Pavel Durov: “last year we had two attempts to bribe our devs by US agencies” | Hacker News Reader

Pavel Durov: “last year we had two attempts to bribe our devs by US agencies”

twitter.com

76 pointsgukov30 comments

Signal is on Github. You can audit it if you want.

Telegram is closed source with a proprietary crypto implementation.

I know how I'm placing my bets.

Also Telegram is a much better target for US intelligence, it's the messenger app of choice for ISIS.

There's no way to verify that the Github code is what is distributed by the app stores. To look for backdoors you would have to audit the binaries, which you can also do with Telegram. And you have no visibility into the server code.

It's end-to-end encryption, so the server code doesn't matter. You can compile it yourself if you like. Are you arguing that's no better than how Telegram is distributed?

Only secret chats are end-to-end encrypted, normal chats are not

Maybe not but in the desktop app world there is. It's called deterministic builds, where multiple people can build the same code and come up with the same exact binary. Gitian[1] is a popular one used by cryptocurrencies.

Basically multiple people independently sign the build so even if one is corrupted it would be difficult to corrupt them all (especially since an independent third party can replicate the results).

If there were an easy way to verify the hash of a binary on a mobile device, this could be an option.

[1] https://gitian.org/

Right. And while I generally dismiss conspiracy theories about the protocol itself (and have great respect for moxie), the one thing I find very curious is their staunch refusal to implement anything other than phone numbers as account identifiers.

There's absolutely no reason not to, and yet they refuse to do it.

Usability. As soon as you install Signal you can message all your friends (whether they have Signal or not), because you already have their phone numbers.

This feature is possibly the most important reason that Signal has seen widespread adoption in average-user land.

Signal makes a lot of decisions for you, which enrages nerds but is actually pretty appealing to everyone else.

> As soon as you install Signal you can message all your friends (whether they have Signal or not), because you already have their phone numbers.

I had no idea. I've played with Telegram before but didn't know anyone else that used it. I was looking for new SMS app now that Google Hangouts no longer allows you to consolidate carrier and Google Voice SMSs in the same app - switching to Signal means I get the security portion for free.

You're probably already aware but just to be clear, Signal offers no security related benefit while using ordinary SMS. Only messages sent to another Signal user are encrypted.

Yep, intuitively I saw no way it could, except perhaps offering at-rest encryption of message history on your device.

It gives me one place for encrypted and plain-text SMS though, and that's a very good thing.

Of course, you could always use a trustworthy service that downloads and builds the source directly from Github and signs it with its own key - like F-Droid...

...if Moxie allowed it. Instead you're just supposed to trust him.

I much prefer SilenceIM anyway, which uses the same encryption but tunnels it over SMS. Like Signal used to, before that feature was inexplicably dropped in favor of becoming another internet instant-messenger. Which requires giving root on your phone (!) to the largest ad company in the world (!) to use properly.

Here you go, reproducible builds. https://whispersystems.org/blog/reproducible-android/

Removing encrypted sms messages wasn't inexplicably dropped. It was explained and one of the reasons was terrible security https://whispersystems.org/blog/goodbye-encrypted-sms/

You don't need anything Google on your phone to use Signal. https://signal.org/android/apk/

While I disagree with their rationale, the person you were replying to was displeased with how Signal depends on Google Play Services for push notifications. Last I checked that was a dependency that Moxie felt was essential to providing a good UX.

you don't need Google play services to use signal already for months

I stand corrected on the reproducible builds. That's great. It would be greater if they worked with F-Droid so I could download a verified build from there, instead of making me choose between Play Services or APK sideloading.

I don't buy any of the arguments on that SMS page. In particular, all the hand-wringing about metadata rings hollow when OWS has simply arranged it so that they have access to it all, instead of a federated network of telcos, making themselves a single point of failure. The only argument that rings true is "iOS does not have APIs that allow us to programatically send/receive SMS messages", but I don't feel I should have to suffer for Apple's failings.

