Hacker News Reader

Hacker News

Comment by tptacek | Hacker News Reader

Full thread

tptacek

Just to get a sense of the track record here, Signal Protocol powers:

* Facebook/WhatsApp

* Google/Allo

* Microsoft/Skype

* Signal

Signal is also the basis for the protocols for Wire (Proteus) and Matrix (Olm).

View on HN

Please note that of the technologies tptacek listed, the only one which can operate in distributed fashion is Matrix.[0]

Wire is attempting to pursue decentralization, but federation is not (yet) in their roadmap.[1]

[0] https://github.com/matrix-org/synapse

[1] https://medium.com/@wireapp/wire-server-code-now-100-open-so...

I don't know why but Signal has always scared me - I think it was when I noticed it sharing my contacts with their server to "find my friends" when I never consented.

Matrix is looking good, but again not P2P only federated. This is why we are trying to do fully P2P end-to-end encryption like with https://hackernoon.com/so-you-want-to-build-a-p2p-twitter-wi... .

They don't.

https://signal.org/blog/private-contact-discovery/

We’ve designed the Signal service to minimize the data we retain about Signal users, so the only information we can produce in response to a request like this is the date and time a user registered with Signal and the last date of a user’s connectivity to the Signal service.

Notably, things we don’t have stored include anything about a user’s contacts (such as the contacts themselves, a hash of the contacts, any other derivative contact information), anything about a user’s groups (such as how many groups a user is in, which groups a user is in, the membership lists of a user’s groups), or any records of who a user has been communicating with.

All message contents are end to end encrypted, so we don’t have that information either.

https://signal.org/bigbrother/

Ok, that's good. But Signal automatically created conversations with people in my address book who I never want to communicate with via Signal. I didn't authorize it, the app did it automatically. I think that is what the grandparent was referring to.

https://support.signal.org/hc/en-us/articles/115005045728-Do...

> my contacts with their server to "find my friends" when I never consented.

I installed Signal a few days ago for the first time and there was definitely a prompt, with an easy way to skip it, before it did the find my friends thing.

The wording also seemed to indicate that only the hashes of the numbers would be uploaded but not sure if thats actually true.

How does Signal know my contact is using Signal?

Signal periodically sends truncated cryptographically hashed phone numbers for contact discovery. Names are never transmitted, and the information is not stored on the servers. The server responds with the contacts that are Signal users and then immediately discards this information. Your phone now knows which of your contacts is a Signal user and notifies you if your contact just started using Signal.

https://support.signal.org/hc/en-us/articles/115005045728-Do...

You can add OMEMO (XMPP) to the list: https://conversations.im/omemo/

Don't forget EFF also gave it high grades. What I also would like to see is meta-data eliminated in chat, without having to bring up a Tor server on your phone. Build it into the client/software, make it invisible.

The EFF's handing out of 'grades' for this stuff is probably something everyone is better off forgetting.

Their grades are based on objective assessment of properties of underlying technologies. They are very useful for most people, who are not going to study cryptography for several years and then manually reading and checking source code of every communications software product out there.

No. They're not useful. They were a disaster, and EFF should stop doing them (I think they have?)

https://news.ycombinator.com/item?id=10525266

Yes they have. Now they have a set of (kind of weird) guides.

Seems like there's a demand for an objective layman friendly list like this (only more accurate). Wonder if that'll ever happen.

I suspect security will remain a privilege of the technical elite or those that can pay the technical elite. I guess maybe that's OK, but a bit concerning at a consumer level where it could be another factor increasing the class gap.

This press release is about security being increased in a product that has massive reach amongst those who are not the technical elite. There doesn't appear to be any extra fee for it as well.

I don't think things are nearly as bleak as the GP poster sees them - my optimistic guess is that e2e encryption will be the default in most messaging systems within 10 years.

