4 ms·

As tptacek explained very well yesterday, this is not the way crypto systems are built: > Be honest with yourself. Crypto doesn't get beta-tested into resilien

by ch0wn 13y ago

As tptacek explained very well yesterday, this is not the way crypto systems are built: > Be honest with yourself. Crypto doesn't get beta-tested into resiliency. Strong systems start out strong. If you're building something because its your dream to thwart the NSA, don't kid yourself into thinking that you'll get there by first protecting people's Warcraft clans. (https://news.ycombinator.com/item?id=6961260 https://news.ycombinator.com/item?id=6961260)

unicornporn 13y ago
Yes, and don't forget that if the NSA gets direct access (by threat, payment or persuasion) to their infrastructure these bold statements mean null. Secure communication will NOT be centralized.
eps 13y ago
tptacek said it! Hallelujah! One man's beta-testing is another man's peer review. They are fundamentally the same process, so - sure, you can iterate your crypto design into a better shape by releasing a production app and attracting the reviewers by making over-the-board marketing claims. It's certainly more conventional to request feedback during the design phase, but it doesn't mean that doing it ass-backwards is not an option. In the end it's the underlying motivation that matters.
- nav1 13y ago
  They are not the same process at all. Releasing a secure messaging application that leaks data is really not okay.
  - eps 13y ago
    I didn't argue that releasing an app with a weak crypto was okay. I replied specifically to this point - > Crypto doesn't get beta-tested into resiliency and crypto can most certainly be beta-tested into resiliency. It all depends if they are learning from their mistakes.
- aioprisan 13y ago
  Requesting feedback when the app is released to the masses and they actually believe that it's secure and doesn't send out their location data in clear text can have huge consequences, much larger than a small visual bug in an app. It is a core flaw in what they're promising, and you can't iterate over that. How about if a bank did that with their login system? Iterating over fatal bugs and exposing all of their customers to attack? The loss of credibility and fines would quickly put them out of business.

by ch0wn 13y ago

unicornporn 13y ago

Yes, and don't forget that if the NSA gets direct access (by threat, payment or persuasion) to their infrastructure these bold statements mean null. Secure communication will NOT be centralized.

eps 13y ago

tptacek said it! Hallelujah! One man's beta-testing is another man's peer review. They are fundamentally the same process, so - sure, you can iterate your crypto design into a better shape by releasing a production app and attracting the reviewers by making over-the-board marketing claims. It's certainly more conventional to request feedback during the design phase, but it doesn't mean that doing it ass-backwards is not an option. In the end it's the underlying motivation that matters.

nav1 13y ago
They are not the same process at all. Releasing a secure messaging application that leaks data is really not okay.
- eps 13y ago
  I didn't argue that releasing an app with a weak crypto was okay. I replied specifically to this point - > Crypto doesn't get beta-tested into resiliency and crypto can most certainly be beta-tested into resiliency. It all depends if they are learning from their mistakes.
aioprisan 13y ago
Requesting feedback when the app is released to the masses and they actually believe that it's secure and doesn't send out their location data in clear text can have huge consequences, much larger than a small visual bug in an app. It is a core flaw in what they're promising, and you can't iterate over that. How about if a bank did that with their login system? Iterating over fatal bugs and exposing all of their customers to attack? The loss of credibility and fines would quickly put them out of business.