Thanks for playing! Some puzzles require predicting future actions which is rather challenging as a concept, I think the difficulty scales too quickly, if I were to do a full game ever, will definitely take note of the difficulty curve.
As for bot launching, I'm not sure if computer specs have to do with how accurately it can replay the actions, because the replay is done in a fixed update loop, which in theory should provide the most accurate replay, but it may also run less frequently on lower specs machine. It could also be the uppercutting bot launched the wrong bot up, maybe the future 2 clones were right on the uppercutting bot and it uppercut the wrong bot, as uppercut only launches 1 bot. Will be changing this interaction to affect all bots.