Seeing this comment made me try to answer that, just for fun. For improvements, I know that I would have included a more obvious manual somewhere (I didn't know that you could flip the pieces, rather than rotating them with a space bar, until I read that in a YC comment).
The second question is more interesting to me, because I know regexes better than Turing machine variants. I think I could give them the problem of accepting only balanced, nested "parentheses" with arbitrarily deep nesting or something similar (treating blue as open parentheses and red as close). Though they asked something close to that with the nasty "accept only bots with equal numbers of red and blue symbols" so maybe I'm wrong and I should look for some angle with incomputable functions, like the Busy Beaver.
Or you can just cheat and give the play a board that's too small (or limit them to too few pieces) for the problem to make it unsolvable.
The ability to use yellow and green makes me wonder if there's some way to solve the balanced parentheses problem, though. I quit to do something else before solving the "accept only equal numbers of red and blue" problem (alas, I wished it would let me write to balance the tape, then you could just loop until empty, write red blue and accept).
This would sure be a lot more fun than most programming tests. I had a lot of fun designing things like a "red eater" which took care of the .* in the problem asking you to match .*BB$ and trying to shave parts. I know I could get rid of several more if I had known how to flip them. I just hate seeing something and knowing there ought to be a way to simplify it further.
I found this game few days ago. I had few approaches to accept only equal number red/blue. First I attached binary yellow/green counter starting at 1111 to the end of the tape and with each pass of the tape I removed one symbol and incremented the counter if the symbol I removed was blue and decrementing it if it was red. At the end I checked if counter is still 1111
Worked perfectly until robot with only one empty space at the end of his tape was encountered.
The second question is more interesting to me, because I know regexes better than Turing machine variants. I think I could give them the problem of accepting only balanced, nested "parentheses" with arbitrarily deep nesting or something similar (treating blue as open parentheses and red as close). Though they asked something close to that with the nasty "accept only bots with equal numbers of red and blue symbols" so maybe I'm wrong and I should look for some angle with incomputable functions, like the Busy Beaver.
Or you can just cheat and give the play a board that's too small (or limit them to too few pieces) for the problem to make it unsolvable.
The ability to use yellow and green makes me wonder if there's some way to solve the balanced parentheses problem, though. I quit to do something else before solving the "accept only equal numbers of red and blue" problem (alas, I wished it would let me write to balance the tape, then you could just loop until empty, write red blue and accept).
This would sure be a lot more fun than most programming tests. I had a lot of fun designing things like a "red eater" which took care of the .* in the problem asking you to match .*BB$ and trying to shave parts. I know I could get rid of several more if I had known how to flip them. I just hate seeing something and knowing there ought to be a way to simplify it further.