Slashdot Mirror
Broadband Bills Will Have To Increase To Pay For Snooper's Charter, MPs Warned (theguardian.com)
An anonymous reader writes that the UK's Science and Technology Select Committee has been told that ISPs will have huge problems implementing the so-called snooper's charter, and may be forced to raise their prices. The Guardian reports: "Consumers' broadband bills will have to go up if the investigatory powers bill is passed due to the "massive cost" of implementation, MPs have been warned. Internet service providers (ISP) told a Commons select committee that the legislation, commonly known as the snooper's charter, does not properly acknowledge the "sheer quantity" of data generated by a typical internet user, nor the basic difficulty of distinguishing between content and metadata. As a result, the cost of implementing plans to make ISPs store communications data for up to 12 months are likely to be far in excess of the £175m the government has budgeted for the task, said Matthew Hare, the chief executive of ISP Gigaclear."
Good. I'm delighted to hear about this. It's high time that the cost of outrageous government snooping programs are made to fall directly on the public who ultimately vote to support this nonsense.
Oh? You're ambivalent about mass GCHQ/NSA surveillance? OK. Well it'll cost you an extra £11 a month on your telephone bill. Oh you have a problem now?.
Most people will not care about an issue until they see it hit their pocket. Therefore, I say let it.
One of the insightful points made by the head of Gigaclear is that the line between metadata and data is pretty vague. For instance, who are you calling on Skype? "Obviously" metadata .... but if someone is added to a group call in the middle of it, then suddenly metadata might be being mixed in seamlessly with voice and video data. If you post a message to a website like Slashdot that has subject lines and bodies, is the subject line metadata? And if so, how does an ISP extract that and store it separately from the body?
The real cost of this scheme isn't even in the hardware, really, it's in paying large numbers of skilled people to develop a dizzying array of Wireshark filters to try and separate and index the metadata for every imaginable internet protocol.
Of course it's going to cost more. Every time the "snooper's charter" proposal came up with Labour and then the Coalition, the cost was placed at around £2bn at least. Even during the Coalition, it was estimated at around £2bn to do this. Nothing has really changed with the proposals, and yet the government thinks it's now going to cost £175m. I know storage costs are getting cheaper, but the amount of data generated is far more than it was when Blair and his cronies were trying to push a Stasi state on us.
If you post a message to a website like Slashdot that has subject lines and bodies, is the subject line metadata?
The scrutiny is the point. Blanket surveillance is shit for finding actual criminal / terrorist activity because the false positive rate means that your agents will all be tied up investigating bad leads forever.
As a tool to gain insight into a population and thus control over them, it's excellent.
And whle you're at it, itemize the bill.
Line rental mothly: 5 pounds
30 mbit/s package: 10 pounds
fee for us to snoop on you as legally required by the government: 10 pounds
If it costs more, pass the cost on to the customer and LET THEM KNOW.
SJW n. One who posts facts.
Exactly this. Finding "criminal/terrorist" activity can be a needle in a haystack endeavor at the best of times. However, mass surveillance just adds more hay to the stack under the notion that maybe perhaps you'll possibly be tossing in another needle or two. Of course, now you have to sift through 1,000 times the hay just to find one more needle.
Do they need to do surveillance? Sure. But it should be targeted and only undertaken after the proper warrants have been obtained. Is there the possibility that they won't pick up on some activity until it's too late? Yes, but the loss of liberty from mass surveillance isn't worth the tiny perceived increase in security that mass surveillance brings.
My sci-fi novel, Ghost Thief, is now available from Amazon.com.
Have you ever wondered how long it would take to record so much data that to read it would take the same amount of energy as it would take to boil all the water in the world's oceans?
Thankfully we have ZFS but how much data do they really intend to store? It is cheaper to just put everyone in prison and give them a free iphone. As long as tasty meals are provided there shouldn't be many complaints. I shotgun top bunk.
Urban legend has it that back in Old Days of the Revolution, the Chinese Communist Party billed the family of an executed criminal* for the cost of the bullet used to execute him.
There's some dispute to this, of course. It is hard to believe because it would be beyond the pale of decency, even to the extent it would be acknowledged by Communist revolutionaries, to bill you for the cost of their oppression.
