3 ms·
I don't trust any "private" settings in any public social networks, for anything remotely serious. Anything I publish on any social network I'd be also comfort
by nine_k 4y ago
I don't trust any "private" settings in any public social networks, for anything remotely serious.
Anything I publish on any social network I'd be also comfortable seeing posted on every wall in foot-tall letters, with my real name attached to it. Were I not be comfortable with that, I won't post it.
I consider all "private" or "limited" posting provisions on all social networks as paper walls at best, one small devops or SWE mistake from being torn and becoming visible to public. Maybe I'm overshooting a bit, but my past experience shows that it's not much of an exaggeration. So I never use any "private" features and post everything I ever post as openly as possible.
Social media is a colander. Expect anything and everything to leak. Embrace being public.
If you want something with more real privacy, as in not likely to leak if all parties involved take reasonable care, use something that was built with privacy in mind, like, IDK, Signal, or maybe a private Telegram chat, or a private Mastodon instance only accessible through a VPN which you personally and competently operate. Of course it's unlikely to hide you from state-level actors, but at least it's more likely to protect you from simple operator blunders.
- _cenw 4y ago> Anything I publish on any social network I'd be also comfortable seeing posted on every wall in foot-tall letters, with my real name attached to it. You misunderstand. It's layered defense / minimizing exposure. Nobody is suggesting it can't be done, but it not going to be normalized in the current Fediverse, and most potential attackers do not care enough to build throw own search engine. The main reason is quite clearly stated in the article, and I'm not sure if you didn't see it or just don't care because you're not affected. > And, if you’re vulnerable, attack you, shame you, doxx you, SWAT you, try to kill you. A lot of people that are being targeted by the likes of KF or Libs of Tik Tok find nice comfortable corners on the Fedivserse because they're not immediately discoverable by someone typing their name into Google.
- nine_k 4y agoSorry, I do understand. I think that the particular fortress of highly private communication on a network designed for publishing and dissemination cannot be defended efficiently. I'd be glad to be proven wrong, but my last 25 years of experience with social networks of all sorts, from early IRC and web chats to what we have today tells me to keep my current security stance. Anything on a social network has a very high chance of a public exposure, one way or another. For the defense in depth / layered defense to work, the initial assumptions should be different. They should be like Signal's, not like Mastodon's. The whole approach should be paranoid about not letting things seep through by any chance at all. I don't think that such an approach would be seen as productive by most users and developers. So, I view various access control settings in networks like Mastodon as a convenience filter, not as a security mechanism. > not immediately discoverable This is actually both important and achievable! If you never post your public identity, and things which easily link you to your public identity, it's pretty hard to doxx you and harass you IRL, even if everything you publish were publicly accessible. If you don't advertise your presence where everybody is looking, you have a good chance of not being harassed online, because the harassers won't try hard enough to infiltrate remote corners of the internet. A reliable pseudonymous identity is a great mechanism, if you maintain reasonable opsec. I just won't expect a great level of insulation of your content from fellow pseudonymous users, and occasionally even from general public.
- _cenw 4y agoUsually the kinds of people who can direct harassment at you go for the lowest hanging fruit, so any amount of obfuscation or barrier will deter most of them, unless you're a very high profile target (i.e. you're Keffals). The entire model doesn't even rely on being entirely undiscoverable - the point is that you don't need to step back from public life because a bunch of weirdos like stalking you, it can be as easy as having a publicly visible profile with a lock on it. This prevents following, but unlike Twitter you're not entirely excluded from making public posts that anyone can interact with. You can just do a baseline of filtering when accepting followers and deciding which posts to make public or private. You'd be surprised how many of these stalkers and harassers can't make a profile that passes a basic "vibe check" - they either try too hard at putting together a profile or they don't bother at all. Fedisearch and related projects are frowned upon because they would make it much more trivial to find people who are not yet targets based on a soundbite and to go through someone's "permanent record" (most implementations do not backfill posts made before the user was initially followed from an account on the same instance).
- HenriTEL 4y agoYes. I would even argue that Privacy expectations should be much weaker than for the big centralized platforms. Mastodon's instances admins have blanket read access to every post and message, even the "private ones", yet they are not under scrutiny of RGPD regulators like for example Twitter. Are you sure those admins take good security measures, like storing backups encrypted, keeping their tech stack minimal and up to date, using safe configs? Do you trust them for not reading and sharing your conversations? Regarding the Fediverse search, it's even worse as you're supposed to trust the rest of the world for not abusing it and indexing the entire network. Well, that's not how things works. With this kind of large scale systems you should know that if there is the possibility of an event, that event will happen very soon if not already. Either you impose restrictions or you accept the tradeoffs.