I applaud the offering of APK downloads, which Moxie has historically resisted calling them "insecure" (I can see his perspective but when the alternative is giving Google root, I'm afraid we're going to have a difference of opinion on what constitutes "security"). Unfortunately, when using Signal without Play Services, as I understand it battery life is much worse due to the implementation. I appreciate that "push notifications" without a central service is a difficult technical problem to solve efficiently - this is a large part of why I regard the dropping of SMS support, which solved it, as a screwup.

Lastly - if I seem like I'm giving OWS a hard time unfairly, it's because of their repeated insistence on centralised services (both Google's and their own) and resistance to any sort of federation or anything that costs them control or information. I'm glad that the situation is (slowly) improving, but these are indicative not just of technical difficulties, but a deep ideological difference. I am not afraid to say that I am mistrustful of this One True Encryption solution when its authors seem so willing to compromise on fundamental issues.

I am also mistrustful because whenever a "controversial" decision is made, the real reason is always accompanied by copious FUD. The SMS thing is one example. The F-Droid affair is another (Stated reason: F-Droid is insecure. Real reason: F-Droid strips analytics).

The Telegram client source is on Github (https://github.com/peter-iakovlev/Telegram) but it seems to be months out of date and they still don't have the source for the server. Even if both were equal on the OSS front, I'd still trust Signal a lot more, especially considering Telegram's homebrew crypto implementation.

Some context for this tweet: infosec twitter has been going absolutely nuts over Durov's conspiracy theory over Signal's USG funding in the last couple days:

https://twitter.com/matthew_d_green/status/87356403040063897...

I think he meant to say "two unsuccessful attempts". Who knows how many attempts were actually successful. I sometimes wonder what percentage of e.g. Facebook, Google, Microsoft etc are covert NSA/CIA agents. I'd be willing to bet my own money that the number is not zero for any company of any significance.

I think bigger, what percentage of workers in all industries throughout the world are covert government agents of one form or another. Not just US on top of that. I'm willing to bet it's much higher than one would think.

I always imagine the facebook network team is the most diverse team on the planet: They have CIA agents, MI5 agents, Mossad agents, whatever-KGB-is-named-today agents, ... And they all work in perfect harmony.

Pretty weird statement to make. Why are people buying into the drama? The US government funds all sorts of basic research, implementation, and infrastructure work. Modern web would be impossible without government funding. I see no problem with the government funding work on Signal.

Isn't bribing a commercial app developer different from funding basic research?

Yasha Levine has been promoting this view for years now. He doesn't appear to have a complete understanding of the technology he writes about, but still makes these accusations that anything the USA spends money on must be compromised.

See: https://pando.com/2014/07/16/tor-spooks/ another: https://surveillancevalley.com/blog/government-backed-privac... and another: https://pando.com/2015/03/01/internet-privacy-funded-by-spoo...

He brings up some good points, but then jumps to conclusions which don't actually seem reasonable.

I kinda wonder if he will eventually substantiate his claims in book he is working on now: "Surveillance Valley" (planned for release in 2018). Otherwise what he is doing now, with all those grandiose conclusions, seems pretty harmful for actual security and privacy of users.

Does anyone have thoughts on XMPP? If I understand correctly it is a group / private chat protocol similar to email in that there is a protocol that can allow you to communicate to each other. Similar to email there are many companies that offer software that can communicate with any other software using that protocol.

Since it is simply an open protocol if one app gets compromised, users can easily switch to another without having to worry about loosing touch with their contacts.

I'd also be interested to hear an experts thoughts on this. Could this be asked as a question in a thread of its own?

This is standard MO for the feds, so probably not a lie. They do this to everyone.

http://mashable.com/2013/09/11/fbi-microsoft-bitlocker-backd...

Not saying it didn't happen since there's no way to verify/falsify it, but I'd take it with a giant grain of salt. Telegram vs Big Bad USA is a great marketing story for Telegram.

Relevant:

http://securitywatch.pcmag.com/security/319544-what-it-s-lik...

... that we know of ...

Reply on news.ycombinator.com