But not, apparently, in Oceania.
*"criminal" often meant political opponent, not necessarily an actual criminal.
Welcome to the Panopticon. Used to be a prison, now it's your home.
Meanwhile the PM - Cameron - is promising affordable broadband for all
Do these politicians ever talk to each other ?
Did you perchance overlook the key weasel word "promising"?
I am sure that there are many other solipsists out there.
They've also been talking about broadband being some sort of fundamental right one minute, with ominous-sounding ideas about cutting people off for dubious IP-related reasons the next, and then moving government services that many people are legally required to use into on-line systems the day after that.
I'm pretty sure it's all just an elaborate episode of Yes Minister at this point.
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.
It seems highly unlikely that this is the interpretation the government is looking for. They've been quite explicit that they don't just want to know which communications channels you use, but also who you communicate with and the like.
The trouble is that, as many here will understand but I fear many in the government do not, there is no black and white distinction to be made based on some universal technical test to achieve the results the authorities say they want. Leaving aside the usual issues with encryption and reluctant foreign services, you're effectively talking about deep packet inspection in real time of many gigabits/second of network traffic, applying custom processing based on numerous specific protocols and/or service providers to each packet, and then recording the remaining payload after irrelevant parts are stripped.
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.
OK, say you take them at their word and they're just logging sites you visit (as in the domain). Have you ever looked at all the domains you 'visit' when you open a 'modern' web page?
What's to stop a random site from including an iframe or other call to http://dodgy-jihadi-site.com/ in their page? Does that get logged? If not, what's to stop a site from just being a wrapper page that lets you browse dodgy sites without triggering their metadata capture? What's the chances that loads of sites will put malicious img requests in for a 1x1 pixel from dodgy-site?
"Our metadata shows that on the X of Y, you visited 'dodgy-jihadi-site.com'"
"No I didn't, look, I just visited 'random-site.com', it must have pulled something in!"
But as they don't keep the full request 'dodgy-jihadi-site.com/images/1x1pixel.jpg', you have no defense.
This is a complete mess.
I frequently do R&D work in this kind of area and I am familiar with the in-the-trenches details here. It really isn't as simple as you're making out.
For example, you referred to using regular expressions for the decoders, but there are several details you're glossing over. The first is that you're presumably referring to application layer processing, but before you can do that you have to get hold of that application layer data and get it to something that can process it.
Just identifying which application protocol is in use and therefore which analysis tools are appropriate may not be straightforward. Consider the number of protocols that use some sort of control channel to establish connections but then send their data over some arbitrary UDP port, for example. You need all kinds of stateful analysis to do this reliably, and each case needs to be custom written.
Assuming you can isolate the application layer data, you then need to process it, at line rates, to extract the required metadata, which needs to be stored at line rates as well. The TCAM-based filtering built into a switch ASIC isn't going to implement any sort of regex matching, because that's not how a TCAM works. Compiling usefully detailed regexes to run on a sufficiently powerful FPGA is more plausible, but it's no trivial exercise either. (If it were, a large part of the network tools industry would not be the shape it is and some colleagues of mine would have retired very rich by now.) You can do some quite clever things bringing network traffic through a general purpose CPU, and you can build a device around it that could cope with a surprising amount of throughput using high-end but off-the-shelf components today and of course offers the best flexibility, but only if you can support the dramatically higher power, cooling and rack space requirements in your data centre.
It's true that without needing any application layer processing it would still be practical to record all the TCP connections that took place between two connected devices over your network with current hardware. With the co-operation of whoever operated the server-side of a communications network you could then reconcile the connections to figure out who was communicating when, at least up to IP address and port number. But in that case the valuable information is primarily what you're relying on the communication network to provide; what is recorded by the ISP doesn't really tell you anything very useful, certainly not on the level government spokespeople have been talking about when they say "metadata".
In short, extracting what you call "unique entropy" is straightforward, but it's also almost worthless without the real data set you care about to correlate with. You're still relying on some much more sophisticated deep packet analysis and/or the co-operation of at least one of the participants in the communication, and both of those things will normally need to be set up on a case-by-case basis.
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.