[kwlug-disc] Secure IM news

[locklin.jason at gmail.com]

There are certainly two problems in the IM sphere: the lack of privacy/security, and the silos. Signal makes a huge leap in terms of the first -by making good e2e encryption work with good UX. The lack of federation and requirement of GCM means that that advancement can really go no further. The signal servers are going to have all the metadata, and Google is going to track everyone's device.

This feeds into the second problem. With a silos, there is no room for sustained growth. You may be able to convince people to settle on Signal, but convincing them to move on to another system with better metadata security down the road will be very difficult -and what ultimately convinces them to move on will likely be something other than privacy. Silos have always been, and will always be a long-term loss strategy. They will only ever lead to long-term system-hopping, without incremental improvement (of course, silos can be very short-term lucrative for the winners).

The only long-term solution for truly free and private communication is either a federated or p2p network, based on open standards. Of course, those systems are hard to get started because they are technically and economically more challenging. I have certainly installed and experimented with dozens of such, to see them go unused and quiet. I think Matrix has a shot though. It has the unusual confluence of long term potential for truly free and private communication, while actually being very useful now. I basically use it daily as my general IRC client/bouncer for (un-secure) group messaging in large existing groups. I use Signal daily as well, as it doubles as my SMS client, so both have figured out ways to be useful
immediately, while pushing for change. 



On Thu, 24 Nov 2016 04:44:16 +0000
Chamunks <chamunks at gmail.com> wrote:

> "Yep! Signal has made great improvements to OTR. I personally think the
> best change is rather that the ratchet, being based on deterministic
> algorithms (hashes), supports offline messaging in a way OTR couldn't and
> therefore couldn't be used (effectively; you could with xabber but it was
> The Worst Thing) for mobile. Actually, my belief was that OTR also used a
> different key for each message, but it had to negotiate the key actively
> with the partner, which is why you had to have an active both-people-online
> session to use it. Do you know if I'm wrong about that?"
> 
> After some review you are correct.  I'm not entirely certain of the
> difference at this point I always thought / was under the impression OTR
> was per session.
> 
> As for google's GCM, it's literally just a push system/relay and index of
> registered devices on the GCM network it acts as a secure relay for data.
> As you basically have to trust the internet underneath your fibers there
> will be some proprietary systems in there somewhere you're not even going
> to be aware of.  Cisco/Juniper etc they're all proprietary IIRC.  I would
> be more concerned about the data sent via GCM more so than the fact that
> its using GCM.
> 
> The benefit of this system is that you don't add security holes opening up
> sockets to the internet meaning it's more secure than leaving an open
> web-socket/socket.  Or the alternative of having a DHT or something
> tracking the online status of your neighboring devices.  Signal isn't
> intended to be an anonymity network its intended to provide provably secure
> conversations past beyond the common standard.
> 
> If you want something ultra-secure that won't have anyone on it ( the
> biggest problem with a messaging protocol is that someone needs to actually
> use it for it to be useful ) you could try TOX
> <https://wiki.tox.chat/clients> or something via the TOR network.  That'd
> really be the only truly open-source messaging system that would be secure
> end to end and also anon and free of proprietary backing.
> 
> I do really like the concept of open standards though but I think that
> Moxie Marlinspike is rightfully defending his protocol at the moment though
> because if I were trying to build the reputation for my protocol I would
> fend off people who would likely run things poorly and at the moment it
> seems like he's doing a better job than anyone else.  Once you get mass
> adoption maybe he might be convinced into opening the standard.
> 
> 
> 
> On Wed, Nov 23, 2016 at 11:16 PM Nick Guenther <nguenthe at uwaterloo.ca>
> wrote:
> 
> >
> >
> > Le 23 novembre 2016 21:41:08 HNE, Chamunks <chamunks at gmail.com> a écrit :
> > >SignalApp is perfectly open they only use GCM for an efficient method
> > >to
> > >tell clients where to send the message to p2p. None of the messages
> > >actually travel through GCM.  This is how android's designed to work
> >
> > There are multiple layers of openness here. The code is open, but the
> > service is not because they don't federate, and to protect the brand (the
> > argument being that for the non-obsessed-geek crowd, that also protect a
> > security by avoiding imitators) they actively stamp out anyone running even
> > an alternate copy of the server. On top of that, the fact that any part of
> > the infrastructure depends on Google's non-free systems is a lack of
> > openness; isn't one of the core FSF software freedoms the ability to use
> > the software wherever you can? Well, because of the tie into to GCM/Apple
> > Push, you can't run it on BlackBerry 10, you can't run it under Pidgin
> >
> > >
> > >The best part of the signal protocol is that unlike the OTR spec, it
> > >ratchets its keys for every message since its so cheap on the CPU these
> > >days to make more keys.
> >
> > Yep! Signal has made great improvements to OTR. I personally think the
> > best change is rather that the ratchet, being based on deterministic
> > algorithms (hashes), supports offline messaging in a way OTR couldn't and
> > therefore couldn't be used (effectively; you could with xabber but it was
> > The Worst Thing) for mobile. Actually, my belief was that OTR also used a
> > different key for each message, but it had to negotiate the key actively
> > with the partner, which is why you had to have an active both-people-online
> > session to use it. Do you know if I'm wrong about that?
> >
> > I am not sure if you were debating my reluctance with this message. If you
> > were, let me clarify that I'm not reluctant about Signal; it sucks that
> > they have decided they don't have the resources to do federation, but I
> > have peace with that and use and recommend Signal every day. My reluctance
> > is about Matrix claiming "e2e group chats!" without maybe fully thinking
> > through the threat model.
> >
> >
> >


-- 
locklin.jason at gmail.com <locklin.jason at gmail.com>