New Research

[Robotguy]

It's been a while since anyone posted (or probably even looked) here, but I thought I'd post this in case anyone comes searching:

 

How Data Mining The Web Reveals What Makes Puzzles Hard For Humans

 

"Pelanek has gathered hundreds of thousands of hours of data on the way humans solve Sudoku puzzles on websites, such as www.sudoku.org.uk. His data focuses on 2000 puzzles each solved by hundreds of individual players."

 

https://medium.com/the-physics-arxiv-blog/99a5d87e9235

[Larry Hosken]

Thanks for the link!

tl;dr

Radek Pelanek at Masaryk University Brno in the Czech Republic studied
sudoku, hoping to figure out why some sudokus are harder than others.

His data focuses on 2000 puzzles each solved by hundreds of individual

players. (Hooray for online puzzle sites that make this possible.)
It's not just the complexity of the puzzle-solving steps.

His key insight is to ask whether the individual steps in the solution
are independent and so can be applied in parallel; or whether they are
dependent on each other and so must be applied in series. His model of
puzzle solving captures both of these aspects. He says the correlation
with the measured data is 0.95.

> --
> You received this message because you are subscribed to the Google Groups
> "Puzzle Theory" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to puzzletheory...@googlegroups.com.
> To post to this group, send email to puzzle...@googlegroups.com.
> Visit this group at http://groups.google.com/group/puzzletheory.
> For more options, visit https://groups.google.com/d/optout.