MATCH is a time attack: a sample number appears, and you tap its exact match among lookalike options. Wrong answers add a one-second penalty, so speed means nothing without accuracy.
It's a classic visual-search task turned competitive. Your total time goes on the global leaderboard, and top runs come down to shaving tenths of a second.
▶ MATCH — Play free nowOptions multiply and digits grow as you progress, with deliberately confusable pairs (1/7, 6/8) mixed in. Careless scanning gets punished.
Aggregated from 21 personal-best records stored on MIKIRI's own servers (as of 2026-08-07; one record per player, flagged runs excluded). The sample is still small, so read this as where early players are stalling rather than a settled trend.
The median is 30–35s. Beat it and you are in the faster half.
| Result | players | share | |
|---|---|---|---|
| 15–20s | 2 | 10% | |
| 20–25s | 1 | 5% | |
| 25–30s | 7 | 33% | |
| 30–35s | 5 | 24% | |
| 35–40s | 4 | 19% | |
| 55–60s | 1 | 5% | |
| 70–75s | 1 | 5% |
MATCH is a visual search task: find the number that matches the sample among lookalikes. The more the target resembles the distractors, the more you are forced to check candidates one at a time, and search time grows with the number of items on the board.
In the real data, clear times cluster between 25 and 35 seconds. What separates fast players from slow ones is not eye speed but whether the search follows a rule. Wander at random and you will re-check the same corner twice while missing another.
Memorising the whole sample. Re-encoding every digit means the memory fades each time you look back at the board. Hold the last two digits, use them to filter, and verify in full only when something matches.
No scan order. Z-pattern or column by column — either works. Fixing an order is what makes the unsearched area obvious.
Skipping the check. With this many lookalikes, a wrong tap costs you time outright. Re-checking one digit before committing pays for itself.