Hi all,
in case you don't know, https://en.wikipedia.org/wiki/FreeCell is a single-player card game, that became popular after being included in some versions of Microsoft Windows. Now, the English Wikipedia entry about it used to contain during at least two times in the past, some relatively short sections about several automated solvers that have been written for it. However, they were removed due to being considered "non-notable" or "non-Encyclopaedic".
Right now there's only this section - https://en.wikipedia.org/wiki/FreeCell#Solver_complexity which talks about the fact that FreeCell was proved to be NP-complete.
I talked about it with a friend, and he told me I should try to get a "reliable source" news outlet/newspaper to write about such solvers (including I should add my own over at http://fc-solve.shlomifish.org/ , though the sections on the FreeCell Wikipedia entry did not exclusively cover it.).
Recently I stumbled upon this paper written by three computer scientists, then at Ben-Gurion University of the Negev:
* http://www.genetic-programming.org/hc2011/06-Elyasaf-Hauptmann-Sipper/Elyasa...
* There's some analysis of this paper in this thread in the fc-solve-discuss Yahoo Group:
https://groups.yahoo.com/neo/groups/fc-solve-discuss/conversations/messages/...
The solver mentioned in the paper can solve 98% of the first 32,000 Microsoft FreeCell deals. However, several hobbyist solvers (= solvers that were written outside the Academia and may incorporate techniques that are less fashionable there, and that were not submitted for Academic peer review) that were written by the time the article published, have been able to solve all deals in the first MS 32,000 deals except one (#11,982), which is widely believed to be impossible, and which they fully traverse without a solution.
Finally, I should note that I've written a Perl 5/CPAN distribution to verify that the FreeCell solutions generated by my solver (and with some potential future work - other solvers) are correct, and I can run it on the output of my solver on the MS 32,000 deals on my Core i3 machine in between 3 and 4 minutes.[Verification]
===========
Now my questions are:
1. Can this paper be considered a reliable, notable, and/or Encyclopaedic source that can hopefully deter and prevent future Deletionism?
2. Can I cite the fc-solve-discuss’s thread mentioning the fact that there are hobbyist solvers in question that perform better in this respect - just for "Encyclopaedic" completeness sake, because the scientific paper in question does not mention them at all.
===========
Sorry this E-mail was quite long, but I wanted to present all the facts. As you can tell, I've become quite frustrated at Wikipedia deletionism and the hoops one has to overcome in order to cope with them.
Regards,
Shlomi Fish
[Verification] - one note is that all these programs were not verified/proved as correct by a proof verifier such as https://en.wikipedia.org/wiki/Coq , so there is a small possibility that they have insurmountable bugs. Note that I did write some automated tests for them.