yeah, but on twitter you're probably a nobody, the staff have no incentive to read your dms. on mastodon, you're at least a friend-of-a-friend of the operator unless you're on a huge instance.

Others have pointed out that this is fallacious -- even on big services sometimes employees are creepy stalkers, sometimes they're malicious actors of other sorts -- but even if you ignore that on twitter you are a target of advertising and if you think they aren't slurping up all the data they can about you and storing it in a database somewhere, you're fooling yourself.

Anyways, down this logic path is an internet where we somehow put all our trust in megacorporations and absolutely none in our fellow human beings and I dunno about you but one of those sounds a lot more dystopian to me than the other.

twitter has hundreds of millions of users and 7,000 (less now, unfortunately) employees, not all of which have production access. i don't know anyone at twitter, so them accessing my data would need to be a random choice from the whole userbase. they'd need to evade internal checks (which, however weak they are, are infinitely stronger than the average fediverse instance). could it happen? could i be eaten by an escaped zoo leopard during my morning walk?

having my data "read" by a non-sentient advertising model, while irritating, is nowhere near the same as having it read by a human being. that is a false equivalence.

i made no argument about the absolute merit of twitter vs the fediverse. this is simply a downside of small communities that people excitedly migrating to the fediverse with little understanding of how it works or experience with its predecessors will soon run into—twitter is neutral ground, on mastodon you might be arguing with the instance admin's friend and find yourself retaliated against. anyone who used a forum or IRC in the 90's/00's knows what i'm talking about.

Well, now you've posted about it here, so maybe someone will get curious?

As for:

> they'd need to evade internal checks (which, however weak they are, are infinitely stronger than the average fediverse instance).

Did you watch Mudge's testemony? Even if Musk wants to, it'll take a total re-org to get a semblance of security at Twitter, by the sound of it...

Mudge Twitter whistleblower testimony [video]

https://news.ycombinator.com/item?id=32824504

https://www.judiciary.senate.gov/meetings/data-security-at-r...

> having my data "read" by a non-sentient advertising model, while irritating, is nowhere near the same as having it read by a human being. that is a false equivalence.

I agree they're not equivalent.

One is a small harm guaranteed to be perpetrated every day against every single person using Twitter, numbering in the hundreds of millions of people, largely without those people knowing or consenting to it.

The other is a large harm sometimes inflicted on a small number of people, and when it does happen is isolated to a small community where people can find out about it and act on that knowledge how they see fit.

I've been around a long time. I've seen power abused on irc and forums and even small social networks. To me, the idea that we need supposedly benevolent megacorporations to keep us from doing harm to each other is a repulsive idea, far and away above "my instance admin might read something I wrote on their server."

If we've forgotten how to exist in community with each other, we should relearn that skill.

> Admins of <service> can read data on <service>" is just a tautology.

Huh? This is certainly not true for Signal and Matrix, heck even whatsapp and telegram sounds better than some random instance operator.

That said, truly private messages aren't always necessary, as long as the platform is crystal clear about this.

Signal is not a "hosted website," which is more the context we're talking about here. But even on those services, yes, there are ways that the owners of the service could tap or impersonate you through exploiting their own key exchange service. You are trusting that they won't do that.

This might be less true for matrix, since you could in theory be using an open source client where you have somehow guaranteed it will alert you to an attempt to add an unwanted device key to your e2e chat, but on signal you're running a binary you didn't compile against a service you can't see.

I don't think you shouldn't trust them. But you are doing so to some extent.

In the case of signal, they would have to forge the SGX enclave signature (by an intel held key) or release a client that didn’t validate that sig. Definitely possible but if I had an SGX bypass I’d want to use it on something known to be high value, and releasing a non-verifying client would at least be noticeable on android and desktop.

I think the latter (manipulating the client) is far more likely than the former, and I think it would also be pretty difficult to detect in practice. But the point is less "I think they will do this" than "there is still an element of trust here, even if it is a much harder hoop to jump through." I don't think any situation where signal does anything like this is likely.

You don't need to release a non-verifying client. Just one that generates a key which is known to the other side. What about existing clients? "Your identity in the database became corrupted and can't be recovered. Would you like to generate a new key and continue using the service?" or just release a version which is both verifying and lying to you about which key has been verified... or a low effort "hey, new phone, key changed".

Sure. My point was more that - at least on android - you can compare decompiled binaries with the open source implementation, presumably making it reasonably likely that it’d be noticed and reported on.

