# Mission: Blindfolded cubing, aimed at 4x4x4

## Why

I already speedsolve with CFOP, but I have never solved a cube blind, and 4x4x4 is my favourite puzzle.
I want to sit down with a 4x4, look at it once, put on a blindfold, and solve it — reliably, and reasonably fast.

## Success looks like

- A verified blindfolded 4x4x4 solve (memo + execution, no peeking).
- A 4BLD success rate high enough that it is a skill, not a lucky hit — call it 2 in 3.
- Sub-5-minute 4BLD, including memorisation.
- Along the way: a solid 3BLD, since wings (r2) and corners carry straight over from it.
- Able to trace a scramble into letters without a cube in hand.

## Constraints

- 20–30 minutes of practice on most days. Lessons must fit one sitting.
- Recall drilling is split into **three short sessions a day**, WaniKani-style, via
  `speffz-trainer.html`. Lesson design can assume letters are being drilled separately and need
  not spend lesson time on rote recall.
- Corners will be learned as **3-style / commutators**, not Old Pochmann.
  This is the slower on-ramp but the right one here: 4x4 centres are solved with commutators regardless, so the skill is not optional later.
- Already fluent in standard notation, slice moves, wide moves, and finger tricks.
  Nothing needs to be explained in beginner terms.

## Out of scope

- Multi-blind, 5BLD and up.
- Speed-optimising the sighted 4x4 solve (Yau, reduction) — that is already handled.
- Full 3-style alg *memorisation* (hundreds of cases). Commutators will be built intuitively; alg-sheet grinding is a later choice, not part of this mission.

## Route

**Edges first, corners later** — chosen deliberately: an edges-only blindfolded solve is a real
blindfolded solve, reachable in days, and it is motivating in a way that half-learning two piece
types is not.

It also lands a free simplification. With the corners already solved, the edge permutation must
be even, so the target count is always even and **parity can never occur**. The fiddliest part of
blind solving is skipped entirely until it is actually needed.

Wings on 4x4 use **r2**, which is M2 with the slice swapped, so this work transfers almost directly.

| Stage | Content |
|---|---|
| 1 | Speffz lettering, tracing edge cycles ✔ |
| 2 | The M2 shot — setup/M2/undo, the 16 off-slice targets ✔ |
| 3 | M-slice targets, the pair rule, **first edges-only blind solve** ← **current** |
| 4 | Speed and consistency on edges-only |
| 5 | Corners: tracing, then 3-style commutators |
| 6 | First full 3BLD success |
| 7 | 4x4: wings via r2 |
| 8 | 4x4: centres via commutators |
| 9 | 4BLD parity, full solve, then speed |