▲ 77 points
back
23 comments
It's an interesting paper, but seemingly disconnected from at least two important facts about surveillance:
1. Strong encryption works and data in transit and at rest can be secured.
2. Node security doesn't work, and state actors are behaving irresponsibly by creating a market in insecurity by buying zero-day attacks. This is analogous to buying bioweapons from freelance laboratories and hoping nothing really bad happens. How do you "govern" a dangerous practice like that?
The article appears to assume that all nations will accept a subservient role in a monitored world. Nobody will want actual confidentiality for their government and industry. That doesn't seem plausible.
Strong encryption works, but it's doesn't matter much if you can't trust your hardware, operating system, browser or network. Everything is broken.
Strong encryption works and data in transit and at rest can be secured.
Hey, if we can capture the packets today, it's only a matter of time until quantum computers are brought online and AES256 is rendered as strong as wet paper.
(Or so the gamble might go)
For symmetrical ciphers like AES, quantum computers effectively cut the key length in half for brute force attacks. If you used a good key with AES256, you do not need to worry much about quantum computers.
That's true, and AES is a bad example of a QC threat, but if you're talking about packet captures, in practice AES is usually keyed by something that is very vulnerable to QC.
TLDR: Use of surveillance examples to zoom-out to big picture questions about the role of technology in civilization and the future of humanity versus entrenched capital and increasingly powerful technologies increasingly harnessed to cement present-era dynastic power/economic structures.
Well, aint that just the question. If you're not working on world-changing technologies (~= morally defensible alternatives to today's centralized technologies) then quit your job and get with the program.
Node security is just as much a problem. Sure, not as much in a 'large dragnet of the Internet' sense, but de-centralizing (e.g.) Facebook won't help you to trust your hardware or OS more.
Your TLDR above neatly summarizes my feelings on the matter. Thank you.
The global surveillance network that’s currently being built by the NSA, GCHQ and its collaborator agencies in dozens of countries may become a new international institution, like the World Bank or the United Nations, but more influential and rather harder to govern. And just as Britain’s imperial network of telegraph and telephone cables survived the demise of empire, so the global surveillance network may survive America’s pre-eminence."
Sounds a lot like the Central Intelligence Corporation from Snowcrash. Presumably Google glass owners are going to become the gargoyles.
Presumably ... although I suspect that sensors will be cheap and ubiquitous enough that gargoyles in the manner that Stephenson described them will be quite redundant.
In fact, sensors and delivery systems are close enough to being cheap and ubiquitous enough to enable non-governmental "competing" intelligence organizations becoming a problem.
I keep expecting the first publicly known example to be conspiracy nutcases deploying drones into Area 51 or similar... We've already seen what the semi-anonymity of hacking over the network leads to.
Over the next few years, we're likely to see the next step: With robotic delivery platforms becoming small enough and cheap enough to be hard to stop/track (either because they're hard to spot, or simply because they're cheap enough that there can be ridiculous numbers of them), and sensors and networking lets "anyone" capture high quality images etc., while gaining some chance at evading capture by being physically detached from an intrusion.
On one hand there's plenty of potential for abuse against individuals, and for corporate espionage and organized crime. On the other hand, it seems like we're at a stage where it may become incredibly hard for state intelligence and military actors to prevent individual civilians, political groups or organized crime from carrying out intrusions of an unprecedented level.
Just wait 'til someone finds a zero day for Otis equipment that lets you turn an elevator into a remote control assassination tool. Facial recognition is good enough that if you had the right systems in play you could automate the whole kill chain from identification to placement to action.
And here's the thing; it's a tool that's most effective against the midlist, the strivers, the up and coming execs and entrepreneurs who can't afford a real security detail yet. If you start hearing a lot of stories about promising young execs dying by weird household and office mishaps that could involve automated machinery ( like burned to death by scalding water from a dishwasher, electrocutions, garage doors closing at inopportune moments or such.)... you might be seeing traces of an organized campaign.
Just something to think about next time you step into a hotel elevator...
> someone finds a zero day for Otis equipment that lets you turn an elevator...
I'm reminded of the EE-vs-CS joke about how to design a toaster[1]. Engineering-major rivalries aside, I've always liked it for its caution against over-engineering. For some things - such as elevators - we hit "good enough" a long time ago.
I wonder if "non-programmability" will become a popular feature in the future, specifically to avoid these risks. Using non-Turing Complete "Little Languages"[2] or hardware that's fixed to a single circuit could help. Unfortunately, it's probably too late; we spent the last few decades putting CPUs into everything, and now we get to slowly find out how many security holes that additional complexity left us with.
"In fact, sensors and delivery systems are close enough to being cheap and ubiquitous enough to enable non-governmental "competing" intelligence organizations becoming a problem."
Isn't it called Google?
You are correct.
Yes, gargoyles .. but there's so much resistance, thank god, that they will be socially isolated. The problem is as other describe, namely that the less comprehensive intelligence from the masses will equate to that regardless.
Tell your less technical friends not to post shit to the internet.
(Politico-Technical comment)
India shares intelligence with the NSA because (presumably) given the cozy relationship between the USA & Pakistan, the US (via the NSA) is in a better position to figure out if the intelligence has Terrorism implications for either country.
Last I checked, Russia was not shoveling many billions in aid to Pakistan along with F16s & other weapons systems with the misguided hope that engagement is better than outright confrontation of a failed state with Nukes.
So India "sharing intelligence with NSA vs FSB" as proof of some Network effect is bogus. NSA spies on Pakistan extensively via direct & indirect methods, FSB does not (that we know of).
Ross Anderson's own comment:
http://www.lightbluetouchpaper.org/2014/05/27/post-snowden-t...
Yes. Url changed from https://www.schneier.com/blog/archives/2014/05/the_economics..., which points to this.
This was a bad idea, IMO. The part of the paper quoted in Schneier's blog post does a much better job of explaining the "economics of surveillance" than Ross Anderson's comment does.
Unfortunately I don't have time to look at it right now. Should we just link to the pdf instead?
Blog posts consisting of nothing but an excerpt from someone else's work generally don't count as good links for HN.
Whoops. I saw the link to the PDF at the top of Schneier's article, but missed the blog post link at the bottom. Thanks for updating!
Globalization is zero-sum