PDA

View Full Version : Free Will: The Internet Age Edition


Bloggingheads
02-18-2008, 09:03 AM

ogieogie
02-18-2008, 10:15 AM
Young people these days.

I swear.

Bloggin' Noggin
02-18-2008, 10:44 AM
I'm looking forward to some hot libertarian on libertarian action,
though I wish we could have worked the lovely Julian Sanchez into the mix:
http://bloggingheads.tv/diavlogs/8828?in=00:01:39&out=00:02:04

Did Will say that Tim was his tech guru?

Tyrrell McAllister
02-18-2008, 01:19 PM
Did Will say that Tim was his tech guru?

Maybe he said "TeX guru".

threep
02-18-2008, 01:36 PM
I don't claim to know what exactly the capabilities of our intelligence services are, and I'm sure that the whole echelon thing is overblown to some extent, but it seems to me that Will's rant about the overestimation of our intelligence capabilities is, well, wildly overestimated. The amount of collection of signal intelligence and our capability to process it are, most likely, pretty extensive. There's a difference between "hack the whole internet" cliches and more measured takes on what the NSA is (probably) capable of.

bjkeefe
02-18-2008, 03:10 PM
I don't claim to know what exactly the capabilities of our intelligence services are, and I'm sure that the whole echelon thing is overblown to some extent, but it seems to me that Will's rant about the overestimation of our intelligence capabilities is, well, wildly overestimated. The amount of collection of signal intelligence and our capability to process it are, most likely, pretty extensive. There's a difference between "hack the whole internet" cliches and more measured takes on what the NSA is (probably) capable of.

My sense, from a fairly long career in signal processing, is that Will's estimation is less wildly off than yours.

For one poignant example, consider 9/11. I think it's fairly well accepted that considerable intelligence had been gathered prior to that date, but had not been processed.

More generally, the NSA faces numerous challenges: increased use of fiber optic and coaxial cables for transmission (as opposed to radiated data); cheap or free strong encryption for cell phones and Internet-based transactions; onion routers, zombie PCs, uncountable free email and hosting options, and other Web-based ways to mask one's location; the language barrier (especially Arabic and other Middle Eastern languages); the problem of signal detection in voice data (as opposed to text processing), and the sheer volume of the data: of order 1 billion each cell phones and Internet connections.

And then there is always the fundamental limitation in any signal detection problem: how to balance false positives and and false negatives. If you believe in a "1% doctrine;" i.e., you assign a really high cost to missing a communication between two bad guys, then there is no choice but to accept a large number of false alarms. This means at least two things: you need lots of resources to check all the "detections," and innocent people are bound to be falsely accused.

Finally, there is the bureaucratic problem -- data sharing between agencies is known to be really bad. This is especially noteworthy when considering a connection that is half on US soil and half offshore.

I'm not saying the NSA and other government agencies are helpless. But they're really not supremely powerful, either, and it is certainly the case that asking Americans to give up even more privacy is not going to fix the problem. In fact, it's arguable that giving carte blanche eavesdropping permission could exacerbate many of the problems listed above. It's easy to kid yourself that you're gaining something if you're collecting ten times more data than you were before, but the key in this age of data at firehose rates is not to ask for ten more hose's worth; it is to be smart about what you want to focus on. Tim's point about us having a small database of repeated terrorist actions is especially relevant here -- in order to automate a search for signals, you have to have a good idea what you're looking for in the first place. We don't have that, not by a long shot.

cragger
02-18-2008, 08:10 PM
Before swinging the needle back too far toward "its no big problem", consider a few things.

The expansion in "legalized" surveilance is consistent with a general pattern of behavior and activity, all of which has fairly greatly increased the power of the executive. Anything that enhances a ruler's power just might have attractive uses beyond finding the "terrorist needle in a haystack". Not that I would suggest that anyone or any group in our government would ever be attracted to such an idea, nor that they would ever manifest anything beyond the utmost respect for both the letter and spirit of the law, as well as for the principles underlying the Constitution and the idea of democracy. As is well known, we're special in that respect.

