Games

I think I found a typo in a Cryptoquip.

I have four algorithms in my second attempt at writing algorithmic “players” for the game of Squava:

1. Alpha-beta minimaxing
2. Alpha-beta minimaxing with a better static evaluation function
3. Monte Carlo Tree Search
4. Monte Carlo Tree Search with UCT child selection

At this point, all but plain MCTS (#3 above) regularly beat me. How do my programmatic squava players do against each other?

United Launch Alliance, an aerospace industry roll-up of General Dynamics Atlas, Martin Marietta Titan and McDonnell Douglas Delta launch vehicle divisions, put out this incredibly awful set of 9 Sudoku puzzles on Twitter. The purpose was to whip up enthusiasm for the 2021-09-21 launch of Landsat 9.

I made a weird third guess due to coffee deficiency.

9 board tic tac toe is a tic tac toe variant played on 9, 3x3 tic tac toe boards, arranged in a 3x3 pattern:

NYT Wordlebot claimed I had two choices after my 3rd guess. Let’s see if it’s correct.

I was curious about whether the “hint” that you get with an official Cryptoquip makes any difference.

I was also curious about how hard it is to construct a decent Cryptoquip.

Previous Cryptoquip Post

Cryptoquip Solver Github Repo

I wrote a Cryptoquip solver.

My big contribution here is using the “shape” of the words to narrow the choice of clear text letters.