January 5, 2010

Now that Obama has approved airport profiling ...

... of travelers from 14 countries (13 of them heavily Muslim), will progressives retract all the dumb arguments they've made over the years about how profiling can't even work in theory?

Probably not.

The New York Times hosts a debate over profiling:

The Obama administration has announced that it will subject citizens of 14 countries, including Pakistan, Saudi Arabia and Yemen, to intensive screening when flying to the United States (the rule will also apply to those passing through those countries). This means treating people differently depending on where they come from or what passports they hold.

Does it make sense to concentrate security efforts on more limited populations — through profiling, behavioral or otherwise? Is profiling effective, compared to other strategies?

The first contributor says:
Bruce Schneier is a security technologist and author of several books on computer security, including “Beyond Fear: Thinking Sensibly About Security in an Uncertain World.”

Terrorists can figure out how to beat any profiling system.

There are two kinds of profiling. There’s behavioral profiling based on how someone acts, and there’s automatic profiling based on name, nationality, method of ticket purchase, and so on. The first one can be effective, but is very hard to do right. The second one makes us all less safe. The problem with automatic profiling is that it doesn’t work.

Terrorists don’t fit a profile and cannot be plucked out of crowds by computers. They’re European, Asian, African, Hispanic, and Middle Eastern, male and female, young and old. Umar Farouk Abdul Mutallab was Nigerian. Richard Reid, the shoe bomber, was British with a Jamaican father. Germaine Lindsay, one of the 7/7 London bombers, was Afro-Caribbean. Dirty bomb suspect Jose Padilla was Hispanic-American. The 2002 Bali terrorists were Indonesian. Timothy McVeigh was a white American. So was the Unabomber. The Chechen terrorists who blew up two Russian planes in 2004 were female. Palestinian terrorists routinely recruit “clean” suicide bombers, and have used unsuspecting Westerners as bomb carriers.

In reality, as sportswriter Damon Runyon said, "The race is not always to the swift, nor the battle to the strong, but that's the way to bet."
Without an accurate profile, the system can be statistically demonstrated to be no more effective than random screening.

Actually, the link says the opposite, as I'll show below.
And, even worse, profiling creates two paths through security: one with less scrutiny and one with more. And once you do that, you invite the terrorists to take the path with less scrutiny. That is, a terrorist group can safely probe any profiling system and figure out how to beat the profile. And once they do, they’re going to get through airport security with the minimum level of screening every time.

Sure, as long as Al-Qaeda can recruit Mexican grandmothers to be suicide bombers as readily as it can recruit young men with Muslim names.

As counterintuitive as it may seem, we’re all more secure when we randomly select people for secondary screening — even if it means occasionally screening wheelchair-bound grandmothers and innocent looking children. And, as an added bonus, it doesn’t needlessly anger the ethnic groups we need on our side if we’re going to be more secure against terrorism.

A recurrent theme of mine is how the demand for denial of average IQ differences spills into seemingly unrelated issues, like airline security, causing widespread intellectual stultification. The modern liberal mind thinks in black-and-white Manichean terms, rendering it unarmed for dealing with a probabilistic universe.

It's hard to deal with liberal arguments because they tend to be so Gladwellian in their mental rigidity. Here we are, more than eight years after 9/11, and this "expert" picked by the NYT for his wisdom can't imagine any profiling system smarter than he is.

Schneier seems to be assuming that profiling means that 100% of attention would be devoted to people in category X and 0% to people in category Y. The weird thing is, that's common among progressives. They really just don't get it. The conventional wisdom is a form of unilateral cognitive disarmament.

He's like a pitching coach who tells a baseball pitcher, "Your fastball is above average, your slider average, and your change-up below average, but if you only throw your fastball, they'll expect it, so you should choose your pitches randomly, throwing one-third of each."

Obviously, when stated in those terms, it's easy to see the fallacy: there are superior methodologies in-between all fastballs and total randomness. If your fastball is relatively more effective than your other pitches, you want to throw relatively more fastballs. But you still want to "mix 'em up," as every pitching coach from Babe Ruth League onward as told pitchers.

Why can't Americans be as smart about public policy as they are about sports?

Thus, if you read the article Schneier links to behind his phrase "statistically demonstrated," you'll find it's merely a debunking of a braindead "100% fastballs" profiling method:

Press then examines the effect of what he terms a strong profiling strategy, one in which a limited set of screening resources is deployed solely based the risk probabilities identified through profiling. It turns out that this also works poorly as the population size goes up. "The reason that this strong profiling strategy is inefficient," Press writes, "is that, on average, it keeps retesting the same innocent individuals who happen to have large pj [risk profile match] values."

