Can anyone here comment on what kinds of proofs coq is a good tool for, and what kinds of proofs are impractically hard for it to handle? I'm aware that most proofs aren't formalized, and now and again we hear that mathematicians spend a long time trying to assess whether a proof has errors. From that, I infer that there are proof approaches that are challenging to capture in tools like this (or else any proof not obviously correct could be formalised to resolve the issue). What are those hard-to-capture techniques?