Blindfolded solving

A blindfolded solve has two phases on one clock: look at the scrambled cube and memorise it, then put on the blindfold and solve it without seeing it. There is no inspection; the memorisation is timed (WCA Regulation B1a). The blindfold goes on before the first move, and the solver may not look again (B4).

Speedsolving methods work by looking. Blindfolded methods instead solve pieces one or two at a time with algorithms that move only those pieces, so the solver can track the whole solve in memory. One spot of each kind is the buffer: each algorithm sends the piece in the buffer to where it belongs and brings a new piece into the buffer. The solver memorises the order the pieces will be sent, then executes it.

Each sticker that can be a target gets a letter. The common scheme is Speffz: A to D on the top face, E to H on the left, I to L on the front, M to P on the right, Q to T on the back and U to X on the bottom, going clockwise round each face from its top-left piece. Corners and edges use the same letters, so a letter names a sticker only together with the piece kind. The same rule letters the 4x4's and 5x5's wings and centres.

Start at the buffer. Look at the sticker in the buffer's top spot and find where it belongs: that spot's letter is the first target. Then look at the piece in that spot and find where it belongs: the next target. Continue until the next target would be the buffer piece itself: the cycle is closed. If pieces are still unsolved, break into one of them (the first in letter order is a common rule): memorise its letter, follow from there, and memorise its letter again when the cycle returns to it. Pieces in their place but turned (twisted corners, flipped edges) are noted separately.

The memo trainer, on the Learn page, draws a scramble with its letters and traces it for you, kind by kind, so you can check your own tracing.

Letters are memorised in pairs. Each pair becomes a word or an image (AB could be 'abbey'), and the images are linked into a story or placed along a route you know (a memory palace). Many solvers memorise edges as images and corners by the sound of their letters, which they repeat as they solve. Learning a fixed image for every pair takes time but makes memorisation much faster.

On a 3x3 the corners and the edges always have the same parity: an odd number of corner targets means an odd number of edge targets. With letters in pairs, the last letter of each is left alone; one parity algorithm swaps the last corner target with the buffer and two edges at once. Twisted corners and flipped edges need their own algorithms, or can be solved inside a cycle by targeting a sticker of the piece and then its other sticker.

In Multi-Blind the competitor names the number of cubes beforehand, at least two (H1a), memorises all of them, then solves them all blindfolded. The time limit is ten minutes per cube for fewer than six cubes, and an hour otherwise (H1b). The result is the number solved minus the number unsolved; below zero, or with only one cube solved, the attempt counts as unsolved (9f12c).

Multi-Blind memorisation leans on long-term memory: each cube gets its own route or part of a route, solvers review earlier cubes as they go, and execution recalls one cube at a time. Choosing how many cubes to attempt balances the time limit against the risk of forgetting.