The very next paragraph of the article linked to by Schneier explains that non-braindead profiling is the best method:

According to Press, the solution is something that's widely recognized by the statistics community: identify individuals for robust screening based on the square root of their risk value. That gives the profile some weight, but distributes the screening much more broadly through the population, and uses limited resources more effectively. It's so widely used in mathematical circles that Press concludes his paper by writing, "It seems peculiar that the method is not better known."

Peculiar, indeed. But as Napoleon supposedly said, "Never ascribe to malice that which can be explained by incompetence."

My published articles are archived at iSteve.com -- Steve Sailer

44 comments:

Anonymous said...

Christian Ganczarski and Mustafa Setmariam Nasr (who carried a Spanish passport for most of his career) are worth thinking about here. They didn't look stereotypically Muslim, yet both of them were pretty important figures in the jihadist network. (Ganczarski was an actual al-Qaida member-accounts vary on whether Nasr was.) Why don't you mention them, Steve?

Schneier is arguing against racial profiling, you are arguing for it-and you're also misrepresenting the argument from Press.

Press then examines the effect of what he terms a strong profiling strategy, one in which a limited set of screening resources is deployed solely based the risk probabilities identified through profiling. It turns out that this also works poorly as the population size goes up. "The reason that this strong profiling strategy is inefficient," Press writes, "is that, on average, it keeps retesting the same innocent individuals who happen to have large pj [risk profile match] values."

According to Press, the solution is something that's widely recognized by the statistics community: identify individuals for robust screening based on the square root of their risk value. That gives the profile some weight, but distributes the screening much more broadly through the population, and uses limited resources more effectively. It's so widely used in mathematical circles that Press concludes his paper by writing, "It seems peculiar that the method is not better known."


(Note for Press: your system isn't something that can be reduced to a sound bite, nor does it justify panicky introductions of methods that would have been rejected by Yuri Andropov, and it would actually reduce harassment of air travelers-which means that it would require our airport security goons to change their world view, so forget that!)

Steve, you're also ignoring the fact that al-Qaida has been able to recruit converts-in addition to Ganczrski, John Walker Lindh and Adam Gadahn come to mind, too. If you want to get slightly obscure, information about Muriel de Guage isn't too hard to find. Lindh, Gadahn and de Guage were all wasted as assets, but I wouldn't bet money on al-Qaida being stupid indefinitely.

The really sad thing about this is that you long ago identified the best methods for reducing the risk of terrorist attacks on the US: stop the "invade the world" policies and quit giving Israel a blank check to do whatever they feel like.

And just to forestall the people who will come on here and engage in the usual fetishization of Israel's airport security, I'll point out that they had a spectacular failure of their profiling system in the 1970s. That's without getting into how much their system depends on Israel being a small country whose visitors are mostly Jews.

rightsaidfred said...

The Chechen terrorists who blew up two Russian planes in 2004 were female.

The "screen every passenger to the max" mentality assumes unlimited resources for security. As soon as we realize we have limited resources for security and that terrorists have limited resources for recruitment, statistical methods begin to make sense.

If women are 90% less likely to be terrorists than men, we can reduce screening resources dedicated to women by 50%, direct those resources toward the men segment, and be safer in the long run. If Al Queda wastes resources recruiting Mexican grandmothers or middle aged white women from Wisconsin, that is to our benefit.

l said...

[b]...you long ago identified the best methods for reducing the risk of terrorist attacks on the US: stop the "invade the world" policies and quit giving Israel a blank check to do whatever they feel like.
[/b]

The problem with that thinking is that it's "antisemitic". That's what Ron Paul found out a week ago on the Larry King Show.

Next?

Truth(er) said...

This is funny because "computer security" has, in one of its arsenal of defenses, readily available methods for dealing with certain threats.

For example, lets say a hostile server is sending a denial of service attack from, I don't know, Saudi Arabia. One method of dealing with this server is prophylactic: block the IP address of the offending machine. Sure, there are workarounds...but that would certainly be a first step.

Nobody would suggest that the existence of a possible workaround means you let the server be.

Bruce Schneier's computer analogy would be the equivalent of saying that blocking random IP addresses would be more effective against DDSA then targeting the offending system.


