OLL recognition
Reading the case from the stickers instead of counting them, and the single sticker that separates each pair of lookalikes.
By the end you will be able to
- Narrow 57 cases to a handful with two glances rather than by counting stickers
- Read the side stickers as part of the case rather than as an afterthought
- Name the sticker that separates each pair of lookalikes you keep confusing
- Practise recognition on its own, away from the algorithms
Once a group is in your hands, the time it costs you is recognition, not execution. A ten-turn algorithm at four turns a second takes two and a half seconds. Finding out which algorithm it is takes most people longer than that for months.
Counting stickers does not work
The instinct is to count the yellow stickers on the top face. It tells you almost nothing. Twenty-one of the fifty-seven cases show exactly five yellow stickers up there, centre included, and another twelve show four.
Reading the shape rather than the count is better, and still not enough on its own. Fifty of the fifty-seven cases share their top-face pattern with at least one other case; only seven — OLL 17, 20, 25, 28, 39, 40 and 57 — can be identified from the top face alone. Everything else needs the sides.
Two glances
Take the case in two passes rather than trying to see it whole. The first pass is the edges, the second is the corners, and between them they cut fifty-seven cases down to at most twelve.
First, the edge shape. Ignore the corners entirely and look at the four edge stickers on the top face. They make a cross, a line, an L, or a dot — the same four shapes you learnt for two-look.
Then count the corners showing yellow on top. The answer is none, one, two or four. Three is impossible, for the reason the beginner method gave: the twists of the four corners have to balance out, and three corners already facing up leave the fourth with nothing to balance against.
| Edge shape | No corners up | One corner up | Two corners up | All four up |
|---|---|---|---|---|
| Cross | 2 cases | 2 cases | 3 cases | solved |
| Line | 4 cases | 4 cases | 6 cases | 1 case |
| L | 6 cases | 8 cases | 12 cases | 1 case |
| Dot | 2 cases | 2 cases | 3 cases | 1 case |
The third glance: a single sticker
Inside a box, the cases differ only in where the remaining yellow stickers point — and almost always, one sticker decides. Finding that sticker for each pair you confuse, and writing it down, is what turns a three-second recognition into a half-second one.
Here is the pair everyone meets first. Both are T shapes: the same T of yellow on the top face, and nothing on the top face separates them.
The three cases with a cross and two corners up work the same way, and you already know two of them. The top face separates the bowtie — its two oriented corners sit diagonally opposite, which is visible from above — and the sides separate the other two. Headlights has its two remaining yellow stickers side by side on one face. The chameleon has them on opposite faces, one pointing at you and one pointing away.
Hold it the same way every time
Recognition is a habit rather than a calculation, and habits need a fixed viewpoint. Decide now that you read every case with the same two faces towards you, and re-hold to that position rather than reading the case from wherever the last F2L pair left you.
- Read the top face at a slight angle, so the front and left side stickers are visible without moving your head. Most people settle on tilting the cube away from them.
- Never turn the cube over to look. If you cannot see the case, you are holding it wrong, and the fix is a habit rather than a rotation.
- The
Uturn you make to line the case up is part of the algorithm, not part of the recognition. Decide the case first, then decide the turn.
Where the recognition should happen
The last three turns of the final F2L pair are the most valuable in the solve. Your hands are busy with something they know, so your eyes are free — and the top face is already showing you most of the case, because inserting a pair does not touch the edges you are about to read.
Getting the edge shape during that insertion, and the corner count immediately after, takes the OLL pause from a second to nearly nothing. This is the same skill as lookahead in F2L, applied to a place with less going on, because there is only one thing to look at.
Practising recognition on its own
- Name, do not solve. Set up a case, say the name out loud, check, move on. Sixty cases in five minutes, and no turning at all beyond the setups.
- Time the naming. Anything over a second and a half goes on a list. The list is your next practice session.
- Drill the confusions, not the set. If you mix up the two knight moves, drill those two against each other twenty times. Drilling all fifty-seven when six are the problem wastes an hour.
- Write down your own check. For each pair you confuse, write the one sticker that decides, in your own words. The act of writing it is most of the benefit.