Don't underestimate the resources that can be thrown at the surveilance problem. Throw in money by the billions year after year and you can get a pretty impressive amount of computing power and algorithm development. It will never stop each and every individual or small group, and it may well not stop nearly as much potential damage as other means attainable with less societal risk and for less cost, but there are certainly things that the approach will yield. If these techniques did not give information and advantage, commercial firms with vastly smaller resources and who have to actually use their own $ rather than simply taking them from the population would not engage in data mining. Don't underestimate the power of unlimited money and a decade or so of effort. After all, that got us to the moon.

I think you would find that the NSA can monitor any over-the-air broadcast in the world, and has been able to for some time. The telecom issue, which has been festering for years, has that same capability moving into the land-line digital world. In some ways it is getting easier technically despite the increase in volume. With analog voice, you had to go to the telephone office and clip onto the wires of an individual phone and have someone listen to it. With digital data, you sit at your desk with your feet up and let supercomputers parse the stream. Encryption is a potential privacy enabler, but the government has been trying very hard to prevent that, and to stop deployment of any system without government-accessable backdoors. Recall the Clipper chip?

Even though terrorist attacks in the US or Europe are pretty rare and represent a small proportion of people who meet a sudden and violent end, and though people reasonably suspected of connection with terrorist groups may be few, the ability to monitor all communications from those few people can be valuable in terms of locating an ever-expanding web of connections.

One can also look for patterns in communication usage. Someone who calls "suspect" areas of the world a lot, and visits "radical" websites regularly for example, might be a candidate for digging down deeper. You might tag a real terrorist through a connection with someone who would never pull a trigger themselves, but who might act as some sort of support enabler.

As you say, there will be a lot of false alarms in a net that sweeps widely. There was once a pretty widely held belief that in our system, "better that many guilty go free than that the government unjustly punish one innocent". That sentiment may not be totally absent yet, but you won't have to look far to find those who will argue against it. Flip it around and false alarms cease to be a problem. Waste of resources? When the deciders are playing with house money, do the losses really matter? Any attack that slips through any net is just another reason for a more powerful government no?

Now put on your paranoid hat. Its right over there, under that stack of books about the abuses of government power over the last half century or so. Just how handy might this sort of surveilance power be, for things other than ferreting out the next dozen A-holes with knives and spoons who plan to crash some planes?

It should be obvious, that once given the expanded govenment power and capability, it will not long be restricted to "terrorism". Who would want to argue that it could and should not also be used to go after any and all sorts of lawbreakers from sex offenders and killers to deadbeat dads and pot smokers, parking scofflaws and tax evaders? Nobody running for office for sure. These will be "special anti-terrorist powers" for only as long as it takes for any current resistance to fade away. Someone will be able to sit and sort job candidates based on how much they donate to an "acceptable" church vs. how much they spend on beer, or whether they have ever purchaced a sex toy. Its all just part of the data flow.

Further, suppose for example you had some sort of conviction that is really important that you and your like minded cohorts hold power. Nobody could imagine that someone just liked being in power, so pretend its your fanatical belief in your necessity, and the necessity that your views hold forth. Imagine further that you found a certain shallowness to the political system, and that you were willing to trash your opponents in any way to maintain office. Its a stretch I know, but try. If such a hopelessly hypothetical individual as you are imagining had access to all sorts of information based on pervasive surveilance and found that a potential opponent, call them Ruby Julie Yanni for example, was involved in some personal peccadillio, might that be useful in playing the scandal card and trumping any policy debate? Maybe a quiet phone call to a sympathetic reporter, blogger, or radio talk show blowhard?

