What happens if you download a YouTube video more than one time? You won't get the same video files: the hashes won't match. YouTube itself and/or FFmpeg combining streams together = non-deterministic data, sadly. (FFmpeg is more to blame I think; if you downloaded just an audio stream or video stream from YT multiple times, I guess it would be the same each time.) Real example - 1st download: >$ ipfs ls /ipfs/QmWvaCyeUpUnyNvLL6QzW4GeLEHnPPRSLEVBwCkzJkjXQP >[...]QmW6KkhztTusP1ASFDji8HqnFZnw3JvaurSvkYmenAvfW9 51445407 Through_the_Fire_and_Ponies_PMV.mp4-Apple_Dash-20120409-youtube-960x720-FGpdrxHr3Cw.webm[...] 2nd download: >$ ipfs add --only-hash --raw-leaves --chunker=size-1048576 $f >added QmdwALb4k3SS1qeJw2w7xfRgiAg9Rfd8DJsAQSxkSqT9We Through_the_Fire_and_Ponies_PMV.mp4-Apple_Dash-20120409-youtube-960x720-FGpdrxHr3Cw.webm > 49.06 MiB / 49.06 MiB [...] 100.00% >$ # same 51,445,407-byte filesize, but no match (QmW6...vfW9 =/= Qmdw...T9We)
Bytes are different even at the start of each file: >$ ipfs cat QmW6KkhztTusP1ASFDji8HqnFZnw3JvaurSvkYmenAvfW9 | xxd | head -n18 >[...]000: 1a45 dfa3 9f42 8681 0142 f781 0142 f281 .E...B...B...B.. >00000010: 0442 f381 0842 8284 7765 626d 4287 8104 .B...B..webmB... >[same] Lavf58.76.100D..[same] >00000110: 0000 0000 0000 42d7 8101 73c5 88ab 290e ......B...s...). [ <- different ] >$ cat /z9/youtube/Apple_Dash_UCYApxt-C8NrVKJhjkj-B76w/Through_the_Fire_and_Ponies_PMV.mp4-Apple_Dash-20120409-youtube-960x720-FGpdrxHr3Cw.webm | xxd | head -n18 >00000000: 1a45 dfa3 9f42 8681 0142 f781 0142 f281 .E...B...B...B..