PDB P1261566

[andrew-buchanan]

NBb2k1r/KpPp1ppp/1P1P4/1PpPp3/5p2/2P5/8/8

I set the config parameters 1000 & Yes.

Basically, the retro argument should be this.
(1) All white captures occur by pawns on queenside: axb, fxe, exd, exd, dxc
(2) So wPg&h never left their files, and were waylaid.
(3) All black pawn captures occur on queenside: axbxc, cxdxe, exf
(4) Black can't retract ↶e7-e5 because the 5 black pawn caps are all on dark squares

So bottom line candidate solutions should be ↶Ke8-f8, ↶Kg8-f8, ↶Rg8-h8, ↶f5-f4 (with no captures).
(Only DP reasoning can distinguish between these three, which is out of scope for Retractor.)
Thanks so much for all your work!

[hwatheod]

This is a known weakness in Retractor right now, where it doesn't do pawn capture analysis beyond counting how many there are for each side. It can't figure out "waylaid pawns" (item 2) and similar types of analysis that require analyzing what the pawns did/didn't capture.

By the way, a simpler reason that Black can't retract e7-e5 is because it would add one more White capture of the Black f8 bishop on its home square. Retractor detected this.