The idea that surveilance is more powerful than you suggest, and might even in some imaginary world be put to some other handy use beyond catching the odd "terrorist"? Pure fantasy no doubt. As Frank said "it can't happen here". Not now, not with another hundred billion dollars and terraflops. Guardians of truth and reality will be slamming down their Pabst and grabbing keyboards to explain that fact even as I scuttle back to my tin foil hat. Got to block out the voices of Jefferson, Paine, and Mill!! Must be Will and all these libertarians on BHTV lately. Yeah, its their fault.

bjkeefe
02-18-2008, 11:33 PM
cragger:

Excellent response.

I want to start off, in my own defense, by saying I consciously avoided mentioning the potential abuses of increased eavesdropping because I wanted to focus on the points I was trying to make without seeming like I was motivated purely by BDS. I am glad that you jumped in to point out the potential dangers of the government watching the citizenry too much, though.

Some quibbles:

Don't underestimate the resources that can be thrown at the surveilance problem. Throw in money by the billions year after year and you can get a pretty impressive amount of computing power and algorithm development. It will never stop each and every individual or small group, and it may well not stop nearly as much potential damage as other means attainable with less societal risk and for less cost, but there are certainly things that the approach will yield. If these techniques did not give information and advantage, commercial firms with vastly smaller resources and who have to actually use their own $ rather than simply taking them from the population would not engage in data mining. Don't underestimate the power of unlimited money and a decade or so of effort. After all, that got us to the moon.

I don't underestimate what can be obtained with a blank check, but I do maintain that it's all too easy to overestimate what can be accomplished by throwing resources at a problem. You say that private companies do data mining, and therefore must perceive a benefit, but I'm not sure I buy this. For example, I have been a registered user of nytimes.com since shortly after they went live. I visit that site several times a day, I always stay logged in, I don't block cookies, I post numerous comments, write them email, participate in their surveys, and every other thing you could think of to expose myself to them, and as far as I can tell from their specific directed ads, they still -- after a decade -- have no clue about my interests.

I also don't accept the Apollo mission as a reasonable analogy. That was not (primarily) a signal detection problem. The trade-offs to be considered in putting humans on the Moon and returning them safely were transient; i.e., practically all limitations involved the current state of the art in rocketry and life support. Essentially, no problem was fundamentally insoluble. By contrast, the limitations of signal processing are inherent to the theory. A better analogy would be thinking about going to the Moon and pretending we didn't have to consider the force of gravity. That would be crazy, and it's equally crazy to think that false alarms will magically vanish with more disk space, faster CPUs, and an order of magnitude increase in capturing mostly irrelevant data.

Algorithms, I admit, represent more of a potential game-changer. But I still don't believe that anyone is going to come up with a way to filter voice or email or IM to the degree that blows the existing theory of signal processing out of the water.

I think you would find that the NSA can monitor any over-the-air broadcast in the world, and has been able to for some time. The telecom issue, which has been festering for years, has that same capability moving into the land-line digital world. In some ways it is getting easier technically despite the increase in volume. With analog voice, you had to go to the telephone office and clip onto the wires of an individual phone and have someone listen to it. With digital data, you sit at your desk with your feet up and let supercomputers parse the stream. Encryption is a potential privacy enabler, but the government has been trying very hard to prevent that, and to stop deployment of any system without government-accessable backdoors. Recall the Clipper chip?

I did acknowledge the capabilities of monitoring over-the-air transmissions, and I grant that, in some ways, the digital age makes monitoring wired transmissions easier. However, I think you're mistaken to think that backdoors and export restrictions are any sort of comfort. From what I understand about cryptography, there are practically unbreakable encryptions schemes readily available to anyone who wants them. I say "practically" in the strict sense -- maybe the NSA can decrypt something encoded with, say, the Blowfish or AES schemes, but they certainly can't do this in anything approaching real time. And these example encryption schemes don't require any special equipment -- an undergraduate in math could implement them from scratch, just from the literature. And importantly, don't forget the language barrier. Doesn't matter if you can decode the transmission if you don't speak Arabic or Farsi or Hindi, let alone any of the myriad of dialects of these.