In fact, computer security experts make all kinds of suggestions that they readily admit have workarounds: patching systems; making sure anti-virus is up to date; changing passwords periodically, etc. All of these have workarounds, and yet no one suggests you don't use them.

And Schneier's list of terrorist candidates all have on thing in common: they are almost all non-white.

frost said...

'Napoleon supposedly said, "Never ascribe to malice ..."'

Wildly off topic but does anyone know who this quote should actually be attributed to? This is usually referred to as Hanlon's Razor, a gem of wisdom from the Unix fortune file. I've heard many other attributions but the most appealing, to me, was to some 19th century British politician, possibly Gladstone, that I read in an airline magazine back in the 90s. It does have the flavor of a Parliamentary put-down to it.

Anonymous said...

Is there anyone else from Palantir in these comments?

I mean, look, off the record OBVIOUSLY "profiling" is the only way to actually stop crimes and terrorist attacks. That is what every single statistically informed national security analyst knows.

We just can't say anything about it because of the great and the good!

I mean for chrissake. Nationality is just the first and relatively the most PC variables to include.

You then include things like:

* gender
* religion
* race
* past travel history
* past purchases
* past phone calls
* past emails + electronic fingerprint
* and all the same features on your social network

At that point it's game over. Anyone who has worked at Palantir will tell you that you can get pretty darn good accuracy on identifying the profile of a terrorist.

In technical terms it is basically an imbalanced logistic regression problem with several hundred positive examples (i.e. all the jihadis we have apprehended over the past years) and millions upon millions of negative examples.

See Owen 2007 for an entree to this literature:

http://sci2s.ugr.es/keel/pdf/specific/articulo/owen2007.pdf


Anyway, Schneier intentionally fuzzes the issue. All those guys are Muslims. Hence they all have something in common. Hence that coefficient is highly nonzero. F'in DUH, right? You don't need to have a Phd in Statistics to know that a PhD in Statistics could identify a terrorist given some data, a cup of coffee and a command line.

dearieme said...

If I ran an intelligence service that wanted to penetrate Muslim terror groups, I'd welcome a screening system that encouraged them to try to recruit middle-aged white women from Alabama.

Anonymous said...

Steve didn't misrepresent the argument from Press. He said shake it up, like Press said: "square root of their risk value," in which race would of course be a weighted element.

Schneier is a joke. Consider. "The problem with automatic profiling is that it doesn't work. [It] makes us all less safe." Has it been tried? Less safe compared to an absolutely random approach, i.e. no approach at all? "We're all more secure when we randomly select people for secondary screening — even if it means occasionally screening wheelchair-bound grandmothers and innocent looking children." Why? No answer. But what directly follows is probably the real answer. "[A]s an added bonus, it doesn't needlessly anger the ethnic groups we need on our side if we're going to be more secure against terrorism." Yup, we're afraid of hurting the feelings of (a small sub-subsect of) NAMs. Discrimination! Racism! The ultimate evil! Schneier to the rescue! Hey, I have an idea. Instead of our being afraid of hurting NAMs' feelings, why not have it so that these NAMs are afraid of hurting our feelings?

Accordingly to Schneier, our (and only our) racism contributes to terrorism, racism consisting of nothing more indicative than finding risible the following:

"Terrorists don't fit a profile and cannot be plucked out of crowds by computers [computers?]. They're European, Asian, African, Hispanic, and Middle Eastern, male and female, young and old."

Pardon me while I laugh my ass off. Where are they all these things? In his mind? If in the first days of Jackie Robinson's playing in the major leagues, someone had declared: "We can't generalize about the racial composition of baseball. Heck, a baseball player could be any race, so any guess about the average baseball player's race would be worthless," everyone would have shrugged at the speaker's idiocy. But all it takes today is a mention of "Timothy McVeigh" and "Unibomber" (who didn't fly) and the moronia think Schneier has a point.

The OVERWHELMING MAJORITY of terrorists are Middle Eastern types, not "wheelchair-bound grandmothers and innocent looking children."

Ask a group of Las Vegas bookies which one is plotting to interfere with their flight: Abdul from UAE or a white 14-year-old girl from Nebraska in a miniskirt. So we can't be smarter than a bookie, because it might hurt a non-white's feelings and we "need them on our side"?

But give Schneier his props. He's fightin' white racism all the way to doom. Mazel tov, you crazy diamond!

Old Rebel said...

I think it's telling that both Iraq and Afghanistan are included in the list.

Didn't Bush invade those countries to eliminate the terror threat in both?

