The New Sudoku Players' Forum

[SpAce]

an exact Hodoku score for a hard puzzle is a bit of a lottery (or maybe I don't know enough about Hodoku to keep the scores consistent)

I wouldn't worry about the Hodoku scores because they're cumulative and thus depend on the full solve path. It makes them quite arbitrary when there are multiple possibilities for the same technique with varying degrees of effectiveness. The only part we can control is the hierarchy of different techniques, but Hodoku decides the order of applying instances of the same technique. If it happens to pick an effective instance early, the score is obviously lower than if it first uses all the irrelevant ones. I don't know the exact algorithm it uses for that, but I think we should consider it random for all practical purposes, making the cumulative scores nondeterministic.

SER is not perfect but I think it's much better for these kinds of comparisons than any cumulative scores. For your latest pair of puzzles the SER scores are 4.6 and 8.3.

[tarek]

The technique selection in most solvers is biased towards the current sudoku configuration and towards how the solver searches for eliminations in the current technique. This may result in under or over rating a puzzle

It was suggested that the puzzle should be solved & rated based on the canonical form which removes the configuration bias.

There is also the techniques that are of similar rating (which one do you choose 1st?)

Another and possibly better way for electronic solvers is to apply all techniques & elimination available at this stage that are of similar rating/level (essentially batch solving)

Sukaku Explainer has this feature available using the command line solver with the -b1 and -b2 options (This speeds up solving puzzles too in most cases)

Tarek

[StrmCkr]

Another and possibly better way for electronic solvers is to apply all techniques & elimination available at this stage that are of similar rating/level (essentially batch solving)

Thats how mine works but its super slow especially when it hits als where there can be in exceas of 90k varations at a step