Surprising ? Yeah btrfs can probably recover a lot more than a few bits (never looked at how it works), but what he is saying for "non trivial amount of errors", doesn't seem wrong, and is related to Hamming distance (https://en.wikipedia.org/wiki/Hamming_distance), and in particular things like this quote "Thus a code with minimum Hamming distance d between its codewords can detect at most d-1 errors and can correct ⌊(d-1)/2⌋ errors.". Did I missed something obvious ?