Jason said...

Schneier is in no way incompetent. He is very familiar with probabilistic approaches, in particular, Bayesian classifiers. The same technology that is used to filter your email could easily be applied to selecting airline passengers for additional screening.

Feed it any number of characteristics, each with a statistical weight calculated by looking at the corpus of historical terror attacks. Sex, race, country of origin, religion, marital status, favorite music, choice of pet, presence on watch lists, whether their passport's ever been stamped "Pakistan", amount of checked luggage, whether or not the ticket is round-trip, and on and on. Some of these weights will be high, some low. Sometimes seemingly irrelevant characteristics turn out to be rather good predictors once you crunch the numbers. Combinations of multiple low-weighted factors can also be higher than single high-weighted factors, so you don't have to worry about a terrorist being overlooked because of the lack of a stereotypically Muslim name.

Is it perfect? As anyone who's ever received spam in their inbox knows, no. But it works very, very well. And in this application, it would only be a first level filter, used to more efficiently direct human efforts.

Anonymous said...

Terrorists don’t fit a profile and cannot be plucked out of crowds by computers. They’re European, Asian, African, Hispanic, and Middle Eastern, male and female, young and old.


Actually, Schneier is correct, it's just that he's a time traveller from the future when the US and China are at war, and China has bred a strain of females for terrorism who are not afraid of exploding and who believe that they will go to a reward that includes 72 virgins (or is it grapes, can't remember) after a glorious death for the fatherland.

Anonymous said...

One problem with profiling and other such schemes is that however good and sensible they may sound, you've got joe average security cop implementing it. One time coming back from Puerto Vallarta my luggage got searched. The agent looked at the three books I had. Two were innocuous, the third was about Khomeinism in Iran. The agent seemed alarmed, gazed at it over and over, then went scurrying off to see a supervisor. Apparently, the average dope thinks something like that is a how-to book, not merely something to be read to keep informed. Movie star biographies are on the approved list, I suppose. Since I'm a white male with an ethnic sounding last name you would think they could make some differentiations there. But they can't. Nowadays they always seem to profile me as some money launderer, questioning and searching me. All I'm doing is taking a vacation. Yes, pay attention to the Muslims, that's 90% right there. But, no system is going to work perfectly so some will slip in. That's the reality we have to accept. The security people implementing it all are the usual gum-snapping American dolts who worship football, so don't expect miracles.

greenrivervalleyman said...

I think we've found a winner in the "Find-a-Worthier-Liberal-Punching-Bag-For-Sailer-Than-That-Nitwit-Malcolm-Gladwell". Schneier is actually a smart cookie in his area of formal training- mathematics and cryptography. His submission to a NIST-sponsored competition to find the next generation hash function (one of the basic tools in modern digital security) made it to the second round and could prove the eventual winner.

Yet Schneier is the typical semi-autistic nerd who cannot fathom that the real world does not operate according to the same laws of infinite elasticity as his statistical models. In cryptography I can keep ramming any series of bits I want into the maw of the digital gatekeeper in the hopes of finding a weak spot. In the real-world I do not have infinite resources to recruit, train, and deliver suicide bombers. If I'm running an Internet book store I can be sure hackers from across the world will eventually try breaking in to steal my customer's credit card #'s. In the real-world I keep my valuables in a Staples-bought mini-safe and don't worry that someday the Russian mob will send safe-crackers to empty its contents.

Sorry, Bruce, but the security measures invented centuries ago by only slightly above-average intellects and field-tested over that same length of time (such as locks, and armed sentries, and steel vaults) have proven much more resilient in "meat-space" than the digital counter-measures invented over the last 40 years for cyberspace.

Terrorists don’t fit a profile and cannot be plucked out of crowds by computers. They’re European, Asian, African, Hispanic, and Middle Eastern, male and female, young and old.


Of course they do. Even in Schenier's own cherry-picked list the overwhelming majority are Muslims, either born or bred after a stint in the joint or some other equally calamitous downscale life event. But finding and synthesizing such patterns, what with all the thousands of dependent and independent variables, would require computing resources greater than any heretofore known to man! Why, you'd need something with the raw processing power of an Irish police detective, and who has the money for more than a handful of those!

Legaleye said...

There is also the 14th amendment problem. If the Federal government mandates TSA to give additional screening to Arabs, Middle Easterners or Muslims, it will probably end up being challenged on constitutional grounds.

Deckin said...

