Wouldn't it just be a law requiring ISP's to kill peering upon the government's request? Am I missing something here? If it isn't feasible, I would think that's because it isn't legal, not because of a technical hurdle.
If the "Kill Switch" bill passes, the legal question will be answered. As no-one's optimistic the bill will fail [1] commentary has moved on to: can it actually be implemented?
Personally, I don't see that being a very interesting question either. Once it's law, "Kill Switch"-Compliance will get built into the big switches and routers. It would only be a technical challenge to implement in the near term. And I don't think you can be paranoid-enough to think the government plans on hitting this switch in the near term, yet be rational enough to believe the government would wait for legal authority to do so.
So whether it's feasible in the next 5 years is pretty academic.
[1] If people aren't going to take issue with standing constitution-free-zones covering 80% of the population, the growing surveillance state, increasingly-intrusive security theatre at the airport and greenlit assassinations of suspected terrorist US citizens, an argument about internet access during hypothetical 'states of emergency' has a snowball's chance in hell.
I think you're right on. People complained that CALEA was technically infeasible, but those people just didn't understand it; ISPs are obligated to make CALEA work and they complied even if it meant redesigning their networks. If ISPs have to update their OSS systems to support the "kill switch", they will do it.
That's actually the example [CALEA] I had in mind as well.
I don't see a huge technical hurdle here, other than maybe a script to disable all peering interfaces for every ISP in the US. This could be done by hand just as easily should Uncle Sam come knocking.
The real issue, to me, is the why behind this. The only real motivator is to prevent people from organizing to overtake a government they perceive as corrupt. Is it really worth exploring crippling our economy and stifling free speech at the same time? Forget the technical hurdle, what about the constitutional one?
Interesting that we also claim to have a way to "force" internet on a country who kills it, but at the same time are looking for a legal basis to kill it ourselves.
> "Forget the technical hurdle, what about the constitutional one?"
We haven't had much luck with that one lately.
I don't think the "Why" is quite so transparently dystopian. They don't want to turn off the entire internet. It's just another attempted end-run around the judicial process, to make it easier to further political and economic goals. e.g. filtering WikiLeaks or BitTorrent.
The idea that they would need this to disconnect critical infrastructure to protect it from cyberattack is laughable. Any critical infrastructure that could still operate independent from the internet should not have a connection to the internet, and if it did should certainly have it's own disconnect capability.
The only reason to put disconnect capability on the ISP or backbone carrier is to do it against the will of the target facility. If they wanted to protect things like the Hoover Dam [1] they'd just issue/enforce some government regs. [2]
[1] The Dam Authority has already taken issue with being a talking point in this debate. Pointing out that, no, they are not foolish enough to have dam controls connected to the internet.
[2] I'm pretty sure these already exist, as regards air-gaps for critical infrastructure and security requirements for networks that do have a connection to the public internet. There may not be a unified national service to flip connection-kill-switches, but that would be resolved with a government network project, not a new law. The government already has legal authority over infrastructure.
While I don't think it is technically impossible, I don't think it is as quick and easy to deploy and support as CALEA. For example...
- ISPs would have to be able to characterize all their connections and subnets as being part of the critical infrastructure and of which parts of the critical infrastructure those users or networks were members.
- This entails insane tracking and management - literally thousands of smaller CIP organizations might be served by one ISP - they have to know in real time what IPs/ranges/etc, and on what ports, those organizations are operating
- Central management or the ability to engage the 'kill switch' implies that there is some means of remotely (read: outside the ISP control) engage the kill switch and remove those organizations from the network
All of those points (and many more I can include later) make this vastly different from CALEA, where a warrant compels ISP personnel to engage collection on an IP or target of interest and to turn over that capture to the authorities. That puts the control in the hands of the ISP - the kill switch would put the controls of ISP assets in the hands of the government.
The thing about the article is that it points that it would be very hard to determine how to "limit traffic to critical infrastructure". But "conveniently", when a state has a mandate that "must be done" while being hard to do surgically, that state has mandate to muck-about as it sees fit (that's why search warrants are supposed to specific, etc).
So "conveniently", it may turn out that to implement the bill, ISPes would be required to install flexible remote controlled switches of various sorts. Meaning that the real impact of the bill would be the configuration of a remote-controlled censoring system. IE, a Great-Firewall-of-America "by accident" (yeah).
Instead of a Kill Switch, I think it would be possible to make a Flood Switch. When enough shit would hit the fan, to make the president issue the Kill order, all bets would be off.
Let's say Americans are using Twitter or a foreign website to organize riots in a civil war, I don't think once the order is given and such a bill is in place, that the government has to ask nicely and force compliance. It will put all government computers and some very big tubes into DDOS'ing whoever is publishing something they don't want published at that very moment.
If I were these hypothetical guerillas, I wouldn't have one site, I'd have hundreds. Good luck DDOSing 600 independent sites at once, even if you're the government. I'm pretty sure killing the ISPs is simple by comparison.
You could certainly bust the internet into a lot of small unconnected subnets, simply by telling all the backbone providers to disconnect peering. I would expect that .mil and continuity of government circuits would be tagged as exempt, and would continue to work.
I dunno what happens if your provider happens to be someone like Hughes or DirectPC. That's one heck of a large subnet.
And note, if a 'kill switch' is ever required, the agency that will enforce its adoption by ISPs? Just as with wiretap requirements and broadcast language/content censorship, the FCC.
