There’s no such social graph to speak of. Signal does not know who is speaking to whom.

You have no source for that other than Signal's "just trust us" claims.

There are plenty of chat services that aren't centralized and hosted in the USA.

You don’t have to trust anybody when you run your own server,

You have to trust the people that wrote the code.

or you use a server that doesn’t collect information it has no business collecting.

Again, you're trusting the authors of the code.

Which is fine, but it's a choice to trust them.

But again, you either read the source to confirm there's nothing nefarious, or… you trust the programmers.

Which is not a problem, but it is a choice to trust. All I'm pointing out. :)

It's really the opposite, considering its a centralized server hosted in the USA. Signal's "Trust me bro" is not an adequate security model.

Sure… and my point is that you have to trust those services that aren't hosted in the USA. It's a choice you have to make. I'm not judging either way, just pointing out because what I responded to in the comment to which I replied was:

The problem is that you just have to trust them

Which is true of open source unless you read the code and can verify nothing nefarious exists; which is true if you use a service in a country you trust; which is true no matter what you're doing.

Not all entities are deserving of the same level of trust - some are more trustworthy than others - but you are still making a decision to trust someone unless you write the code yourself or verify the code yourself.^[And had the capability and time to do so]

So what? Her app is also probably selling data too, and it it isn’t it will be sold any day now

You have to trust the people that wrote the code.

There's a big difference between having confidence in open source code that has been audited by many people, and knowing for a fact that the service collects specific information. In the former case, you can never be absolutely sure that the code is not malicious so there is always a risk, but in the latter case you know for a fact that the service is collecting inappropriate information and you have to trust that people operating the service are not using it in adversarial ways. These two scenarios are in no way equivalent.

Which is fine, but it’s a choice to trust them.

It's a choice to trust the entire open source community around the project and all the security researchers who have been looking at the code.

Frankly, I have trouble believing that you don't understand the difference here and are making your argument in good faith.

I used it for my family chat, which I naively thought was private. I moved the fam to signal last year

And the client, too.

Signal is good because it is easy to get others on board because of its simplicity, while having good encryption and security. However, SimpleX is just a better option. No phone number required, the same encryption, you can make a new profile whenever you want and you can have however many you want at a time. The groups can hold thousands of people. You can have it automatically make a new anonymous profile for you for each contact and group so that you can't be tracked across them apart from your writing style. There are no usernames, you add each other through links, which can be temporary or permanent, and you can add or remove a link whenever you want.

Sorry for that big wall of text, I just spat out all the things SimpleX is better for. I think that it's platform is the future.

All these apps owned by corporations are just black boxes where you send information and nobody knows for certain what they do with it.

Best case, they parse it, cross it with other data and make it profitable (for them, not for you).

Worst case... Who knows...

I was using it because the API was easy and it didn't have a problem sending images from my back door camera to my cellphone.

At some point, it got REALLY slow at sending, so I moved to NTFY

Well yeah everything is a choice when trust is the matter, but there is a difference between choosing a community project that can be audited by different transparent parties and choosing a private company on their own servers (even on source available projects)

When you build a backdoor into your "encrypted messenger" its just a surveillance app

I don't know man, those benches are nice for a variety of positions. Plus, you are already naked..

Frankly, I have trouble believing that you don’t understand the difference here and are making your argument in good faith.

Let's back up to what I replied to in the first place:

You don’t have to trust anybody

I even took the time to quote that, because it's important.

Of course there are different levels of trust. But what you said is flatly wrong and misinformation, if you want to get technical about it. Arguing in bad faith? I beg your fucking pardon, friend.

Just becuase it's less likely to find nefarious code in open source doesn't mean it doesn't exist. There ahve been multiple cases of it found in open source code. Blindly trusting something because it's open source or you host it on your own server is a very very false sense of security, especially in the context of the larger discussion, which came about in regard to what information is exposed by certain messaging clients.

It's also a matter of the importance of what you're doing.

I wrote a little CRUD app a while back to track me giving my cat medication. I sanitized inputs, but I left it open without a login on my server, just an obscure URL that didn't get published anywhere. All you could do was click a button to indicate the cat had been medicated, or another button to delete the latest entry. That was plenty of security for that. If I was writing a banking app, I'd use a bit more.

So yes, in the same way as that, hosting something you use to chat with friends about whatever is one thing; trying to communicate secretly from a country where your comms might lead to being put to death is quite another. And in the latter case, it's important to know that no matter what you use, unless you wrote it or read all the source code, you are trusting others with your life. Perhaps you feel comfortable doing that, but you should be aware of it.

So no, this is not a discussion in bad faith at all, it is valuable on multiple levels.

Precisely.

And it's worth repeating here - the level of trust needed is affected by the nature of what you might lose if that trust is broken. For non-important things, trusting a third-party company is probably fine. If you're in a country and being found out might mean you get put to death, though, the stakes are a bit higher.

We as a community should, in my opinion, pick a “good enough” solution for the majority of the people we interact with.

I'd probably suggest Deltachat. It's decentralized and has always on encryption, but is so incredibly simple and easy to onboard and use, and doesn't require a phone number or even an email. It also works on all platforms with a single app.

Why did they do it to themselves to name it after the herpes virus tho?

The only people who know this are people operating the server.

We all know this, for reasons I've already stated.

See the link I provided above.

Your link is broken.

Yup, that's precisely what it's a filter for.

100M people is not a filter...

Trust me bro is not a viable model

No one said anything about that? That is not the model.

The reality of the situation is that Signal asks users for information it has no business collecting

The business is connecting users. It's one of the reasons it is the most viable private and secure chat platform. It's why I have a dozen connections on Signal and literally 0 on every other platform. Because you actually know who's using it. You can have the most private and secure messaging system in the world but if you can't use it to actually chat with anyone, then what good is it?

