This algorithm is essentially Minimax, but it chooses the second-best option for your opponent if there is enough of an evaluation difference between your opponent's first and second-best options and there are a substantial number of move choices on your opponent's turn. That is, it prioritizes traps by assuming that your opponent will make a mistake but that, if they get it correct, your evaluation is still in the "drawish range."
It is primarily useful if there is no clear winning variation (pure MiniMax yields no clear win) but that one wants to find positions where your opponent has a low probability of randomly picking the best choice, thereby acting as a kind of trap-finding chess second.
In order to avoid forced moves that aren't traps, it only considers moves that aren't a check or capture/recapture.
There are other conditions that can be added for what constitutes a trap candidate, like weighing the trap evaluation difference based on the number of moves in the position.
This algorithm can be optimized because only first and second-best options need to be chosen, so all of the child nodes don't need to be sorted.
This can be further optimized using alpha-beta pruning on the half of choices that you make and an alternative alpha-beta pruning on your opponent's choices. I will be working on that alternative ongoingly.