- cross-posted to:
- tech@pawb.social
- cross-posted to:
- tech@pawb.social
This blog post, and some of its comments are pretty interesting and concerning at the same time. Not really sure if in the end that means that nothing other than centralized controlled messaging can be as cryptography safe.
Any comments?
Looking in the spec design document instead of digging through the source code is normally enough research in other projects where the spec design document is properly filled out. It’s a mistake on OMEMO’s part that the spec design document didn’t include the truncation step in 0.4.0, and this mistake was fixed in their 0.7.0 version. Either way, as I said this is a positive outcome because now we have clarification.
I briefly recounted the points made in the article and I think this was the only one that was against OMEMO directly. Soatok made another post days earlier about why nothing on the market is currently better than Signal, and it makes sense that the other 3 points are still being leveraged against XMPP+OMEMO as they exist in reality, and not against OMEMO alone. It doesn’t matter that OMEMO 0.7.0 is sufficient if nothing is using it, and the various implementations have their own issues. If you were to want to use XMPP+OMEMO today, you’re likely using Gajim, Dino, or Conversations, or someone you’re talking to is. These are still on 0.3, and this is a point that again is important to bring up and potentially solve. If no one is talking about the problem, it will not get solved.
RE quantum encryption, we know what quantum computers are capable of. You’ve suddenly turned into a quantum expert so you must now know that you only really need to protect your asymmetric encryption with PQ, and you would also know that you can combine PQ and traditional algorithms together in cases where you don’t want to degrade existing security if there are flaws in the newer PQ algorithm. This would be the “serious” response, and “serious” software like Signal agrees.
I was in the specs before as well, just not as clearly spelled out.
As for the other reasons why Soatok thinks Signal is better, well those are cherry picked and highly opinionated. There are similar lists of reasons from equally respected security researchers (that have less of a e2ee tunnel vision), that rule out Signal as a serious option due to its centralised and single vendor approach.
Which brings me to the last point. Yes, Signal is a snake-oil vendor that tries to hide the various glaring security issues of their model behind a state of the art e2ee system. But that’s just a fig-leaf not really all that different from how WhatsApp claims to be secure due to them adopting e2ee.
Post-quantum encryption is an active R&D field with no proven to work solutions yet. In fact, solutions that are proudly announced as finally having solved it are regularly silently retracted as other researchers find that they actually offer less security than current state of the art encryption algorithms.
In the Signal article yes, by design those are his opinions on what the minimum requirements are for “beating Signal”, which are not all objective truths. These articles come from a response to people evangelizing one messenger or another to him, some of which have “stronger” security but negatives in other places that make them unacceptable for widespread use (especially for non-techies). In the XMPP article I think the remaining points are very fair in terms of how the XMPP ecosystem works today.
Okay, that’s enough of my time. These replies are for the benefit of others, and I hope I’ve said enough on that for people to make their judgments with more info that what you initially were responding with.