4 ms·

It's a thoughtful rebuttal, but I'm not sure it hits the point. As usual, there's nuance here. The top selling points of Signal are security and privacy. If yo

by growse 5y ago

It's a thoughtful rebuttal, but I'm not sure it hits the point. As usual, there's nuance here. The top selling points of Signal are security and privacy. If you want these things, then there's a lot of things that maintaining complete control of the protocol gives you. From the original "Ecosystem is moving" essay: "By contrast, WhatsApp was able to introduce end-to-end encryption to over a billion users with a single software update". That's the point of centralisation. Control means agility in the face of adversary, which when you're trying to optimize for security and privacy is pretty useful. The comparison to HTML is therefore wide of the mark, because the whole point of HTML is to be interoperable. A better comparison would be with something like TLS/SSL, and then you can judge how effective the combination of slow-moving protocol evolution and a security focus has worked over the last few decades. There is no bad choice. There are just compromises that you have to make depending on what you choose, and some of those compromises are more suited to some objectives than others.

ddevault 5y ago
TLS/SSL is a better comparison, and it's a protocol which has enjoyed widespread and unprecedented success in securing the privacy of users worldwide, despite being decentralized. The tradeoffs Moxie was thinking of when rejecting federation had more to do with centralizing his power over Signal rather than any kind of technical limitations.
- tialaramex 5y ago
  For TLS we've had to drag everybody along, kicking and screaming. You call this "widespread and unprecedented success" but the reality is that later this week you're going to see even more outrage and despair from people who didn't like being centralized and so, of course, they're screwed because the price of their "freedom" is the eternal vigilance they couldn't be bothered with. draft-moriarty-tls-oldversions-diediedie (now RFC 8996) only exists because of this. Moxie doesn't need to worry about security problems with a protocol he shipped last century but for TLS we had to actually document that people should stop using that - because of course they are anyway. In the Web PKI we have in practice even more centralization, in principle it would be possible to operate a public CA that simply isn't trusted by Mozilla, Apple, Microsoft or Google, but in practice you need all of them to have a viable service. The de facto situation is that you need logging with Google (and any other service, but Google is mandatory) and you need a sign-off from Mozilla, or you are dead in the water. Now, I think the Web PKI is arguably the only global PKI with something approaching public oversight and it's a huge success. But it's not decentralized in the way you mean, not at all.
  - ddevault 5y ago
    It's not without drawbacks, but TLS gives almost everyone good security/privacy by default. And for those who really need it, anyone who does even a basic amount of research can gain very good levels of confidence in the privacy of their TLS setup. Also, there are degrees of federation. It doesn't have to be a wild wild west of the sort we see with projects like Mastodon. A closed federation can have strict rules on who can participate, can enforce constraints on auditing and rapid updates, and so on, and still be a federation. In fact, we see some of this in TLS - not everyone can participate, but there are dozens of independent organizations working within an internal governance models. Another example is EFnet, which is a federated chat network which has regulations on who can participate: http://www.efnet.org/?module=docs&doc=16 http://www.efnet.org/?module=docs&doc=16 Moxie conveniently ignores all of this because it serves his interests to be ignorant of it. It has been repeatedly brought to his attention. Moreover, he has put out legal threats to any project which tries to fork Signal to research these features.
    - tialaramex 5y ago
      > In fact, we see some of this in TLS - not everyone can participate, but there are dozens of independent organizations working within an internal governance models I think what's happened here is that you don't understand the difference between TLS, a network protocol, and the Web PKI, a Public Key Infrastructure and so you've sort of muddled both together, which is revealing on its own. You've also imagined a fairytale outcome that doesn't match reality. Moxie's strategy works and yours didn't, and you're here telling everybody you don't agree but reality doesn't care whether you agree with it or not.
      - ddevault 5y ago
        >I think what's happened here is that you don't understand the difference between TLS, a network protocol, and the Web PKI, a Public Key Infrastructure and so you've sort of muddled both together, which is revealing on its own. I understand the difference, and the condescending snark is not appreciated. I don't think the difference is very important in this case. It's also revealing, perhaps, that you don't understand that the "Web PKI" is used far and wide in domains outside of the web. >Moxie's strategy works and yours didn't, and you're here telling everybody you don't agree but reality doesn't care whether you agree with it or not. Oh, bugger off. The only group who has even made a serious attempt to do it properly is Matrix and one "meh" implementation does not damn an entire design.
- UncleMeat 5y ago
  And an absolutely tremendous number of devices are still using versions of TLS with serious weaknesses and there is very little pressure actually moving them to more current versions. Thus, tradeoffs.
- Seirdy 5y ago
  Are you aware of a federated, decentralized approach to e2ee messaging that doesn't require trusting the servers? i.e., without the metadata leakage of email, xmpp, Matrix, etc.? The only options I know of have poor support for offline messaging (and kill the battery) due to being peer-to-peer. I'm not yet aware of any federated approaches that prevent a server from connecting a sender and recipient, but would love to be educated otherwise.

