Notes: - This is a testing project to test a Pure Monte-Carlo search approach with the simplest game there is, Tic-Tac-Toe! Overall, this bot can play well, but the way I did it, not wanting to implement anything else, makes it so that this bot can fail, especially with 'subtle' lines or when one line has better results, even if the opponent can force a win. From my research, I think I would use some sort of minimax or negamax algorithm instead, as I can search possibilities; I just need to learn how to determine evaluations and cut off potential lines. - This project allows you to determine the "difficulty" of the bot (basically, how many simulations it runs, from 100 to 5,000, though it may not affect accuracy all that much) and whether you want to go first or not. The project displays the bot's evaluation of the position up in the corner, where a positive number means you have better results, and a negative number means you have worse results. It is not really a true evaluation, in the sense that it relies on randomness and the law of large numbers. - I realize this performed even worse than I thought it would on Scratch compared to TurboWarp, so I had to tweak a couple of things. Overall, the project works fine; however, this approach on its own isn't ideal. - This project and the thumbnail aren't really all that polished, but I just wanted to get this out as soon as possible. I tried something different with the user input, so be wary with your clicks. - This is a bit of a precursor to any other ideas I have with randomness, and it looks like I won't be using this if I really want accuracy, and that this isn't all that feasible in Scratch, even for a simple game like Tic-Tac-Toe. - Do your own reading on this, reader! - This project I think, is pretty well programmed in certain areas, especially dealing with issues, and my own custom blocks that do exactly what they're supposed to efficiently. _________________________________________