Steve, you're bang on with the comments about Manichean dualism and the inability to deal with probabilistic thinking as the core of progressive cognitive pathology. It suffuses every aspect of every debate with them, from environmental nuttery to the refusal to acknowledge even the possibility of HBD.

As for the first anonymous, John Walker Lindh and his ilk are the dogs that didn't bark. That's attributed to Al Qaeda's incompetence, but why not to the fact that it's hard to get kids from Marin county to blow themselves up on board of crowded jetliners?

Anonymous said...

"Never ascribe to malice that which can be explained by incompetence."

And never ascribe to incompetence that which can be explained by the nihilists' desire for the terrorists to SUCCEED.

Granted, it can be difficult to distinguish nihilism from incompetence, if for no other reason than the two motivations both ending in the same outcome [although, in fairness to the incompetent, it's possible that they could randomly & unintentionally wander their way into an unexpectedly successful outcome - the same can't be said for the nihilists].

But as a mature, rational, sober, legally-emancipated adult, at some point you have a duty to ask yourself: WHY?

Why does the same thing keep happening over and over and over again?

And the answer is that there are very powerful people [like that raving lunatic, Bruce Schneier, and his sponsors at the NY Times] who WANT it to happen, and, in fact, they are so powerful that their philosophy is the dominant intellectual force in all of the upper echelons of our society.

bob said...

the question is what is one allowed to condition one's probability estimates on. Currently we are legally obliged to throw away information on crime rates, IQ, etc .... before even getting to the airport. The throwing away of information implies that it's free or we are too rich.

liamascorcaigh said...

As Polonius might say "That's good! 'Unilateral cognitive disarmament' is good."

Not as good as "mobléd queen", maybe, but, hey, you're a Steve not a [Wil]liam.

short white coat said...

Profiling is a tricky topic because of PC and the way American history is remembered. Racial discrimination and bigotry are considered the great sins of America: black slavery, Japanese internment, anti-Chinese immigration policies, segregation in the south, and even anti-communism--remembered by many Jews as a kind of antisemitic pogrom.

Also, certain minority groups are very powerful and threatening in this country. White people are afraid of black violence and Jewish money/media. For blacks, profiling means cops picking on 'suspicious' looking brothers. For white and Jewish liberals, black issues & concerns are of paramount importance if only because blacks vote as a bloc for the Democrats. So, even as white and Jewish liberals are afraid of black crime and live in mostly white neighborhoods, they make a lot of sanctimonious noise about the evils of profiling. And, Hispanics have jumped on the anti-profiling bandwagon since so many of them are illegal and don't wanna be 'harassed' by white cops.
And, gays don't like profiling since being labeled 'gay' in the past could mean getting fired, not promoted, or no apartment lease.

So, profiling means different things to different people. For most law-abiding patriotic whites, profiling is no big deal since they themselves won't be affected. Profiling will mean authorities paying more attention to potential muggers, illegal aliens, or foreign spies--mostly non-whites. Even though whites are the majority, they've been tainted with the sin of 'racism', and thus they cannot make a strong moral case for profiling since the progressive media will spin it as 'racist'. Therefore, on the moral level, the non-whites have an upperhand when it comes to profiling--largely due to American history and white abuses of power in the past. (Of course, white conservatives are kinda schizo on this issue. Though many will insist on the need for profiling of black/brown criminals and Musliim terrorists, they insist that we should harbor NO suspicion regarding the Jewish community though it is most dangerous to white American interests. ACLU, SPLC, News Media, Open Borders, etc are largely funded and run by liberal Jews.)

Of course, Muslims are a different story. Most Americans feel no historical guilt regarding the Muslims--though US aided in the creation of Israel which dispossessed the Palestinians who have a right to be mad as hell.
(It's somewhat different in parts of Europe like UK and France because they'd once imposed imperialist rule over areas like Algeria, Egypt, and Pakistan.)
Also, lots of Muslims do seem pretty crazy or stupid. So, most Americans are not be averse to profiling Muslims.
The problem is if profiling of Muslims is accepted, it may then justifiy profiling of non-Muslims. This is what AIPAC spies of Israel, black gangsta types and their political representatives, Hispanic Illegals and Raza, and a whole bunch of HIV gays are afraid of.

albertosaurus said...

There is a Darwinian perspective on profiling. Imagine two creatures - one of whom has developed a system to assess potential danger based on a probabilistic weighting of various indicators and another creature who reacts to all potential threats equally. We don't seem to find many of the latter kind of creature, do we?

