Probe
Five candidate sets a day. One job: pick the guess that splits them best.
Loading today's rounds…
Each bar is one round: your probe's elimination as a share of the best probe's. Bars fill as you play; 100% means you matched the best available split.
Scores and streaks are stored only in this browser. The share string carries round bars and your total — never any words.
What this game trains
Mid-game Wordle turns on one skill: gathering information while the field is still wide. Probe isolates it. Each round shows every word in the site's working answer pool that still fits a real opening line, and asks for the one guess — any accepted word, not just a candidate — that best splits them into distinct feedback groups. Small groups mean the next reply almost names the answer; one big clump means a wasted turn. The arithmetic is the same partition math as the site's solver, and the sibling game Solver Says grades the move you would actually play. Solver Says tests the decision; Probe trains the instinct behind it, and Last Row closes the loop with pure endgame deduction. For the same discipline outside words, How Sure? scores how well you know what you know. Five rounds a day, easiest first.
How scoring works
Every probe gets a value from one measure, expected elimination: group the candidates by the feedback pattern your probe would receive, sum (group size)² divided by the candidate count to get the expected remainder, and subtract that from the count. It is the average number of candidates your probe rules out. Your round score is that elimination as a percentage of the best available — "best" meaning the lowest expected remainder among all 14,855 accepted guesses for that candidate list, with ties broken by the smallest worst-case group. When several probes tie on value, the reveal says so.
| Daily total | Band | What it means |
|---|---|---|
| 95–100% | Optimal eyes | Round after round you matched the best available split — for these candidate lists, no accepted guess gathers more. |
| 80–94% | Sharp | You gathered most of the information on offer; the misses were near-best probes, not wasted turns. |
| 60–79% | Steady | Solid splits with real information left on the table — usually one round where a big clump survived. |
| 40–59% | Drifting | Your probes let large groups of candidates through together. Look at which letters each reveal actually tested. |
| Below 40% | Guessing | The field barely shrank. The habit to build: before probing, scan the candidates for letters no guess has tested yet. |
The daily total is the plain average of the five round scores, so 100% means five best-available probes. A share bar fills at 80% — the level where a probe stops costing you turns against the best line.
Rounds are built offline from the site's 2,360-word working answer pool and its reconstructed accepted-guess list; candidate counts describe this site's pool, not what the NYT will choose. Every round is screened so that probing a candidate itself is strictly suboptimal — the best candidate probe trails the best probe by at least 0.75 expected candidates. That is the built-in lesson: while the field is wide, a word that cannot win can still tell you the most.