4 ms·
> purported security failures > people are looking to block it instead of silently explore its flaws This is not a good heuristic for determining if security
by dsacco 9y ago
> purported security failures
> people are looking to block it instead of silently explore its flaws
This is not a good heuristic for determining if security failures are "purported."
In point of fact, Telegram is widely lampooned by every self-respecting professional cryptographer who has written about it. There's nothing "purported" about Telegram's security failures - they are empirically demonstrable, and have been exposed through multiple cryptanalytic reviews.[1][2]
Frankly, I don't think I've ever seen anyone defend Telegram here on HN who actually has professional crypto experience (whether academia or industry), or any other similar proxy for credibility in the field. The popular contention is that (poor, harmless) Telegram is plagued by a persistent astroturfing campaign perpetrated by the likes of Moxie Marlinspike and Thomas Ptacek in order to elevate Signal's status. That's:
1) not true, in my opinion (though to be fair at least one of those people is obviously biased); and
2) irrelevant, because we have the benefit of empirical rigor to instruct our opinions of secure messaging systems. We don't need to rely on infosec ideologues on HN or Twitter.
Telegram is very much like climate change. There is a widespread consensus among the informed (read: academic and professional cryptographers) that Telegram's security failures exist, and that these failures are empirically demonstrable. At the same time, there is a controversy led almost entirely by the uninformed (read: non-cryptographers) that denies Telegram's security failures and undermines attempts at demonstrating them through accusations of shilling or misdirection.
To put it very succinctly: there are no valid arguments that Telegram has an optimal security model from the perspective of cryptanalysis and cryptographic design best practices.
____________________________________________________________
1. https://www.alexrad.me/discourse/a-264-attack-on-telegram-and-why-a-super-villain-doesnt-need-it-to-read-your-telegram-chats.html https://www.alexrad.me/discourse/a-264-attack-on-telegram-an...
2. https://cs.au.dk/~jakjak/master-thesis.pdf https://cs.au.dk/~jakjak/master-thesis.pdf
- out_of_protocol 9y agoOk, valid argument: telegram servers DO receive and store plaintext version of every conversation done (except private chats), while, as a reference, whatsapp don't do this, applying end-to-end encryption by default. So, basically, hack telegram servers -> you have all the convos. Bribe some employers, locate and grab their servers -> see above. Also, how would you know they're not connected with some 3-letter agency already?
- reitanqild 9y agoFWIW - I have no way of verifying this - Telegram says they try to work around this by (IIRC) : - Using some kind of crypto with multiple keys - These keys are located in multiple independent datacenters in independent jurisdictions According to what I read this was supposed prevent leaking of user data just by bribing/coercing/etc or suing any Telegram employee or datacenter provider. As I've mentioned before I don't use Telegram for security, I use it because 1. Facebook failed its Whatsapp acquisition so badly 2. It is way more user friendly Also keep in mind that when I left WhatsApp for Telegram WhatsApp still didn't have e2e.
- Aloha 9y agoThere is a difference between practical security and theoretical security. Telegram has some well known theoretical security flaws - its been well documented as such. But from a practical point its more secure than most of what it replaced - SMS and Facebook Messenger. I use telegram as a cross platform iMessage - if iMessage was cross platform there wouldnt be much room for for telegram - if WhatsApp had a desktop client, same deal - if Signal was more user friendly (it also launched without a desktop client) - same deal. In short, Telegram is good enough for the purposes that people are using it for - once MProto has been breached - then I'll be concerned - but right not all the attacks are theoretical, and not practical.
- replythrow 9y agoAre people damn reading what they link? There's not a single actual practical vulnerability in the second paper. http://telegra.ph/mtproto-security-01-17 http://telegra.ph/mtproto-security-01-17 And the first one is outdated.