Remember that before cheering on the FCC to have authority over which ISPs are sufficiently 'neutral'. (And don't be surprised when 'neutral' gets redefined over time into 'compliant with the FCC's political biases'.)
From a brute-force PoV, I imagine you can have armed FBI agents in the offices of major ISPs in an hour or so. If they block their customers and they also block their resellers, then really you can handle most of the Internet with probably a dozen 2-man teams. It's not really that critical if it's deemed too hard to block two guys on the same local ISP from chatting.
Not to mention that even rudimentary steps - knocking out DNS or even just Google, Yahoo, and Bing - would put 95% of Americans effectively offline.
Metcalfe's Law works both ways - if you cut off even 3/4 of the internet, it's now only a square-root of a square-root as valuable as it used to be.
If the kill switch is built into routers, expect that to be used by hackers. You'll have to use open source router code like dd-wrt to avoid it.
This is a bad idea and is trying to change how the Internet works. The Internet was designed to stay on even if there was a nuke attack. ISPs have been handeling DOS attacks for years successfully without a kill switch.
14 comments
[ 4.2 ms ] story [ 34.4 ms ] threadPersonally, I don't see that being a very interesting question either. Once it's law, "Kill Switch"-Compliance will get built into the big switches and routers. It would only be a technical challenge to implement in the near term. And I don't think you can be paranoid-enough to think the government plans on hitting this switch in the near term, yet be rational enough to believe the government would wait for legal authority to do so.
So whether it's feasible in the next 5 years is pretty academic.
[1] If people aren't going to take issue with standing constitution-free-zones covering 80% of the population, the growing surveillance state, increasingly-intrusive security theatre at the airport and greenlit assassinations of suspected terrorist US citizens, an argument about internet access during hypothetical 'states of emergency' has a snowball's chance in hell.
I don't see a huge technical hurdle here, other than maybe a script to disable all peering interfaces for every ISP in the US. This could be done by hand just as easily should Uncle Sam come knocking.
The real issue, to me, is the why behind this. The only real motivator is to prevent people from organizing to overtake a government they perceive as corrupt. Is it really worth exploring crippling our economy and stifling free speech at the same time? Forget the technical hurdle, what about the constitutional one?
Interesting that we also claim to have a way to "force" internet on a country who kills it, but at the same time are looking for a legal basis to kill it ourselves.
We haven't had much luck with that one lately.
I don't think the "Why" is quite so transparently dystopian. They don't want to turn off the entire internet. It's just another attempted end-run around the judicial process, to make it easier to further political and economic goals. e.g. filtering WikiLeaks or BitTorrent.
The idea that they would need this to disconnect critical infrastructure to protect it from cyberattack is laughable. Any critical infrastructure that could still operate independent from the internet should not have a connection to the internet, and if it did should certainly have it's own disconnect capability.
The only reason to put disconnect capability on the ISP or backbone carrier is to do it against the will of the target facility. If they wanted to protect things like the Hoover Dam [1] they'd just issue/enforce some government regs. [2]
[1] The Dam Authority has already taken issue with being a talking point in this debate. Pointing out that, no, they are not foolish enough to have dam controls connected to the internet.
[2] I'm pretty sure these already exist, as regards air-gaps for critical infrastructure and security requirements for networks that do have a connection to the public internet. There may not be a unified national service to flip connection-kill-switches, but that would be resolved with a government network project, not a new law. The government already has legal authority over infrastructure.
- ISPs would have to be able to characterize all their connections and subnets as being part of the critical infrastructure and of which parts of the critical infrastructure those users or networks were members. - This entails insane tracking and management - literally thousands of smaller CIP organizations might be served by one ISP - they have to know in real time what IPs/ranges/etc, and on what ports, those organizations are operating - Central management or the ability to engage the 'kill switch' implies that there is some means of remotely (read: outside the ISP control) engage the kill switch and remove those organizations from the network
All of those points (and many more I can include later) make this vastly different from CALEA, where a warrant compels ISP personnel to engage collection on an IP or target of interest and to turn over that capture to the authorities. That puts the control in the hands of the ISP - the kill switch would put the controls of ISP assets in the hands of the government.
So "conveniently", it may turn out that to implement the bill, ISPes would be required to install flexible remote controlled switches of various sorts. Meaning that the real impact of the bill would be the configuration of a remote-controlled censoring system. IE, a Great-Firewall-of-America "by accident" (yeah).
Let's say Americans are using Twitter or a foreign website to organize riots in a civil war, I don't think once the order is given and such a bill is in place, that the government has to ask nicely and force compliance. It will put all government computers and some very big tubes into DDOS'ing whoever is publishing something they don't want published at that very moment.
I dunno what happens if your provider happens to be someone like Hughes or DirectPC. That's one heck of a large subnet.
Remember that before cheering on the FCC to have authority over which ISPs are sufficiently 'neutral'. (And don't be surprised when 'neutral' gets redefined over time into 'compliant with the FCC's political biases'.)
Not to mention that even rudimentary steps - knocking out DNS or even just Google, Yahoo, and Bing - would put 95% of Americans effectively offline.
Metcalfe's Law works both ways - if you cut off even 3/4 of the internet, it's now only a square-root of a square-root as valuable as it used to be.
This is a bad idea and is trying to change how the Internet works. The Internet was designed to stay on even if there was a nuke attack. ISPs have been handeling DOS attacks for years successfully without a kill switch.