CheckerCruncher is entering public beta! It works reasonably well in
the major browsers at desktop resolutions. It’s got a database of
22,000 games, 2700 tactics problems, and 10 friends testing it. There
are lots of features I still want to add but the puzzles are both
interesting and fun. I’ve routinely found myself playing checkers
instead of programming and it’s past time to show it to a wider
audience and get some feedback from strangers on the internet. I’m
both excited and nervous.
The heart of the site is the problem generator; it scans the database
of games looking for problems, builds the solutions, and saves them
to the database for users to practice on. These problems are the core
experience of the site so every improvement helps. I’d love to hear
your feedback both about the problems and the site. I know a little
about game programming but almost nothing about checkers so the
heuristics it uses are crude and some of the puzzles it builds are
ugly.
The worst are the ones
like problem 20
with with lots of winning lines, some of which
are crushing but the computer wants only the fastest win. Try-again,
try-again, try-again, try-again, wrong you lose, is extremely
frustrating. I’ll disable these as I find them, point them out on the
forum. I can’t give you your points back but I can spare someone else
the same pain.
Others
like 640
end too early which can be confusing or
like 683
go too long ending with a whimper instead of a bang.
These types of errors I can’t do as much about yet because they’re so
common. Still, I’ll disable the most egregious examples if people
hate them.
The best problems (and there are lots of these!) are ones where it
gives you the satisfaction of finishing the combination and then
ends. Problems
like 127
are a joy to play. I think these are great at
revealing the depth and beauty inherent in the game of checkers.
They’re also fantastic learning tools, my play has improved a great
deal from these problems. I hope you enjoy solving them as much as I
have building them.
Some of my favorites:
The next iteration of the generator will be more discriminating in
the positions it turns into puzzles. This should avoid generating
most of the frustrating puzzles with lots of ‘try again’ lines. It
will also be smarter about choosing where to end the puzzles. But
before I get to building that there are a lot of features I want to
add for the current problem set. Some of my ideas are listed below
but please share your own ideas, I want to hear them.
Possible improvements in the next few weeks/months.
-
Problems selected by rating instead of randomly.
-
Piece movement sounds.
-
Better mobile and browser support.
-
Better user statistics.
-
Thousands more problems.
-
Showing common errors from other users.
More difficult Ideas that I’ll pursue as time permits.
-
Difficulty settings.
-
Problem comments, tags, and star ratings.
-
Leader boards.
-
Different piece sets.
So that's where CheckerCruncher is at. This site has taken two years
to build and I’m super excited to start sharing it. I think it’s
already a wonderful learning tool and there are tons of improvements
still coming. Please enjoy it and let me know what you’d like to see
changed. Good luck!
Regards,
Brooks.