Seriously not trying to be a smart aleck here. It seems like you've just improved an inefficient approach. You're level 3 so beyond me python-wise. What made you choose this approach over ord()?

Agreed. Clever yes. But how can an incomplete solution be a best practice?

Really like the efficiency. Nice one.

I think the first two list comprehensions are easy enough to read. My only thought was, it's iterating throught the source list twice just to break it into evens and odds.

Agreed. In any test where letters appear, there are other problems like too long and/or missing parens or dash. They need to isolate the problem of a non-digit appearing where there should be a digit.

Similar to ost-k's solution. It gives a false positive when a string with a literal '?' appears where there should be a digit. e.g. (333) ?22-1111

I like the idea, but a false positive would result if a string were submitted with a literal 'x' where a digit should appear.