4 ms·

The sentiment of your message almost makes it looks like you are trying to say Whatsapp is no worse at E2E than the others? This is not the case. Signal for ex

The sentiment of your message almost makes it looks like you are trying to say Whatsapp is no worse at E2E than the others? This is not the case. Signal for example has open source which allows to verify that it does not use the message texts for commercial purposes so we can with good reason assume that the messages are at least E2E encrypted properly within the app and at least Signal servers. Yes, of course if you have root access to the device itself, or otherwise hack it, you can compromise any messenger. But that's not even in the same league as having basically a message spying built-in, turned on, always on, inside your damn messenger app itself. Whatsapp calling their app "E2E" in their marketing is a spit in the direction of the users that have the technical knowledge to understand how it really works. It is inaccurate in all the ways that matter. It is accurate only in one technical way that is completely irrelevant in the real world, just put there so they could use the phrase in the marketing while not caring about the true intent behind E2E.

bostik 6y ago
> * The sentiment of your message almost makes it looks like you are trying to say Whatsapp is no worse at E2E than the others?* That was not my intention. I'm trying to say that E2E implies a very specific threat model, and that WhatsApp are in fact in position to subvert theirs in pretty straightforward ways. Their group messages have never been E2E, which means that if they were to force a client update where all communications are always group chats and UI hid this fact, the users would be none the wiser. They could also use their client-side content filtering to build keyword histograms and upload those periodically to their servers, without breaking their E2E. In fact, I was trying to point out that they do not necessarily need to inspect or store message contents. WhatsApp is owned by a marketing analytics giant. With all the noise about E2E and metadata, people forget (or ignore) that traditionally intelligence about communications has been primarily about traffic analysis ("metadata"). Tapping into the communications has been of course a valuable goal, but knowing the communication patterns, frequencies, memberships and direction/timing of communications within groups has been enough to build valuable intelligence. Sure. Access to content allows to do keyword and semantic/NLP based targeting. But the aggregation of marketing cohorts and their various relationships is likely a much more valuable asset. These relationships are also known as the social graph. And E2E, as implemented in WhatsApp, does not protect against it. They know who you communicated with, when, and where you were at the time. Signal on the other hand have done a lot of work to enable not only E2E protected, but also properly untrackable group communications. > But that's not even in the same league as having basically a message spying built-in, turned on, always on, inside your damn messenger app itself. You hit the nail on the head. If you can't trust the client, practically any and all E2E promises are worthless. We agree on this one. You also touch upon a wider problem across the messaging technology space. The term end-to-end-encryption has been hijacked as a high-value keyword by every snakeoil salesman. It confers a high level of trust, precisely because when implemented correctly, it provides guaranteed message content confidentiality. But even in this thread, we see that the term E2E is routinely used to imply even higher standard: that of anonymous communication. Anonymity, confidentiality and integrity are all aspects of communications security. End-to-end can guarantee the last two, assuming the endpoints remain secure or at least trusted. Getting the first one included is going to require a lot of hard work, and in case of WhatsApp, would go directly against their owner's motives.
- fsflover 6y ago
  > I'm trying to say that E2E implies a very specific threat model, and that WhatsApp are in fact in position to subvert theirs in pretty straightforward ways. I disagree. For me, E2E implies that the company itself cannot read my messages. It's not true for Whatsapp, but it's true for Signal/Matrix.
- throw_me_2020 6y ago
  If WA did things like silently degrading/removing E2EE, wouldn't it be discoverable by an independent security researcher? WA seems large enough that the security community would put in that effort periodically.
  - bostik 6y ago
    Well, yes. Obviously. One would hope. But if they were to do so, it could be done so that there likely wouldn't be anything in the visible application or its behaviour to highlight the change to a regular user. Unless you somehow see that the key ratcheting is in use and can confirm the two-sided key state out of band with your peer, you can't tell without disassembling the client. However, this feels like derailing quite far from the original topic. The contract and assumption of E2E protection unavoidably relies on trusting the client(s) and the devices they run on.

bostik 6y ago

> * The sentiment of your message almost makes it looks like you are trying to say Whatsapp is no worse at E2E than the others?* That was not my intention. I'm trying to say that E2E implies a very specific threat model, and that WhatsApp are in fact in position to subvert theirs in pretty straightforward ways. Their group messages have never been E2E, which means that if they were to force a client update where all communications are always group chats and UI hid this fact, the users would be none the wiser. They could also use their client-side content filtering to build keyword histograms and upload those periodically to their servers, without breaking their E2E. In fact, I was trying to point out that they do not necessarily need to inspect or store message contents. WhatsApp is owned by a marketing analytics giant. With all the noise about E2E and metadata, people forget (or ignore) that traditionally intelligence about communications has been primarily about traffic analysis ("metadata"). Tapping into the communications has been of course a valuable goal, but knowing the communication patterns, frequencies, memberships and direction/timing of communications within groups has been enough to build valuable intelligence. Sure. Access to content allows to do keyword and semantic/NLP based targeting. But the aggregation of marketing cohorts and their various relationships is likely a much more valuable asset. These relationships are also known as the social graph. And E2E, as implemented in WhatsApp, does not protect against it. They know who you communicated with, when, and where you were at the time. Signal on the other hand have done a lot of work to enable not only E2E protected, but also properly untrackable group communications. > But that's not even in the same league as having basically a message spying built-in, turned on, always on, inside your damn messenger app itself. You hit the nail on the head. If you can't trust the client, practically any and all E2E promises are worthless. We agree on this one. You also touch upon a wider problem across the messaging technology space. The term end-to-end-encryption has been hijacked as a high-value keyword by every snakeoil salesman. It confers a high level of trust, precisely because when implemented correctly, it provides guaranteed message content confidentiality. But even in this thread, we see that the term E2E is routinely used to imply even higher standard: that of anonymous communication. Anonymity, confidentiality and integrity are all aspects of communications security. End-to-end can guarantee the last two, assuming the endpoints remain secure or at least trusted. Getting the first one included is going to require a lot of hard work, and in case of WhatsApp, would go directly against their owner's motives.

fsflover 6y ago
> I'm trying to say that E2E implies a very specific threat model, and that WhatsApp are in fact in position to subvert theirs in pretty straightforward ways. I disagree. For me, E2E implies that the company itself cannot read my messages. It's not true for Whatsapp, but it's true for Signal/Matrix.
throw_me_2020 6y ago
If WA did things like silently degrading/removing E2EE, wouldn't it be discoverable by an independent security researcher? WA seems large enough that the security community would put in that effort periodically.
- bostik 6y ago
  Well, yes. Obviously. One would hope. But if they were to do so, it could be done so that there likely wouldn't be anything in the visible application or its behaviour to highlight the change to a regular user. Unless you somehow see that the key ratcheting is in use and can confirm the two-sided key state out of band with your peer, you can't tell without disassembling the client. However, this feels like derailing quite far from the original topic. The contract and assumption of E2E protection unavoidably relies on trusting the client(s) and the devices they run on.