Wordle tells you where. This game doesn’t — and we measured what that costs
5 min read
People reach this game by way of another one, so the honest thing to do is put them side by side properly rather than claim a difference in the abstract.
The difference is one rule. In a positional word game, each letter of your guess comes back coloured: right letter right place, right letter wrong place, or not in the word. You are told which letters, and where. Here you get two numbers and nothing else — how many letters landed in the right place, how many are in the word somewhere else. Which ones is never revealed.
That sounds like a small edit to the rules. It is not, and it is measurable.
The same guess, scored two ways
Here is a fair test. Take our answer pool and our dictionary — identical word lists, so nothing is hiding in a difference of vocabulary — and find the best possible opening guess twice. Once under positional colouring, once under a tally. Then count how many answers survive.
| Board | Best under colours | Left | Best under tallies | Left | Ratio |
|---|---|---|---|---|---|
| 4 letters (600 answers) | RALE | 39.91 | RATE | 93.44 | 2.3× |
| 5 letters (646 answers) | IRATE | 12.54 | PRASE | 74.37 | 5.9× |
| 6 letters (1117 answers) | RATINE | 10.53 | SEATER | 106 | 10.1× |
On five letters, coloured tiles take you from 646 candidates down to about 12.54. You are nearly done. The same pool, scored as a tally, leaves 74.37 — 5.9 times as many — and you have barely started.
Why the gap grows with every letter
Look at the ratio column. It is 2.3× on four letters, 5.9× on five, 10.1× on six. The longer the word, the further apart the two games get, and there is a clean reason.
Count how many different replies each rule can produce. A positional grid assigns one of three colours to each of n letters, so there are 3n possible patterns. A tally is a pair of numbers adding to at most n, so there are at most (n+1)(n+2)/2 of them.
| Word length | Colour patterns (3ⁿ) | Distinct tallies | Seen against our pool |
|---|---|---|---|
| 4 | 81 | 15 | 43 vs 12 |
| 5 | 243 | 21 | 104 vs 17 |
| 6 | 729 | 28 | 271 vs 20 |
One grows exponentially, the other quadratically. Every letter you add triples the vocabulary of a coloured grid and adds one to the vocabulary of a tally. That is the whole story: a six-letter positional guess can answer you 271 distinguishable ways, and a six-letter tally 20.
This does not make it a better game
It makes it a different one, and it is worth being straight about which parts get worse.
- You will not feel clever early. A positional game gives you a foothold on guess one. Here, guess one just narrows the field from impossible to merely large.
- It asks more of your working memory. Nothing is coloured in on the board for you. The constraints live in your head, or on the scratchpad the game gives you for exactly this reason.
- It is slower. The boards hand you eight guesses rather than six because the information per guess is roughly half.
What you get in exchange is that the puzzle stays a puzzle all the way to the end. In a positional game the last two guesses are frequently bookkeeping — the answer is already determined and you are typing it in. Here the deduction runs to the final row, because the board never does any of it for you.
So why eight guesses and not six?
Because six would be unfair, and we measured that rather than guessing. Every limit in the game comes from running two solvers over every answer in every pool and taking the ninetieth percentile of a solver that searches the whole dictionary, plus one.
That measurement turned up something counterintuitive that is worth its own post: the four-letter board — the one labelled Easy — is in an important sense the hardest board of the three.
Where to start
Play the regular board first — five letters, eight guesses — and open with SLATE, which our measurement of every legal opener puts within half a word of the best in the dictionary. Expect to still have seventy candidates afterwards. That is normal here, and it is the point.