return-youtube-dislike
return-youtube-dislike copied to clipboard
(Feature Request): K-anonymity
Extension or Userscript?
Both
Request or suggest a new feature!
Currently, the video ID is sent in plain text, allowing the operator of the backend to potentially track users' watch histories. These are known to be very valuable advertisers, so it is conceivable (though IMO unlikely) that the current operator is sharing this data with advertisers, data krakens and other immoral businesses. Users have to trust them to not do that.
The Sponsorblock extension was faced with a similar issue and implemented K-anonymity. With a sufficiently small prefix, it allows users to request the 3rd party metadata for a video without exposing (much) information about which exact video the user is actually viewing.
This is approach is also used in https://haveibeenpwned.com/Passwords.
Since the back-end is closed source and couldn't be remotely assessed if it wasn't either, creating such a near-zero-trust system is beneficial to RYD and its users alike.
Ways to implement this!
No response
Can you work on this?
- [ ] Yes
- [X] No
Base on the example provided in that github page, the user requests for one video, the server returns ten results. That means your server traffic is multiplied by approximately ten times (you still need ACSII for communication that is eventually sent over PHY). It also requires more processing power; at least, you will need ten times of memory just to store the characters.
Let's see if Anario's server can handle this.
If there are hardware load concerns, I'm sure people will help to crowd-fund that specific purpose.
How would we implement K-anonymity for disliking a Youtube video? I'm interested in implementing this
I don't think it's possible to do that for submission, only for requests.
I don't think it's possible to do that for submission, only for requests.
The only way to make submissions anonymous would be to take video id hashes instead of the actual plaintext IDs. With that K-anonymity wouldn't even be that relevant, but still it would be an improvement, since sometimes it's obvious what the video is, even just from the dislike amount and how often it's visited. But I don't think this will be the route to go with this project.
If that won't be done, yeah, K-anonymity will only be useful for requesting the data, however that's still the huge majority of the requests, and so it's still worth implementing, even if sending the data will still send the full video id.
Is not this issue pretty much this same thing as this one? https://github.com/Anarios/return-youtube-dislike/issues/72
Base on the example provided in that github page, the user requests for one video, the server returns ten results. That means your server traffic is multiplied by approximately ten times (you still need ACSII for communication that is eventually sent over PHY). It also requires more processing power; at least, you will need ten times of memory just to store the characters.
Let's see if Anario's server can handle this.
This can be a premium feature.
Note: This is a joke
How would we implement K-anonymity for disliking a Youtube video? I'm interested in implementing this
Disliking can't really be anonymous due to anti-botting mechanisms.
Also if we ever wan't to have a public database then this would ruin any such aspirations.
if #45 would of benn done then someone will open a PR(i might be able too if its in a launge i know)