided one.

Two rules that cover most boards

First, after an opener that gives you nothing, no greens and no yellows at all, guess two should share no letters with guess one. Five fresh letters, full stop. Ten tested letters after two rows is a board you can work with. Eight tested letters and two repeats is a board that's still hiding.

Second, a letter you've proven belongs in the word may be left out of the next guess on purpose, if leaving it out buys you three untested letters and you're not near the end. This feels wrong the first time you do it, because the guess can't be the answer. Let it not be the answer. You're buying the map, and you still have rows to spend the map with.

There's one exception worth naming. If the field has already narrowed to two or three candidates, stop probing and start attempting. Information you can't act on is trivia.

Try both boards yourself

The daily-game solver on Pangramist takes the board as you enter it: what you guessed, and what each letter came back as. Load the CRANE board from this guide, play the STORE line, then start over and play the RINSE line, and watch how the candidate list shrinks in one and sulks in the other. It's the fastest way to feel the difference, and it beats reading about it.

The solver works only from the board state you type in. It doesn't know, fetch, or publish any game's daily answer, and its words come from ENABLE, a public domain list, not the official tournament lists.