Spaces, colons, semicolons, periods, Lines vs CSV. Plus as far as I know he didn't pre commit to any algorithms or how he did the weighting in the file. Think 1-4 sha3, 5-7 blake2s instead of repeating the hash for each line. Also he ordered them randomly vs bunching.
The space for the file is way bigger than an intuitive first glance.
Also note that sharing the text may not work. He should have shared the file. CR, LF, or CR/LF could cause issues for people trying to replicate.
