Why we pivoted then
Family 6 had reverted seven times in a row (0080-0086), past the five-revert rule, because the agreed order of directions (type 1, then type 2) came first. Once type 1 and the move-based part of type 2 were closed, the last open items of type 2 were keys built from two parts.
What happened before
- 0080 put a repeating key on both sides of every move, solved as a linear system: nothing beyond chance.
- 0081 joined both autokeys to K3's rule, grid turns and columnar: chance.
- 0082 and 0083 covered hand padding and every triple quarter turn, 1.7 million routes: chance.
- 0084 tested double columnar with one width of 7 in all 5,040 orders: chance.
- 0085 reran keyword columnar with the true key order after issue #26 found the inverse reading in ten records: chance.
- 0086 counted the ? as a 98th symbol in a full 7 x 14 grid: chance.
The change we believed was needed
K4's doubled letters line up in columns of seven. A key made of a column word plus a row key seemed the one construction that would line them up by design, so the next experiments tested a column word plus a row key, and the same with a shift per stretch of letters (0087-0088). 0096 later showed that premise was wrong.