Haystack project responds to 'security concerns', looks like it's falling apart
blog.jgc.org
blog.jgc.org
It's easy to assume that knowing how to code with sockets and call into a crypto library makes you more clueful than the people running "the great firewall" or whatever it is Iran has. They're trying to censor the Internet, how smart could they be?
Well, they're smart enough to pay someone a million dollars to break a circumvention tool without giving it a second thought. The Tor project, which actually has some software security talent, couldn't clear the "adversary with $1MM" bar (they screwed up Diffie-Hellman). And Tor has an "easy" problem; stealth circumvention is much harder.
Get circumvention at all wrong and you achieve the opposite of what the tool is intended for: you put a big red flag on people breaking their local laws. The risk/reward structure here is totally broken, even before you consider how likely it is that everyone's machines in the country you're trying to "help" are already rootkitted.
Don't build circumvention tools.
I think a state-sponsored circumvention tool, backed by one or more western governments, would --- if procured properly, which would never happen, and instead would involve Lockheed or SAIC --- stand enough of a chance against a foreign adversary to be worth considering. Short of that, you run right into a very simple wall: once your tool starts to matter, Iran will happily spend many millions of dollars to have it turned.
But the police state won't be defeated by broad-scale usage of some technical tool. Small-scale usage by a small core group of hackers who constantly adapt (and, frankly, constantly suffer attrition by the police and face the non-zero chance of making one mistake that allows the police to catch them all) might be possible, but if we're going to defeat a police state it will have to be on something other than a purely technical level like that.
Could the entirety of your arguments against cryptosystem design not be applied to SSL in its infancy? Aren't the resources of the attackers always larger than the resources of the developers?
There's clearly a large gap between what currently accepted protocols can do, and the capabilities of zero knowledge proofs and secure multiparty computation. It's certainly smart to reuse existing blocks rather than recreating their errors, but that doesn't mean all crypto protocol design work is antiquated.
(although please don't think I'm saying distributing Haystack to users in repressive regimes was anything but criminal malpractice. Even TOR says to not rely on it for strong anonymity)
Since I never argued that all crypto protocol design work was "antiquated", I'm not sure how to respond to that. What I've said a bunch of times is: it is prohibitively difficult to get stealth circumvention tools right on the first try, and the adversary being circumvented could pay the couple million it would take to turn any such tool, without even noticing the cost.
1) Why were they writing their own system to do this? Haystack sounds similar to Tor, which is probably more tested.
2) How is hype being allowed to succeed in decisions made to address very real causes? The glamour of hacker culture appears to have won, over the science of hacking.
Hell, The Guardian made him Innovator of the Year: http://www.guardian.co.uk/media/2010/mar/29/austin-heap-mega...
In doing so, he likely made the “fundamental economics” problem that you mentioned even worse. During Iran’s election, people everywhere were anxious to help protestors in any way they could. Legitimate anti-censorship groups may well have benefited from increased media attention and a resulting increase in funding, had Heap not stolen the spotlight. Lay people could have been educated on the real challenges in thwarting censorship. Instead many were led to believe that the problem had already been solved by a boy genius, and that it was simply a matter of sending him enough thumb drives. Heap’s dishonesty and the lax journalism that encouraged it should rightly be discussed.
But either way: a better team would almost certainly not have produced a tool that the Islamic Republic couldn't have turned against its innocent users, because no team we can conceive of can match Iran's ability to fund countermeasures.
People are inherently captive to the narrative fallacy. They adhere to stories of "us" against "them", and what a great "them" they find in Iran, which stones women and hangs people for writing blog posts! But in the real world, the narrative is irrelevant. The person who figures out how to turn Haystack^2 into a dissident marker and arrest warrant generator isn't going to be a Muslim, and almost certainly will never have set foot in Iran --- based on how governments currently acquire exploits, they probably won't even know they're working for Iran. There are many hundreds of very, very smart people who will write security tools for $1MM (their handlers will probably take the other $9MM for themselves).
And perhaps to your main point: it’s also not like the Haystack project would have succeeded if only its members had a bit more technical know-how, or if they had just had been willing to open their source, etc. The challenge in thwarting the Islamic Republic’s censorship measures is almost impossibly enormous.
However, this challenge could conceivably be met. Perhaps another Phil Zimmermann might come along one day and revolutionize steganography to bring us a step closer. Unfortunately this becomes less likely with a consummate self-promoter like Austin Heap publicly claiming to have already solved the entire problem.
Nobody smart enough to solve this problem would have the hubris to release it to the public on their own.
Tor conceals your origin from the endpoint (web server etc).
Haystack conceals the fact you are communicating and about what, from a man in the middle, by pretending to be innocuous web requests.
Tor anonymizes the origin of your traffic, and it encrypts everything between you and the Tor network and everything inside the Tor network, but it can't encrypt your traffic between the Tor network and its final destination. If you are communicating sensitive information, you should use as much care as you would on the normal scary Internet — use HTTPS or other end-to-end encryption and authentication.
From what I understand, there was a diagnostics mode that allowed an attacker to identify running copies of the program -- essentially the exact opposite of what it's supposed to do. Even with Haystack's proxy servers shut down, there may still be a risk to any user who has it installed.
This could have easily been prevented. Many people with security backgrounds, including myself, contacted Austin Heap and asked for technical information. He declined. Instead, they went ahead and distributed it to vulnerable live testers.
On the flipside, while the goals of the project seem noble enough, I can't help but wonder why they didn't just try to value-add to something like Tor.
Regardless, I'd much rather they stop distributing while they get things sorted out than offer a false sense of security to their users who could potentially be using Haystack in ways that, if caught, might put their lives/careers in jeopardy.