On the other hand, this sort of integration is a symbolic baby step of little practical impact or increase in security. Not only are these features not enabled by default, they're so deeply buried in the UI they're hard to find even when you know they're there. It's easier to accidentally send someone an animated gif of tapirs playing poker and smoking cigars on FB Messenger than it is to deliberately start an encrypted conversation. Skype and Allo are not much better.

> No. They're not useful. They were a disaster, and EFF should stop doing them (I think they have?)

> https://news.ycombinator.com/item?id=10525266

Seems like most of the critiqued stuff has changed. By the time i got to EFF recomendations Signal was the winner and use of telegram and cryptocat was discouraged.

"Their grades are based on objective assessment of properties of underlying technologies."

What were the grades for apps/protocols that run on closed, proprietary basebands with, in some cases, DMA ? Or on devices with embedded, parallel computers that can run arbitrary java programs uploaded, at any time, by the carrier, without the users' knowledge[1] ?

All fails, right ? Straight "F"s ?

[1] These are called "SIM chips"

I love this. The more encryption is used, the less journalists talking to their sources, lawyers talking to their clients, etc. will stand out.

Unfortunately most of this is opt-in and not opt-out. Opting in for E2E encryption in a Facebook message really stands out.

Facebook has that? I can't find it.

It's called secret messages or something similar. May only be available in the messenger app...

Correct. It's kinda out of the way unless you know what you're looking for. On iOS:

- Open the app to the "Messages" tab.

- Tap the compose button in the top right corner

- Tap the "Secret" button in the top right corner.

In the mobile messenger app (not lite) , click the 'new conversation' button, and top right there's a switch marked with a lock.

Yes, that is indeed a shame, although I'm glad at least WhatsApp has it by default, which is a lot of messages already.

Do you have an opinion on how Keybase handles their encrypted chat?

I believe Keybase is different because they have so much of their own key infrastructure. I haven't heard anything about them linking up with Signal.

Can't say I am really happy about it. The more the Signal protocol is used the more money is spend on attacking it.

"If you haven't been hacked by the Chinese you got to ask yourself, does the shit you're doing really even matter?"

I prefer for Signal to be attacked and to matter than to be unattacked and to be irrelevant.

Indeed, it's incredibly selfish and naive to depend on security by obscurity here when you can significantly up the bar from basically one step above full-text all-access for every global spy agency... to a serious encryption system.

Regardless, the exposure ship sailed long ago anyway when WhatsApp with their billions of users, including plenty of terrorist groups in the middle east and africa, decided to adopt the platform.

Additionally, breaking crypto systems like this in practice, if possible, pretty much always means targeted attacks. Which is a significant difference from the type of mass surveillance which Skype et al currently allow.

Even Blackberry bragged about building a system for the NSA/CSEC to provide real-time data access to every BBM being sent in the Toronto area during G20. It's very likely that Skype has a similar system and this change (should) make that fundamentally infeasible.

s/Chinese/US agencies/

But being battle-tested early on is the only way to ensure there are no bugs later on when everybody is using it.

That's the opposite of how you should feel. The more money spent attacking Signal, the stronger secure messaging gets. The best thing that could possibly happen for messaging security is for someone to discover a flaw in Signal. Unlike a lot of messengers, where a flaw is unlikely to teach us something other than "people should use Signal Protocol instead of terrible ad hoc protocols", a Signal flaw advances the state of the art.

That's a good thing. Well-studied protocols are safer protocols, and significant money can pay for the best expertise in the world.

The only way you can know if a protocol is secure is if you have a lot of eyes on it trying to break it nine ways from sunday. This is nothing but a good thing.

Moxie Marlinspike is one of my tech heros.

Speaking of track records,

https://www.theguardian.com/world/2013/jun/06/us-tech-giants...

Facebook

Google

Microsoft

They cut off 1 letter off the url by replacing "june" with "jun"

I think you'd either want the full month name or all the months to be abbreviated to the same number of characters. You can't do 2 characters because of June and July, but three works. 4 would be sort of weird.

Reply on news.ycombinator.com