Indeed Marsh's Rule of Progressive Encephalization (intelligence tends to increase over time) may be driven by the comparative advantage of bigger brains to weigh threats better.

There is a lot of evidence and theory (Egon Brunswick) to suggest that people or indeed all vertebrates weigh probabilities. Vertebrate nervous systems also adjust those weights based on experience.

My little dog won't attack raccoons anymore - not after that first time. He wakes me up and watches me attack them. How did he get so smart? - about a billion years of evolution.

Profiling isn't new. In fact it's natural and inevitable given a human nervous system. The liberal mistake is to think it is edifying to overcome your instincts.

liamascorcaigh said...

@Anonymous

The fact that non-Middle Eastern pro-Jihadi outliers could penetrate security is a non-argument. No security system is infallible or ever will be. But a reality based, statistically governed approach would be a significant improvement on the present dispensation in actually catching suicide operative before they boarded. It would also severely attenuate AQ's efficiency by forcing them to recruit outside their natural constituency.

Furthermore, while the likes of Lindh et alii might be zealous enough to fight against the West, they would be more unlikely to engage in suicide tactics per se. After all even among the most devout and acculturated - Muslims by birth and upbringing - only a small proportion are prepared psychologically to kill themselves for the cause.

What Steve is suggesting would greatly reduce the pool of recruits and that pool in itself would not be optimal.

van doren said...

Whether profiling works or not, let's drop the REALLY STUPID stuff, like checking old Polish-American or Chinese-American ladies at airports.

Polymath said...

Here's another example of willful obtuseness:

http://blog.nj.com/njv_bob_braun/2010/01/princeton_university_researche.html

Before misdirecting billions more Federal dollars it might help to recognize the dependence of the proposed course of action on the assumption "all ethnic groups have identical distributions of every propensity".

DCThrowback said...

I always the "malice...incompetence" quote was from Robert Heinlein. Or Robert Hanlon, of Scranton, PA.

http://en.wikipedia.org/wiki/Hanlon's_razor

Either way, it's a keeper.

ben g said...

Steve,

It seems like you're only looking at one side of the ledger. Does it concern you that profiling would adversely affect people who are inaccurately suspected because of their race?

It's an easy emotional argument to say that people dying outweighs such a concern, but if you stop and think about it democracy is about finding a balance between such concerns and the safety of the nation. For example, I'm sure we could lower the number of violent deaths by allowing police to further invade peoples' privacy, etc. but most people don't support that.

gumbic jones said...

If not profiling, we should at least do some filtering. It's like when you PAN for gold in a riverstream, you don't so much look for the gold as wash away the lighter elements like dirt. At the end of the day, you end up with some gold powder.

So, instead of looking FOR terrorists, let's filter out those least likely to be terrorists. I say let's have panning than profiling.

Carbon Lake said...

If profiling is so bad, how come we allow auto insurance companies to charge higher rates for young males and old folks? Isn't it because young males are more likely to be reckless and old folks are more likely to have weak eyes and slow responses?

If we allow auto insurance companies to effectively profile people in the way they charge premiums, why can't we profile young Muslim males at airports, dangerous-looking young black males in the streets, and odd-looking Mexicans in big cities who are likely to be illegal?

truth said...

Profiling Muslims is good policy.

Thank God that Muslims are such a small minority in the USA.

Otherwise, profiling would not be acceptable.

Americans don't really like Muslims so I don't expect too much outcry about profiling Muslims (aside from the usual suspects).

Svigor said...

If the Federal government mandates TSA to give additional screening to Arabs, Middle Easterners or Muslims, it will probably end up being challenged on constitutional grounds.

Well, if it mandated that everyone submit to the same "how well does your profile match that of criminals?" criterion, then they might have an easier time with the black robes. At least they'd increase their chances of forcing a "judicial activism" foul.

Whiskey said...

"Jack Dunphy" an LAPD officer who writes pseudonomously has an article in Pajamas Media detailing how profiling works.

Scheiner is like all computer software experts, dismissive of the human factor of expertise. No computer expert system (I should know, I've been responsible for building a few) is as accurate and "deep" in profiling as an experienced human. Cops and other experienced, well trained individuals can quickly assess in seconds the totality of a man's risk. They do this by noting body posture, various racial and religious backgrounds, attention levels, gait, and so on.

The Israelis who use this system do not simply pat down every Muslim. Only those who are behaving like threats, instead of businessmen pre-occupied with getting to the bathroom and a cup of coffee before boarding, or wondering about their meeting, or a thousand other things.

