Lesson 3 · Blindfolded cubing
Six stickers left, one rule to go with them — and then you put the blindfold on.
Today ends with a real blindfolded solve. Not a drill, not a simulation: a scrambled cube, one look, blindfold on, solved. Everything below exists to get you there.
Every setup in lesson 2 had one job: bring the target to where M2 can reach it,
without disturbing the M slice. That works right up until the target is on the M slice.
You cannot set up a piece that lives in the slice you are about to turn.
There are six such stickers — the buffer's own U and K are never targets, which leaves:
Every algorithm here is machine-checked: each is confirmed to exchange the buffer with its target and apply M2's side effect, exactly as the sixteen setups do. Sources are the Speedsolving wiki M2 tables.
This is the one piece of M2 that catches everyone, and it follows directly from what
M2 does. Recall the side effect: every shot also swaps UF with DB.
So after an odd number of shots, UF and DB are currently swapped. If your next target lives on one of them, the piece you want is really in the other place.
Count your targets from one. If C, I, W or S falls in an even position — the second half of a letter pair — use its partner's algorithm instead.
C ↔ W I ↔ S
A and Q need no rule at all. They live on UB, which the side effect never touches, so they behave the same in either position.
This rule is not folklore. Running traced memos through the algorithms: with the rule, 2018 of 2018 scrambles solve. Without it, roughly one in four.
M2, undo —
applying the pair rule when an M-slice target lands in an even position.Hold the cube in a fixed orientation the whole way through: white up, green front, all the way from tracing to the last move. Losing your orientation mid-solve is the most common way a first attempt dies, and it is entirely preventable.
This gives you a scramble that leaves the corners solved, so you can go straight to edges. Apply it, trace it, type your memo, and check before you commit. When it matches, put the blindfold on and solve. The moves are there afterwards if it goes wrong.
It will, a few times. The failure tells you where the fault is:
| What you find | Almost always |
|---|---|
| Two edges swapped, rest solved | You missed the pair rule on an M-slice target |
| Three edges cycled | One target wrong in the memo, or shot out of order |
| An edge flipped in place | Right piece, wrong sticker while tracing |
| Centres rotated, edges fine | Odd number of shots — a target got dropped |
| Total chaos | Lost cube orientation partway through |
Check the memo before blaming the execution. Tracing errors outnumber turning errors by a wide margin at this stage.
Primary source
J Perm — How to Solve a Rubik's Cube Blindfolded. Worth rewatching now rather than at the start: the parts about holding orientation, committing memo to short-term memory, and staying calm mid-solve mean far more once you have a method in your hands than they did in lesson 1.
When you get one — tell me. And if you get a near miss, tell me what the cube looked like at the end; the pattern of what is left is usually enough to name the mistake exactly.