Don’t Succumb to Security Nihilism
You might have read today’s shocking Guardian and New York Times articles outlining the many ways that NSA and GCHQ have defeated crypto on the Internet, and have influenced tech companies to insert back doors into their commercial security products.
But pay close attention to this paragraph in Guardian’s article:
The agencies have not yet cracked all encryption technologies, however, the documents suggest. Snowden appeared to confirm this during a live Q&A with Guardian readers in June. “Encryption works. Properly implemented strong crypto systems are one of the few things that you can rely on,” he said before warning that NSA can frequently find ways around it as a result of weak security on the computers at either end of the communication.
Giving up and deciding that privacy is dead is counterproductive. We need to stop using commercial crypto. We need to make sure that free software crypto gets serious security and usability audits.
If we do this right we can still have privacy in the 21st century. If we give up on security because of this we will definitely lose.
Legacy comments, imported from previous version of this blog:
I'm sorry, but this is a totally unrealistic response. It's narrowly focused on the issue of encrypting data and not on the larger issues of pervasive surveillance and its implementation. When you can see the entire data stream, the parts that are cloaked to you stand out by their contrast.
Can we still use strong, non-commercial crypto? Yes. But we're then attaching a big red flag to our communications that says "Hey, NSA - here's something you will really want to look at - this is a guy who's trying to hide something from you!" (And as one poster pointed out, we're only guessing whether it even works or not.)
Part of why commercial crypto worked (we thought) was that everyone was using it. Although obscurity by itself is no solution, in combination with other tactics it's an important piece of the puzzle.
Everyone is using open source crypto (check out the https part of your browser window right now, and how do you connect to your mail server?) for increasingly large parts of what they do on the internet. I don't believe that NSA has all of the keys, and I don't believe they have defeated all cipher suites.
Most crypto attacks work by making the attacks much more efficient. For example, rather than taking billions of years of computation, that can be done in 4 weeks. This crypto is broken, but it doesn't make it worthless. It's still way more expensive for NSA to conduct mass surveillance when everyone uses this type of weak crypto compared to everyone using plaintext.
And it's not all weak. And many of these attacks against crypto are impossible to do in an undetectable way, where-as attacks with crypto not involved can often be completely undetected.
Normal people don't have the data hygiene to use strong crypto. Suggesting to the public that they do so is irresponsible.
The best argument for strong crypto is to make its use pervasive so that broad surveillance is costly, not that it can be relied upon.
Ahh, so you've already given up.
Well, it can't get any worse for them, or for us. I don't see the problem in laying out to people their options, even if all of them currently are hard. At least it will help everyone pinpoint the problem. If more people become more acquainted with the issues, that will mean more people willing to do something to help solve it, even if it is just donating to the right kickstarter projects..
Wholeheartedly agree with this statement - however, I must express some sympathy for your first commenter's point of view. A very large part of the population has no experience with cryptography, and there is a certain amount of technical proficiency involved which may be overlooked by those with experience. The problem is that there can be a learning curve, even if it's not necessarily that steep, but the right to privacy shouldn't depend solely upon one's proficiency at cryptography.
It's possible that until there are easy to follow, well-written references and step-by-step instructions which can be reliably implemented by anyone from the savviest engineer to the average layman/woman who uses the computer for sending cat pictures to her grandchildren, a large number of people will simply bypass this whole discussion. Which is too bad, because it affects them too.
He didn't say which ones they haven't cracked, and there's no way to know. We rely on experts looking for holes in a protocol to know whether there really are any. Since they didn't find these but the NSA did, we now know that that method guarantees nothing.
Yes, I've certainly given up now. This is it. Snowden can try to soften the blow, but the truth is that crypto was never workable after all. You can only trust a one-time pad cipher. Everything else is hopeless.