[socalgal2]

Do they? Tons of extremely popular human generated libraries are absolute trash. Just as an example, nearly all of the JS zip file libraries are dumpers fires. Same with QR code libraries and command line parsing libraries.

This feels like if you want to know if the code is good or bad, read the code and check the tests. Assuming human = good, LLM = bad does not make much sense given the amount of bad human code I've seen.

Sure, if the code is from a repuatable company or creator then I'd take that as a strong signal quality over an LLM but I wouldn't take a random human programmer as a strong signal over generated.