Thanks. No, right now the filter only looks at path length. Counting distinct shortest paths is a good idea and nearly free, since the same BFS can count how many ways reach each pair at its distance.

The misleading first move I like even more. Moves aren’t reversible here (a blocked courier just stays put), so it needs a backwards search from the goal pair over the move graph, but at 900 states that’s cheap too. Noted both, I want to see which one matches how hard the days actually feel.

Ha, fair. The side arrows are plain text glyphs, and that font draws them smaller than up and down. Noted, thanks for the screenshot.