I agree, metadata collection and the obligatory matrix.org server are the biggest problems for privacy, but I don't think you made that clear in your comment.

Accept defects != ignore

Why are they sitting in a sauna?

Just trust me on this one, 80C+ room just isn't the place for that sort of exercise. And it's not like your heart rate isn't up already lol. It's a terrible experience

Session is a security downgrade. It doesnt support forward secrecy which is hella important.

X

What are the big three, and what is the bar it's jumping over to be consider big?

People keep finding significant vulnerabilities in its cryptography and the Matrix team tries to deflect or create strawmans for why it isnt actually a vuln. Soatok found a vulnerability in 2024 by just browsing the source code for tiny bit of time, and again just two weeks ago after looking for a couple hours. In both cases, Matrix then responded to his vuln report with hostility, saying it wasnt actually a vulnerability. He is sitting on another vulnerability.

Having a cleartext mode is a security downgrade and no secure messenger should support cleartext. It only barely got functional forward secrecy recently. VoIP in most Matrix clients (and servers) still use Jitsi backend which isn't E2EE, even with the release of the newer (secure) Element call protocol. Matrix leaks tons of metadata, such as usernames, room names, emoji reactions, generate URL embedded previews. Rooms arent encrypted by default. It is also a UX nightmare and often times you cant decrypt your messages.

Matrix is not secure. You'd be better off with XMPP and OMEMO which has its own problems and isn't secure either. Sill better than Matrix.

Yeah with a lot of research they may be able to figure out that semi-anoymous account messages semi-anoymous account at this time. But have no idea what was sent.

Which is true of open source unless you read the code and can verify nothing nefarious exists

Not at all. Not everyone needs to audit open source, only a few interested experts do. Most importantly, auditing is possible because its out in the open.

The just trust me model of signal means its impossible to audit, unless they give us their centralized database and server code.

Session actually does implement a form of forward secrecy through the Session Protocol. https://getsession.org/blog/session-protocol-v2

My original comment that you replied to was explaining the defects. People are free to decide whether they want to accept them or not. Your comment is saying that it's harmful to discuss these defects which implies that we should just ignore them.

A messenger app Musk touted as better than Signal some time ago

It's a turd of an App, it's not even close to ready for prime time.

No, we don't all know this. What we actually know that people like you say this and expect the rest of us to trust you blindly, which is itself concerning.

Your link is broken.

Your browser plugins are broken, the link is fine. That said, here's non archived version https://www.washingtonpost.com/technology/2021/06/15/faq-data-subpoena-investigation/

100M people is not a filter…

Given world population and modern data analysis capabilities it absolutely is.

No one said anything about that? That is not the model.

That's literally the model. Signal asks you for your phone number when you register, what happens with that information after that is only known to people operating the server. Let me know what part of that you're still struggling to understand.

The business is connecting users. It’s one of the reasons it is the most viable private and secure chat platform. It’s why I have a dozen connections on Signal and literally 0 on every other platform. Because you actually know who’s using it.

That word salad has fuck all to do with the point I made, which once again, is that you have to trust people who operate the server in how they handle this information.

You can have the most private and secure messaging system in the world but if you can’t use it to actually chat with anyone, then what good is it?

Ah yes, because there's absolutely no conceivable way to verify whom you're connecting with aside from sharing your phone number with an American company. You couldn't possibly use any out of band channel to verify who the person you're communicating with is.

good for you

I even took the time to quote that, because it’s important.

What's important is that you're quoting me out of context, and that makes all the difference. The actual statement you're replying to is:

You don’t have to trust anybody when you run your own server, or you use a server that doesn’t collect information it has no business collecting.

The fact that you proceed to quote me out of context and then accuse me of being wrong shows that you lack even a modicum of intellectual integrity. Then you proceed to make a straw man arguing against something I never claimed.

Just becuase it’s less likely to find nefarious code in open source doesn’t mean it doesn’t exist.

So yes, this is very clearly a discussion in bad faith, where you're arguing against a straw man while ignoring what I actually wrote. It's especially incredible since I even followed up with a more detailed explanation which you just ignored:

There’s a big difference between having confidence in open source code that has been audited by many people, and knowing for a fact that the service collects specific information. In the former case, you can never be absolutely sure that the code is not malicious so there is always a risk, but in the latter case you know for a fact that the service is collecting inappropriate information and you have to trust that people operating the service are not using it in adversarial ways. These two scenarios are in no way equivalent.

Do better.

Build sauna, host interview in sauna, sauna is tax writeoff

From my perspective, I'd rather have an enemy of country to have my personal data. It's far harder to use for them than for my country.

Because the most common transmission vector is via family members? (HSV 1)

E.g. Your crazy uncle or news max loving grandparents

It seems that forward secrecy is still in development from the blog you showed.

I still wouldnt use session for the reasons stated in this Soatok's (a cryptographer) blogs. Even if they fix(ed) these problems, I have no trust for their security implementations. Why use session instead of something like Briar?

https://soatok.blog/2025/01/14/dont-use-session-signal-fork/
https://soatok.blog/2025/01/20/session-round-2/

most importantly it has a severe network effect problem. you will not find anyone to message on it

how much data do the FBI get from telegram? do you think the russian owner (who got arrested in france for refusing to make changes demanded of him by glowing authorities) is very likely to give any of your info over to american 3 letter agencies?

I'm not advocating for using Session specifically, I just listed it as a viable alternative to Signal. Given that it's forked from Signal presumably it's an easier switch for people who like the general mechanics of Signal and its encryption system.

Understood.

It's not.

Things like SMS also are used for a lot of illegal activity by people who don't know better. Commonly used doesn't mean much.

Lying is worse and you already said you don't trust it at all.