Hacker News Reader

Comment by ENTP | Hacker News Reader

I, for one, shudder at the thought of Facebook getting my banking messages. Telegram has been a great alternative although it's not perfect.

View on HN

I'd be much more nervous about Telegram getting that data.

It's a toss up:

- WhatsApp uses reputable cryptography from reputable sources, but combines it with a UI so much designed for ease of use that it throws out the possibility of real security

- Telegram uses extremely questionable cryptography to begin with from people who probably shouldn't be designing cryptosystems

If you want real security, go use Signal or better, OTR where socialist millionaire's is still an option for key verification.

> OTR where socialist millionaire's is still an option for key verification.

Can you explain the socialist millionaire comment? I don't get it.

https://en.wikipedia.org/wiki/Socialist_millionaires

Other people already linked the wikipedia article, but basically the thing it allows for is verification of a key without trust of it and provides effective proof you're not being MITM'd.

I can ask a question or a series of questions to my friend and we can use zero knowledge proofs to determine we're talking about the same answer to verify his key without disclosing anything about that answer.

Also WhatsApp is closed-source, but Telegram's client is open-source.

Not everything with Telegram is open source like Signal. I still use Telegram because it's designed better and updated much more frequently.

I'm using both, too. :) But unfortunately I don't use Chrome, so I can't use Signal on my PC yet.

There are some FF-related pull requests on GitHub, but it seems that guy hasn't signed the contributor agreement. Maybe we'll get an FF-version soon.

Why is a contributor agreement needed? Can you post the link to the PR?

Telegram is actually not open, they just throw some sources to github or as zip files at their's langing page. That's why i quit Telegram and built open source "Telegram" - https://actor.im and replaced spagetti-encryption with enhanced version of Signal's one.

Considering it is pretty much the gold standard for crypto, how did you enhance it?

We added one extra level based on Russian encryption. So if Curve25519 or AES encryption will be broken there are still be russian level. To break our crypto Russia and US need to cooperate that is impossible.

I've read up a while ago on the questionable cryptography (not an expert in that at all myself) but couldn't find a lot of actual substance. Certainly with claims [0] [1] being that Telegram offers nice sums for breaking it, which gives me as a consumer the idea that someone who actually knows this stuff would love to hit that down, if it was something in my area of expertise offering that I'd love to try and hit it.

I get that it might all be some kind of cheat, but an actual cryptography researcher publishing that it is would, obviously, get a lot of bad karma going in the HN circles (for example) about these contests.

[0] https://telegram.org/blog/winter-contest-ends

[1] https://telegram.org/blog/cryptocontest-ends

https://eprint.iacr.org/2015/1177.pdf

What have you read about the questionable cryptography in Telegram that you found unpersuasive? That's a better starting point, because the formulation you use here is often used on message boards to bait people into doing research work that is then dismissed.

Your comment includes an assertion that there aren't substantial complaints about Telegram's security (there very much are, BTW). I think the onus is on you to defend that argument.

You're absolutely right if my goal was to get people to post these papers, which reading back can easily be how it is interpreted.

My point was more that, as a 'consumer' of this product without any background knowledge, their contests seem extremely convincing to the point of being slightly ridiculous. A 300,000 dollar bounty, while participants get quite some more access than they would normally get while attacking this system, gives me that feeling of it being actually quite secure, besides some possible attacks. If they're possible, if there are complaints, why can't they be made real and substantiated?

My first post is indeed not delivering the message I intended, hope this one shows it better.

This is one of the rare instances where me and Bruce Schneier agree unequivocally, so maybe you can just take Schneier's word for it --- remembering, when you do, that this happens so often in security that Schneier was moved to write a little article about it!

https://www.schneier.com/crypto-gram/archives/1998/1215.html...

Thanks a lot, that was very insightful and exactly the kind of insight I was looking for.

I think for me as an outsider it's really easy to fall for that trap, where certainly point 3 in his article was a realization that I would look at it the same way, perhaps spend a bit of time in my evenings on it but never spend, say, 3 months of research on the off chance that they might pay and that I might be the first winner/best winner, compared to a good (hourly) pay.

Doesn't 'real security' quickly escalate to not having a cell phone at all?

The best opsec advice for normal people seems to be not to op.

I wasn't advocating Telegram as an alternative transport for financial data. More that I'd rather use telegram for im than anything facebook owns.

Which reminds me of the announcement while ago about Whisper Systems is helping WhatsApp implementing E2E [0]. I don't think I've heard ever since about it, and not even sure if that functionality is built into WhatsApp now...

[0]: https://whispersystems.org/blog/whatsapp/

Reply on news.ycombinator.com