Shot in the dark on Super Mario 64 Big Star Secret
Mar 16, 2024 10:28:36 GMT
MediaMonster, MantleM, and 3 more like this
Post by sindex on Mar 16, 2024 10:28:36 GMT
Hi; I know this is a pretty old and dead topic at this point, but I didn't want to bump such an old thread.
A bit over a week ago I took a deep dive into the video. I mean an excruciating deep dive. I am hyperfixated on the Internet Archive.
When requesting a video, YouTube looks to an API called "videoplayback", found on the root of an arbitrary server. From late 2009 to some time in 2014, these servers were hosted on "c.youtube.com" (eg. "rr4---sn-tt1elnel.c.youtube.com"). Since 2014, they have been hosted on "googlevideo.com", which was previously used for Google Video before it shut down.
Before late 2009, an API called "get_video" was used, but it only required the video ID (eg. dQw4w9WgXcQ), so it's irrelevant to what I'm focusing on.
"videoplayback" links have plenty of parameters to strengthen security, ranging from an expiration timestamp to the IP of the requester, though only one can be used to identify a video: the ID. Not a video ID, rather a 16-byte hex string (which I presume to be a hash of some sort) called the content ID. Similarly, an API called "initplayback" contains that ID, though far fewer parameters otherwise.
I used to think that the content ID could only be retrieved by looking through the HTML of a video that was public. This was until I checked Super Mario 64 big star secret and found an "initplayback" link.
In the HTML of Super Mario 64 big star secret, a video which has *long* been deleted off of YouTube, you can find the following link: rr2---sn-tt1elnel.googlevideo.com/initplayback?source=youtube&oeis=1&c=WEB&oad=3200&ovd=3200&oaad=11000&oavd=11000&ocs=700&oewis=1&oputc=1&ofpcc=1&siu=1&msp=1&odepv=1&id=4cf016a471b74566&ip=CENSORED&initcwndbps=2771250&mt=1710582788&oweuc=
Any suspicion of that link being fully derived from the video ID is lost when looking at any non-existent video, where no such link is present. You can check by using CTRL+F in the source and searching "26id=".
Any suspicion of the ID being unrelated to the video is lost when the content ID found in the HTML of the following deleted video ā www.youtube.com/watch?v=wmr0d4ztEBE ā can be traced back to the following archived "videoplayback" link: r16---sn-a5m7ln7s.c.youtube.com/videoplayback?id=c26af4778ced1011...
All of this proves beyond a reasonable doubt that Google has, and can easily access (through the video ID), information regarding deleted videos. Additionally, the content of the page of a private video heavily mirrors the content of the page of a deleted video, as can be seen here: www.youtube.com/watch?v=xDKb4BML14o. This could mean anything, but it's promising that the two show the same amount of data.
IIRC, lotusman already contacted Google for the video to some extent, but I felt I might as well share my findings regardless since it's 6:30 in the morning. I hope people find it interesting for general searching if not this search in particular.
(Also, I did use the content ID to check through all archives of "c.youtube.com" and "googlevideo.com", along with the "get_video" API I mentioned earlier. Safe to say, the video is not on the Internet Archive.)
UPDATE: content IDs are just the video IDs encoded in hex. I think I just entered invalid Base64 and that's why nothing showed up for a "non-existent video".
To be more specific: video IDs are displayed as modified Base64, specifically with '+' replaced with '-' and '/' replaced with '_' along with 1 character of padding removed. For example, the video id -VgS5In-I_c would be represented in proper Base64 as +VgS5In+I/c=
A bit over a week ago I took a deep dive into the video. I mean an excruciating deep dive. I am hyperfixated on the Internet Archive.
When requesting a video, YouTube looks to an API called "videoplayback", found on the root of an arbitrary server. From late 2009 to some time in 2014, these servers were hosted on "c.youtube.com" (eg. "rr4---sn-tt1elnel.c.youtube.com"). Since 2014, they have been hosted on "googlevideo.com", which was previously used for Google Video before it shut down.
Before late 2009, an API called "get_video" was used, but it only required the video ID (eg. dQw4w9WgXcQ), so it's irrelevant to what I'm focusing on.
"videoplayback" links have plenty of parameters to strengthen security, ranging from an expiration timestamp to the IP of the requester, though only one can be used to identify a video: the ID. Not a video ID, rather a 16-byte hex string (which I presume to be a hash of some sort) called the content ID. Similarly, an API called "initplayback" contains that ID, though far fewer parameters otherwise.
I used to think that the content ID could only be retrieved by looking through the HTML of a video that was public. This was until I checked Super Mario 64 big star secret and found an "initplayback" link.
In the HTML of Super Mario 64 big star secret, a video which has *long* been deleted off of YouTube, you can find the following link: rr2---sn-tt1elnel.googlevideo.com/initplayback?source=youtube&oeis=1&c=WEB&oad=3200&ovd=3200&oaad=11000&oavd=11000&ocs=700&oewis=1&oputc=1&ofpcc=1&siu=1&msp=1&odepv=1&id=4cf016a471b74566&ip=CENSORED&initcwndbps=2771250&mt=1710582788&oweuc=
Any suspicion of that link being fully derived from the video ID is lost when looking at any non-existent video, where no such link is present. You can check by using CTRL+F in the source and searching "26id=".
Any suspicion of the ID being unrelated to the video is lost when the content ID found in the HTML of the following deleted video ā www.youtube.com/watch?v=wmr0d4ztEBE ā can be traced back to the following archived "videoplayback" link: r16---sn-a5m7ln7s.c.youtube.com/videoplayback?id=c26af4778ced1011...
IIRC, lotusman already contacted Google for the video to some extent, but I felt I might as well share my findings regardless since it's 6:30 in the morning. I hope people find it interesting for general searching if not this search in particular.
(Also, I did use the content ID to check through all archives of "c.youtube.com" and "googlevideo.com", along with the "get_video" API I mentioned earlier. Safe to say, the video is not on the Internet Archive.)
UPDATE: content IDs are just the video IDs encoded in hex. I think I just entered invalid Base64 and that's why nothing showed up for a "non-existent video".
To be more specific: video IDs are displayed as modified Base64, specifically with '+' replaced with '-' and '/' replaced with '_' along with 1 character of padding removed. For example, the video id -VgS5In-I_c would be represented in proper Base64 as +VgS5In+I/c=