This system requires human expertise, which does not scale like a machine system, needs expensive and time consuming training, and so on. It is however SUPERB in catching prospective bad guys, because it catches far more accurately those who don't "look Muslim" but do have indications of threats.

Said: Russia, Denmark, Sweden, the Pope, China, all of which are quite hostile to both Israel and Jews in general, and do not "invade" Muslim nations, face constant Jihadist terror as do MUSLIM NATIONS THEMSELVES, including Pakistan (which had about 100 people blown up during a volleyball game over Shia-Sunni differences). If Sweden gets terrorist threats over drawings by some obscure artist or the Pope over something he said about ecumenical outreach ten years ago, the prospect of the US EVER avoiding Muslim Jihadist terror is ZERO.

Constant, global Muslim Jihadist terror is a "feature" of the globalized trade that brings you cheap Chinese sneakers and flatscreen TVs from Korea. Everything has its price and cheap stuff is no exception.
----------------
Schneier does not like the fact that a highly trained human system is more effective than a computerized one. One of the facts of life is accepting the limits of any machine system. Threat analysis can only realistically be done by human beings because we have evolved over millions of years to distinguish our fellows who are threats from those who are not.

You can't beat evolution.

Polymath said...

Hmm, that link got cut off, let's see if this works:

http://blog.nj.com/njv_bob_braun/2010/01/princeton_university_researche.html

The headline is "Princeton University researcher finds selective schools contribute to inequality in education" and the article is a screamingly funny (if it weren't so threatening to our pocketbooks) example of Occam's butterknife.

Truth(er) said...

"Also, certain minority groups are very powerful and threatening in this country. White people are afraid of black violence and Jewish money/media. For blacks, profiling means cops picking on 'suspicious' looking brothers. For white and Jewish liberals, black issues & concerns are of paramount importance if only because blacks vote as a bloc for the Democrats. So, even as white and Jewish liberals are afraid of black crime and live in mostly white neighborhoods, they make a lot of sanctimonious noise about the evils of profiling. And, Hispanics have jumped on the anti-profiling bandwagon since so many of them are illegal and don't wanna be 'harassed' by white cops.
And, gays don't like profiling since being labeled 'gay' in the past could mean getting fired, not promoted, or no apartment lease."

Huh. What these groups fail to realize is that terrorists have no problem setting off a nuclear car bomb in New York City, the capital of Jews, NAMs and gays. Heck, even the latest bomber was about to take out a plane over democrat detroit.

This problem is ultimately self-correcting, it seems.

Truth said...

"You can't possibly be this stupid!" Redux.

Eric said...

Schneier isn't stupid, he's just one of those guys who refuses to see the elephant he doesn't want to see.

This debate is reminiscent of the discussion on torture, where the media can line up twenty "experts" from the government and academia to tell us torture doesn't work. And yet all over the world governments effectively employ torture to get information they can't get otherwise.

Whether or not it's a good policy is another discussion, of course, but a discussion that shouldn't be short-circuited by people who want to have their cake and eat it too, i.e. "I think this activity is wrong, but rather than say 'I'm willing to put up with the consequences of not doing it', I'll pretend it doesn't work anyway."

Noumenon said...

If your fastball is relatively more effective than your other pitches, you want to throw relatively more fastballs.

Steve, are you aware that game theory says this isn't true? See here

Steve Sailer said...

Somebody should tell Yankees closer Mariano Rivera about game theory quick before he throws too many more of his best pitch. He might ruin his career throwing so many of his best pitch.

TGGP said...

I think Whiskey has a non-standard definition of "constant jihadist terror".

Mr. Anon said...

"Anonymous said...

Steve, you're also ignoring the fact that al-Qaida has been able to recruit converts-in addition to Ganczrski, John Walker Lindh and Adam Gadahn come to mind, too. If you want to get slightly obscure, information about Muriel de Guage isn't too hard to find. Lindh, Gadahn and de Guage were all wasted as assets, but I wouldn't bet money on al-Qaida being stupid indefinitely."

The way to deal with them is to treat them as what they are - traitors - and hang them. When someone betrays their own kind, they deserve neither sympathy nor pity. And it will discourage more terrorist recruits from the west. Obviously the price of such a choice now is too low. We should raise that price. Our forbearers were smart enough to realize this.

"The really sad thing about this is that you long ago identified the best methods for reducing the risk of terrorist attacks on the US: stop the "invade the world" policies and quit giving Israel a blank check to do whatever they feel like."