Unless they decline to publish their source code for over a year, again. https://github.com/signalapp/Signal-Android/issues/11101

> private messages aren't always necessary, as long as the platform is crystal clear about this

When sending a DM on Mastodon, there is literally a message that pops up saying: "Posts on Mastodon are not end-to-end encrypted. Do not share any sensitive information over Mastodon."

I don't know much about Mastodon, but I know that it's main selling point is that it's decentralized, and it's pretty easy to assume that decentralized means there isn't anybody with special privileges who can read private messages. The same way decentralized finance (blockchain) means there isn't anybody with special privileges who can take your money.

And I would certainly assume that in 2022, any service would be built using encryption for the parts that are private, and aren't DM's private? Why would admins be able to read them? Is there a justification for that?

> it's pretty easy to assume that decentralized means there aren't admins who can read private messages.

I'm not sure why you would assume that? It's not something you run on your computer, it's still a website (or set of websites). Admins of your email can also read your email, if they want, and even with gmail in the mix it's probably one of the most "federated" systems ever built.

> I would certainly assume that in 2022 it would built using encryption for the parts that are private, and aren't DM's private? Why would admins be able to read them? Is there a justification for that?

They could potentially be encrypted at rest, in the database, but that doesn't really help much. The owner of the site would have the keys to decrypt them, and on smaller sites it's very unlikely that there'd be any real chain of custody involved.

If you've ever sent a DM on a forum did you think that was encrypted? It wasn't. Or twitter or facebook for that matter. It's not really practical for any data stored on a central server to be encrypted in a way that irrevocably prevents the owner of the service from accessing it.

> I'm not sure why you would assume that?... If you've ever sent a DM on a forum did you think that was encrypted?

The whole assumption here is that Mastodon is supposed to be better than those, right? Or else why are we switching? Twitter is centralized and can read all your stuff and censor it too. So isn't the point that Mastodon isn't and can't do those bad things?

We expect WhatsApp and iMessage to provide E2EE. Similarly open-source Signal and Telegram are encrypted. So why wouldn't you assume another high-profile open source project isn't adopting those same best practices for the private-messages part of it?

> Mastodon is supposed to be better than those, right?

Here are the ways mastodon is better than twitter:

- It can't be bought by a billionaire man baby

- It can't be coerced into hosting awful people because they drive revenue

- It doesn't require advertising in order to continue existing

- Because of that I'm not being endlessly datamined by adtech every moment I'm using it.

- It can't die because one website goes down, and everyone on it doesn't experience awful performance just because one instance is falling over.

- If I don't like the admins of the instance I'm on, I can move to another instance and bring much of my data with me without having to exfiltrate it with tools that violate the TOS.

- I can use whatever clients I like with it and I never have to worry about the company deciding it doesn't like third party apps and killing them slowly with api rate limits.

There are also a lot of ways it's worse than twitter, though they're mostly along the lines of "some of my friends aren't on it". Things don't have to be "better in every way" than other things to be "better for me (or you)". There are always tradeoffs.

Re. WhatsApp, Signal, Telegram and iMessage are all apps you run on your phone. And if you can read the messages on them from a website (as you can if you turn on a feature for imessage), then the admins of the service also have access to your messages.

Again, we're talking about a website here.

> Re. WhatsApp, Signal, Telegram and iMessage are all apps you run on your phone. And if you can read the messages on them from a website (as you can if you turn on a feature for imessage), then the admins of the service also have access to your messages.

Not true. Web clients for Matrix are open source, and you can self host them if you are afraid of the default host trying to inject spyware to the page

I mean, the website admin in that case can still access your messages. It's just that the admin is you.

That's true. Though I guess that when my parent comment referred to "admin", the were referring to the admin for the homeserver (the one routing the messages), which is different from the one hosting of the web client.

This is why services like Matrix and Signal open source their client. Because for security minded people, securing the client is much easier than securing the server.

Decentralisation is orthogonal to privacy. Your messages can be end-to-end encrypted on a centralised system (e.g. Signal) and not on a decentralises system (apparently here Mastoson).

Decentralisation is about control: the Signal admins could shut down the server and kill the service. For Mastoson, if you stop one server, the others still work.

That's wrong. If it is end-to-end encrypted, then the server admins still cannot read it.

Use e2ee messengers (like Signal) for DMs, use Mastodon (or whatever you want) for public posts.

Mastodon is "service", not "website". You can use dedicated client software.