A thing I feel writing Python is that visually decoding exceptions (the primary form code errors come in) is too much cognitive work, and there are things the interpreter could do to help.
Consider this (long, verbose!) extension trace. Only the pink circled part is important. The interpreter can't know that; that's AI-complete. But the interpreter could know which lines are "my code" and which are "library code", and color-code the background. Then I could just scan for the final "Me" line.