One can also look for patterns in communication usage. Someone who calls "suspect" areas of the world a lot, and visits "radical" websites regularly for example, might be a candidate for digging down deeper. You might tag a real terrorist through a connection with someone who would never pull a trigger themselves, but who might act as some sort of support enabler.

Yes. On the other hand, you might also nab someone like me, who likes to travel to non-touristy places, and who often looks at weird sites on the Web purely out of curiosity.

As I said at the start, you and I seem to be pretty much on the same page about the dangers inherent in the potential for abuse with such systems. I guess where we differ is that you think there could be some use gotten from enhanced eavesdropping and worry that the power might be extended. I think that the usefulness obtained through enhanced eavesdropping in detecting terrorist plots is limited, and (in addition to the abuses) will likely be end up just being "repurposed," in a retroactive attempt to justify all the money spent (cf. missile defense).

Wonderment
02-19-2008, 12:05 AM
The big problem seems to me to be the chilling effect the technology can have on free speech. For example, just knowing what the government can do, in principle, already may be inhibiting a lot of people who are not necessarily paranoic.

The IRS already inhibits some criticism of the government, and the Nixon administration's use of audits to go after political enemies was one of the major privacy abuses of the last few decades. Of course, a lot has been done to protect privacy and prevent abuses since Nixon, but that's all the more reason to be concerned about the Bush administration's domestic spying abuses.

Cragger's observations seem reasonable to me. I don't think there's much we can do about vanishing privacy. We have to try, of course, but it's a losing battle.

Cameras and recording devices are everywhere and may eventually become almost literally ubiquitous. Traces of our identity and behaviors are all over the Internet. It will be increasingly difficult to get off the grid in the future, and we will welcome invasion of privacy in a variety of ways, independent of our national security concerns.

For example, people are likely to love a tiny barcode on their pinky that will plug into their health history from pre-natal to post-mortem. People already love the ability to go on the Internet and see what sex offenders live in their neighborhood. Drug-testing for employment purposes is fine with most people, as long as you don't test for the drugs they use (alcohol, for example).

bjkeefe
02-19-2008, 12:47 AM
Wonderment:

You'll get no disagreement from me on the modern age and its encroachment on our privacy. In response, I ask, so why do we need to give the government even more ability to intrude?

I'm not sure I agree with the chilling effect completely. It's there, to be sure, especially among the well-paid members of the MSM. On the other hand, it's possible for an average individual to give vent to harsher criticism, and with potentially larger audiences and better chances at anonymity, than it was in any time before.

Wonderment
02-19-2008, 01:34 AM
You may be right. In some ways we have more freedom of expression than we ever did. It's hard to predict though. The other day I decided I needed to give some money to help Iraqi refugees. I started searching on the Internet, but I immediately felt a little rush of paranoia: what if I donate through a Muslim charity? What will that trigger? I ended up donating through

http://www.unhcr.org/

but my point is that the "chill" is subtle.

However, the chilling effect may have peaked in regard to public figures.

A minor scandal like renting porn movies, smoking weed or hiring an undocumented nanny is no longer enough to ruin a supreme court or attorney general nomination. Nowadays you may need a serious secret narrative like the one Mickey Kaus maliciously spread about John Edwards or the Larry Craig affair: anti-gay Senator turns out to be cruising airport bathrooms.

Gulliani's candidacy may have been the first of the post-privacy era.

His position seemed to be "Yes, I've done all kinds of shady, embarrassing shit that's contrary to your/our values, and more may come out, but this shouldn't matter to voters because I will perform like an automaton in office. Whatever I did in my private life, I can be counted on to deliver right-wing extremist judges, support torture and wage perpetual war."

Ah, Brave New World...

bjkeefe
02-19-2008, 01:49 AM
Wonderment:

There's another example in support of your later points: There actually is a bit of a meme floating around that wants to criticize Obama for not doing enough drugs. And not from the stoners, either.

Sorry for the lack of link -- I did not think until now that such hilarity merited a bookmark. But it definitely is out there.

