5 ms·
It should be optionally public. Maybe I want to grab a list of those I also want to block. Or negatively weight. Conversely, the follow list can add positive
by echelon 3y ago
It should be optionally public.
Maybe I want to grab a list of those I also want to block. Or negatively weight.
Conversely, the follow list can add positive weight until I say otherwise.
Distributed email spam and trust lists.
There's the potential for creating a strong filter bubble, but if the user can control their own system's behavior it still seems like a win.
- phoe-krk 3y agoYes. For all the cases you've mentioned, I might want to share my blocklist - but it should be an option rather than the default. I may want to share it only with select individuals rather than everyone who might want to use it to scrape it, archive it, monetize it, and stir online drama on my behalf AND on behalf of whomever I've blocked. Seriously, it's worse than on Twitter. Bluesky slowly starts looking like "let's redo ActivityPub but so we can make money off it".
- klardotsh 3y agoThat was basically always the goal. If it wasn't, the BS devs would have just made a Mastodon/Misskey/etc competitor on AP directly.
- preseinger 3y agoyour block list has to be enforced at some point in the chain enforcement means it needs to be read-able by the relevant software "optionally" public means that software that can't see the block list can't enforce it and that means the only way to enforce the block list (reliably) is in your client (which presumably always had read permissions) this is in no way controversial
- kelseyfrog 3y agoThat would be true if holomorphic encryption didn't exist.
- preseinger 3y agohomomorphic encryption theoretically allows a block list to be query-able without being inspect-able, true i'm unaware of any practical implementation that would address the given use case if you know otherwise, please point me to it!
- anamexis 3y agoIt’s not homomorphic encryption, but publishing a list of salted hashes seems like it would fit those criteria.
- ryan-c 3y agoThat would provide very limited benefit due to the search space being so small, and would have serious scaling issues with large block lists.
- anamexis 3y agoIf you used a per-user list-wide salt, there wouldn't be any scaling issues. And the requirements themselves dictate that there's a small search space.
- ryan-c 3y agoIf you used a per-user-list salt, then it would be about as useful as obfuscating phone numbers by hashing them. The list must then be enumerable in N*x time, where x is the time to hash one user, and N is the cardinality of the search space. One second per legitimate hash check? You can enumerate the list in a few hours by doing a parallel search...
- anamexis 3y agoRight, but isn't this inherent in the hypothetical requirements here? Being able to query individual usernames for block status?