3 ms·

Generally you have advocated TextSecure as one of the only good available options. Given the comments you linked to, and the time that has passed. Is this sti

by dsp1234 11y ago

Generally you have advocated TextSecure as one of the only good available options. Given the comments you linked to, and the time that has passed. Is this still your top suggestion? Additionally, in your mind, what criteria are most important for you (if you were going to build your own scorecard, which I know you cautioned against) to help the cause of getting people to use secure messaging? Put another way, if a new secure messaging service came out, how would you identify that it's a TextSecure-like recommendation vs a cryptocat-like recommendation (other than the obvious 'they rolled their own encryption, and did it badly')?

tptacek 11y ago
Most cryptographers, if pressed on the question of which secure messenger they would trust most purely based on sound cryptography, would say TextSecure/Signal. TextSecure is one of two secure messaging schemes --- the other being Silent Circle's† --- designed from the beginning to be cryptographically secure, and not as an instant messenger first and a secure transport second. The other approach, of designing user-friendly instant messaging systems and then layering cryptography on top, has been pretty much an unmitigated disaster. Cryptocat is the post child for these systems --- and the EFF scorecard gave it a perfect score! Some of these systems DIY their crypto --- Telegram is like that, as was Cryptocat --- and suffer accordingly. Others pull in secure cryptography libraries like Nacl/Sodium. Those are marginally more secure, but cryptographic secure messaging is its own problem domain, and Nacl/Sodium doesn't help much with them. For me, the most important question in a secure messenger is whether it's actually secure. The second most important question is how many people can I talk to using it. A very, very distant third question is "how usable is it?" There's no one system that gives you all those attributes, so I suggest looking at secure messaging as 2-3 problems, not just one: https://news.ycombinator.com/item?id=11454172 https://news.ycombinator.com/item?id=11454172 If it was up to me, there wouldn't be scorecards for secure messaging systems. I think they intrinsically oversimplify a complicated problem. But I have a lot of thoughts on what would and wouldn't make such a scorecard less terrible. I'm waiting for EFF to release the "re-done" scorecard to write them up. † Silent circle is at an uncomfortable intersection of being closed source and not particularly popular, so it doesn't get much attention.
- pvg 11y ago
  >designing user-friendly instant messaging systems and then layering cryptography on top, has been pretty much an unmitigated disaster. Hang on, what about iMessage or OTR-over-anything.

by dsp1234 11y ago

tptacek 11y ago

Most cryptographers, if pressed on the question of which secure messenger they would trust most purely based on sound cryptography, would say TextSecure/Signal. TextSecure is one of two secure messaging schemes --- the other being Silent Circle's† --- designed from the beginning to be cryptographically secure, and not as an instant messenger first and a secure transport second. The other approach, of designing user-friendly instant messaging systems and then layering cryptography on top, has been pretty much an unmitigated disaster. Cryptocat is the post child for these systems --- and the EFF scorecard gave it a perfect score! Some of these systems DIY their crypto --- Telegram is like that, as was Cryptocat --- and suffer accordingly. Others pull in secure cryptography libraries like Nacl/Sodium. Those are marginally more secure, but cryptographic secure messaging is its own problem domain, and Nacl/Sodium doesn't help much with them. For me, the most important question in a secure messenger is whether it's actually secure. The second most important question is how many people can I talk to using it. A very, very distant third question is "how usable is it?" There's no one system that gives you all those attributes, so I suggest looking at secure messaging as 2-3 problems, not just one: https://news.ycombinator.com/item?id=11454172 https://news.ycombinator.com/item?id=11454172 If it was up to me, there wouldn't be scorecards for secure messaging systems. I think they intrinsically oversimplify a complicated problem. But I have a lot of thoughts on what would and wouldn't make such a scorecard less terrible. I'm waiting for EFF to release the "re-done" scorecard to write them up. † Silent circle is at an uncomfortable intersection of being closed source and not particularly popular, so it doesn't get much attention.

pvg 11y ago
>designing user-friendly instant messaging systems and then layering cryptography on top, has been pretty much an unmitigated disaster. Hang on, what about iMessage or OTR-over-anything.