by growse 5y ago

ddevault 5y ago

TLS/SSL is a better comparison, and it's a protocol which has enjoyed widespread and unprecedented success in securing the privacy of users worldwide, despite being decentralized. The tradeoffs Moxie was thinking of when rejecting federation had more to do with centralizing his power over Signal rather than any kind of technical limitations.

tialaramex 5y ago
For TLS we've had to drag everybody along, kicking and screaming. You call this "widespread and unprecedented success" but the reality is that later this week you're going to see even more outrage and despair from people who didn't like being centralized and so, of course, they're screwed because the price of their "freedom" is the eternal vigilance they couldn't be bothered with. draft-moriarty-tls-oldversions-diediedie (now RFC 8996) only exists because of this. Moxie doesn't need to worry about security problems with a protocol he shipped last century but for TLS we had to actually document that people should stop using that - because of course they are anyway. In the Web PKI we have in practice even more centralization, in principle it would be possible to operate a public CA that simply isn't trusted by Mozilla, Apple, Microsoft or Google, but in practice you need all of them to have a viable service. The de facto situation is that you need logging with Google (and any other service, but Google is mandatory) and you need a sign-off from Mozilla, or you are dead in the water. Now, I think the Web PKI is arguably the only global PKI with something approaching public oversight and it's a huge success. But it's not decentralized in the way you mean, not at all.
- ddevault 5y ago
  It's not without drawbacks, but TLS gives almost everyone good security/privacy by default. And for those who really need it, anyone who does even a basic amount of research can gain very good levels of confidence in the privacy of their TLS setup. Also, there are degrees of federation. It doesn't have to be a wild wild west of the sort we see with projects like Mastodon. A closed federation can have strict rules on who can participate, can enforce constraints on auditing and rapid updates, and so on, and still be a federation. In fact, we see some of this in TLS - not everyone can participate, but there are dozens of independent organizations working within an internal governance models. Another example is EFnet, which is a federated chat network which has regulations on who can participate: http://www.efnet.org/?module=docs&doc=16 http://www.efnet.org/?module=docs&doc=16 Moxie conveniently ignores all of this because it serves his interests to be ignorant of it. It has been repeatedly brought to his attention. Moreover, he has put out legal threats to any project which tries to fork Signal to research these features.
  - tialaramex 5y ago
    > In fact, we see some of this in TLS - not everyone can participate, but there are dozens of independent organizations working within an internal governance models I think what's happened here is that you don't understand the difference between TLS, a network protocol, and the Web PKI, a Public Key Infrastructure and so you've sort of muddled both together, which is revealing on its own. You've also imagined a fairytale outcome that doesn't match reality. Moxie's strategy works and yours didn't, and you're here telling everybody you don't agree but reality doesn't care whether you agree with it or not.
    - ddevault 5y ago
      >I think what's happened here is that you don't understand the difference between TLS, a network protocol, and the Web PKI, a Public Key Infrastructure and so you've sort of muddled both together, which is revealing on its own. I understand the difference, and the condescending snark is not appreciated. I don't think the difference is very important in this case. It's also revealing, perhaps, that you don't understand that the "Web PKI" is used far and wide in domains outside of the web. >Moxie's strategy works and yours didn't, and you're here telling everybody you don't agree but reality doesn't care whether you agree with it or not. Oh, bugger off. The only group who has even made a serious attempt to do it properly is Matrix and one "meh" implementation does not damn an entire design.
UncleMeat 5y ago
And an absolutely tremendous number of devices are still using versions of TLS with serious weaknesses and there is very little pressure actually moving them to more current versions. Thus, tradeoffs.
Seirdy 5y ago
Are you aware of a federated, decentralized approach to e2ee messaging that doesn't require trusting the servers? i.e., without the metadata leakage of email, xmpp, Matrix, etc.? The only options I know of have poor support for offline messaging (and kill the battery) due to being peer-to-peer. I'm not yet aware of any federated approaches that prevent a server from connecting a sender and recipient, but would love to be educated otherwise.