Joel_Cairo
02-19-2008, 05:23 PM
Will-

It's time for a haircut; you're getting a little bushy on the sides there.

Tao Jones
02-20-2008, 03:11 AM
One thought on advertising as "spinach." Isn't that how television advertisement has worked for decades?

Bloggin' Noggin
02-20-2008, 11:58 AM
This was a terrific episode of Free Will. It may not have been more informative than Will's other, terrific interviews, but it was Will's wittiest performance yet (apart from his first appearance on BHtv with Ezra).

Libertarian-on-libertarian action, despite my initial hopes, appears not to be very steamy. But it does seem effervescent.

I personally would welcome advertising on BHtv if it would free Will and the other amazing diavloggers on BHtv from their "slave labor". I don't want to see a rebellion and a mass exodus of all this talent.

Can't Mickey get a small fee for his Coke product placement? And Will's Kindle product placement made at least one sale of $399 for Amazon.

cragger
02-20-2008, 06:51 PM
I think we are in violent agreement overall. A couple of points though, since we are almost certainly going down this road of ever-increasing government power and intrusiveness.

On encryption, yes there are some "pretty good" as they say, available methods that so far as we know aren't compromised by the "three letter agencies". They are in general breakable only through brute force techniques beyond the capacity of individuals or even modest corporations. Of course, the thing about brute force is that if it isn't working, you just aren't using enough. US govt. with multi-trillion dollar budget = heap big brute force.

Thats about 99% beside the point of course. The catch with encryption is you need to use it. At both ends. How much of your communication involves a strong encryption? What percentage of that in the US? Since most folks don't use it, a communication from an individual, or between individuals, that has strong encryption might be part of a signature that triggers increased scrutiny.

To use your analogy, it all depends on what sort of signature, or correlation function, can be developed for "subjects of interest". The false alarm rate will then depend on how many "signals" you can integrate, and what that correlation looks like.

Yeah, the process will never detect every valid signal, and the false detection rate will be significant, but those building and operating the receiver seem quite happy to accept that.

bjkeefe
02-20-2008, 08:04 PM
The catch with encryption is you need to use it. At both ends. How much of your communication involves a strong encryption? What percentage of that in the US? Since most folks don't use it, a communication from an individual, or between individuals, that has strong encryption might be part of a signature that triggers increased scrutiny.

[...]

Yeah, the process will never detect every valid signal, and the false detection rate will be significant, but those building and operating the receiver seem quite happy to accept that.

Two good points.

I keep thinking about moving to all encrypted email, all the time, and urging my friends to do the same, just to protest the increasing surveillance.

And, sadly, I agree even more strongly that those in power won't care about a high rate of false alarms, or at least, won't care enough to realize the idiocy of trying to listen in on everybody.

themightypuck
02-21-2008, 02:52 AM
On Patents. All of them.

I suspect they only make sense as an incentive where the barriers to entry are high and the cost of a copy is cheap and fast to produce. Pharmaceuticals makes sense partially due to (not unreasonable) government regulation. It is an untestable hypothesis, but I figure if patents never happened our civilization would be further advanced.

jstrummer
02-21-2008, 02:57 PM
I realize this is a non-legal discussion of patents, but patents are not simply, as Tim says, "for physical objects or physical processes." State Street Bank, a case from the 90s, allows the patenting of business processes so long as those processes yield concrete results (however fleetingly).

That's why you could patent a non-physical process like a business process - 1-click.

In addition, is the test for non-obviousness really "

Also, most software is actually copyrighted, and not patented.

jstrummer
02-21-2008, 02:58 PM
I realize this is a non-legal discussion of patents, but patents are not simply, as Tim says, "for physical objects or physical processes." State Street Bank, a case from the 90s, allows the patenting of business processes so long as those processes yield concrete results (however fleetingly).

That's why you could patent a non-physical process like a business process - 1-click.

Also, most software is actually copyrighted, and not patented.