I agree, but what about the "invite the world" policies. You forgot them. If we simply did not admit muslims into our country at all most of our problem would be solved. But, hey, don't let practical solutions get in the way of your dislike for Israel.

Unfortunately, many people who dislike jews, become anti-jews - i.e., the only question they ask is "Is it bad for the jews?", rather than look out for their own interests.

Willy said...

I think Whiskey has a non-standard definition of "constant jihadist terror".

Like Whiskey himself said - "You can't beat evolution."

The Scotch-Irish have evolved paranoia to such an extent that virtually any and everything "are quite hostile to both ["Ulster"] and [the Scotch-Irish] in general" and potentially sources of constant terror.

none of the above said...

Whiskey:

I think Schneier's point was exactly that highly trained individuals could do meaningful profiling, but that you couldn't do a good job of it by very simple rules like "strip search all Muslim males."

none of the above said...

Steve,

I suspect that Bruce was thinking in terms of General McInerney's comments, which you linked to, about strip searching all Muslim males. Or some similar scheme. The argument I've seen floating around is that, if only we weren't so PC, we could stop screening grandmas and white people and such in favor of screening the real risks.

That won't work. In fact, if you look at the series of bombers, you see what looks like an attempt to evade low-tech profiling. While it's surely easier for Al Qaida or other Muslim terrorists to recruit young Arab Muslim men than, say, 45 year old white women, it's probably not impossible for them to recruit 45 year old white women. (And there are more terrorists and wackos in the world than just Islamic fundamentalists.) Even the article he and you linked to about using the square of the risk doesn't work all that well against a thinking enemy. (It's better than nothing, but an enemy is going to respond to your screening strategy as best he can.)

The real problem here isn't political correctness in the racial profiling sense. It's political correctness in the sense of not being able to acknowledge that there are some scary risks that the government can't protect you from. We can (and very well may) wreck commercial aviation before we manage to make it really impossible to take down a plane. Already, we've made flying unpleasant enough that lots of people choose to drive instead, leading to more deaths, since even during the decade with the worst terrorist attack in history, flying is orders of magnitude safer than driving.

none of the above said...

Eric,

I've seen experts argue the other side of that. As I understand it (I neither have nor want this kind of expertise), the problem is that torture tends to produce a lot of false leads, because you'll say anything at all to get the pain to stop. (That's true even if the torturers threaten to check your story and torture you more if it doesn't check out. People have routinely been tortured into confessing to capital crimes, accusing their family of such crimes, etc.) So when you capture some random nobody hanging out with the wrong crowd and try to wring some answers from him, you'll end up chasing down a lot of false leads.

At any rate, the argument for torture has been massively oversold, in the same way as the threat from terrorism.

Most of our anti-terrorism policies are all about protecting politicians from consequences of terrorist attacks. That's why we see security theater at airports, but don't get control of our borders. Security theater at airports creates patronage and looks secure. Controlling our borders would cost some important, connected companies and industries a lot of money.

It's good politics to be in favor of torture, massive surveillance, no-trial detention of citizens and non-citizens alike. Those things make it easier for politicians to say "it wasn't our fault, we did all we could" when the next attack happens. The costs of these actions (innocent people detained or tortured or spied on) don't land on the politicians, but the benefits do. If these policies help prevent even a small number of terrorist attacks (the eavesdropping and detention surely do; I'm not sure about the torture), even at horrible human costs, it's a win for the politicians involved. *They* aren't being tortured to death to extract information they don't have, after all.

If you don't believe me, look at what the decisionmakers actually *do*, instead of what they say. Do we have control of our borders? Do illegal immigrants and drugs have a hard time getting through? Is FEMA or someone ready to sweep in and rescue a city hit by some devastating terrorist attack? Are we even bothering to refuse entry to obvious security risks? Was the aftermath of the failure of CIA and FBI on 9/11 even to fire anyone or cut anyone's funding?

Dealing with those things would have few political benefits, and many political costs. So they don't get done.

wardall angel said...

Start with "whom should NOT be profiled" than with "whom should be profiled".

Parsi said...

I'm glad to see a Sailer on Schneier post. Bruce Schneier really is a major figure in the computer security field. Since computer security snafus interest me, I used to read his blog on a regular basis. But I gave up after one post too many on profiling. I figured that if he could be so illogical on that topic then I couldn't sensibly trust his judgment on less ideologically charged topics.