the part about the "wild" guess is, but this is a counterexample, and something like the reciprocal vs the negative of reals or rationals when moved across the log map would be an example. So, either you're a galaxybrain that just instantly knows if the transformation is structure-preserving in that way, or you're guessing to some degree as well.

The symbols and abstractions have touched me in no-no ways. I miss okaybuddyphd on r*ddit, they knew the pain.

I suppose I could also just say that characters which aren't just drawn asymmetrical, but actually point in a direction as part of their function, look wrong when reversed like this. So, (e) -> )e( is no good, but bed -> deb is fine.