Schneier on Security

A very useful exerpt, given the heated discussion of the previous thread.

One thought: how do we make allowances for those people who go out of their way to avoid guards, dogs, cameras, etc., while maintaing the discipline of only profiling those acting "suspiciously"? Regardless of one's motivation for trying to remain anonymous in a crowd, is it right or even necessary to lump their behavior in with that of those with malicious intent?

You're digging yourself into a hole, Bruce. You come out against racial profiling in most cases and you see random searching as "a waste of money" because "it substantially reduces our liberties, and it won't make us any safer."

Then as an answer to those like myself who have criticized you for poking holes in security plans while not providing your own solutions, you offer up "intelligence and investigation -- stopping the terrorists regardless of what their plans are..."

Unfortunately "intelligence" is not a solution especially for those who espouse a worldview that elevates the values of privacy and anonymity above all others. Without information collection, sharing, and storage, the antithesis of the aforementioned values and something you and the privacy movement have fought to prevent, there is no intelligence or in the very least you have intelligence that isn't very effective.

I'd suggest that you don't need a shovel to get out of this hole. You need a flashlight, one that shines the light on the best way to protect freedom and civil liberties in an open society - more openness.

Dennis Bailey
www.opensocietyparadox.com

Bruce, you're contradicting yourself again. You claim that "the trick here is to make sure perceptions of risk match the actual risks." Then, in the next breath, you make the irrelevant point that "terrorism is not confined to young Arab males." Your left-wing dogma has tainted your sense of reality: of course young Arab males are not the ONLY terrorists in the world, but in fact they have killed more innocents through terrorism than all of their non-Arab or non-male contemporaries combined. Facts are facts: most terrorists fit this profile.

Should race be the only element of our airport security profile? Obviously not. What we need is a return to common sense: we need our screeners to judge the totality of circumstances surrounding an individual, and we need to give them the absolute right to question or detain anybody they deem suspicious, without fear of repercussions for false positives. We need to give the TSA the right to do their job as they see fit. If that means that more young Arab males and fewer 97-year-old grandmothers are strip searched at airports and train stations, so be it.

9/11, 3/11, and 7/7 have proven beyond a shadow of a doubt that being Muslim in the West makes you by definition "hinky." Until members of this subculture learn to assimilate into our populations and to root out the terrorists among them, they will be subject to greater scrutiny.

I think part of the point is that you shouldn't have an "airport security profile". If you're spending your money on securing airports, then the next attack won't be at an airport. Instead of attacking Bruce's words with false dichotomies and talk of holes and flashlights, it'd be good to read and comprehend his point for a change: Someone who wants to fight terrorism needs to be able to say "We'll find out what the terrorists are planning and be able to stop it fast!" not "If they attack this target, we'll be ready for them!" As soon as you say that "if", you lose, because then your plan can only succeed if the terrorists cooperate.

I'm reminded of the blue traffic lights and no pants plan. (http://www.geocities.com/outlawvern/VTILII.html , original site is traffic-limited, mirror at http://www.livejournal.com/users/tongodeon/259489.html .) Someone who says "don't do something stupid" shouldn't be obligated to come up with an "alternative". Not doing the stupid thing is the alternative.

@Chris Walters

You're missing the point though. Bruce is saying that by focusing exclusively on Arab males, that doing so will invite the attackers to change his / her profile.

From the text above:

"At the same time, some real attackers are going to deliberately try to sneak past the profile. During World War II, a Japanese American saboteur could try to evade imprisonment by pretending to be Chinese. Similarly, an Arab terrorist could dye his hair blond, practice an American accent, and so on."

One of the major problems with 9/11 was that drivers licenses were illegally issued from DMV workers. This particular has not been properly addressed.

What would honestly prevent a terrorist from getting a fake drivers license that says Smith instead of a name like mine (El-Wakil is Arabic for The Agent, heh), dye his/her hair blonde, and lighten his/her skin?

And by your quote:
"Until members of this subculture learn to assimilate into our populations and to root out the terrorists among them, they will be subject to greater scrutiny."

Future terrorists cells will have *far* more of a reason to try and assimilate into western culture than standard Muslims, due to the sort of this sort of racial profiling that is going on.

Please keep all this in mind. Thank you.

Here's a convenient little dilemma I'd be interested in opinions on: in previous threads, there have been arguments against the value brought by CCTV systems (which are not preventative in nature, except as a deterrence from a fear of being caught).

Today, we see the police in the UK having used those systems to quickly identify specific individuals with whom they wish to speak, and to release a visual profile of those specific individuals to the public.

If profiling is so deficient, should we ignore the information which these pictures contribute in trying to catch the perpetrators? Surely, some would need to argue, we would be better off looking at the population as a random pool, and go from there: surely we should challenge people randomly, rather than whether they appear similar to the released photos.

The simple fact is that these security cameras do what they are intended to do. And profiling occurs at many levels, with different value at each level. Both are tools, not solutions, to be applied in an intelligent fashion.

"Today, we see the police in the UK having used those systems to quickly identify specific individuals with whom they wish to speak, and to release a visual profile of those specific individuals to the public."

Be careful about that conclusion. It's not a good argument for cameras, because we don't have a control for the experiment: How much harder would it have been for the police to identity the four terrorists without the cameras?

My guess is that it would not have been that much harder. One was identified with the help of his mother, I recall.

Citing that cameras were useful does not demonstrate that they were essential.

@Bruce

"Citing that cameras were useful does not demonstrate that they were essential."

That seems like a red herring and your initial point was far more persuasive. Paul is right.

Cameras are just another tool that can be used for good or bad. What really matters is that "good security has people in charge". Having good people in charge is essential, and they undoubtedly are more effective with the right tools.

Take for example today's eye-witness account: "All described the man as wearing a bulky, winter coat, despite the warm weather, and at least one said he thought he spotted a belt with wires running from it."

http://www.guardian.co.uk/attackonlondon/story/0,16132,1534742,00.html

It might take dozens or even hundreds of spottings by the public to gather the same quality of data as a single security professional. I mean can you imagine a security professional watching a man with wires hanging out of his winter coat walking by a camera during a hot day in July? Now, imagine that one security professional monitoring dozens of locations via cameras with sophisticated alerting and profiling. Granted, visuals may not be essential for all investigations, but we should at least give the technology the credit it deserves as an extremely valuable tool to help manage the vast acreage of space and limited number of qualified officers on the ground. Do not throw out the hammer just because someone didn't know how to hit a nail on the head. Or to use your turn of phrase, the hammer is not essential for nails, but it sure beats trying to use your hand.

In this case the demonstration of the effectiveness should not really be all that difficult; just compare the accuracy of a "witness sketch" to a color hi-res video of a suspect. Or go back to the quote above that "at least one said he thought he spotted a belt with wires running from it"...that kind of lead could take hundreds of hours to validate and clarify, unless it correllates directly to data from a good surveillance operater.

@Paul

"If profiling is so deficient, should we ignore the information which these pictures contribute"

The point is not that profiling itself is deficient, but that it takes skill and fairly clear guidance to use it properly. Or you might just say some people are born with better fraud detectors than others...

So the cameras can either help or hurt the arguments related to profiling, depending on who's running the console.

The question of whether we should ignore information captured on camera has more to do with the discussion about risks/harm and the trade-offs regarding civil-liberties (i.e. unlawful search and siezure, privacy, etc.)

Another thing to note about human security agents on the other end of cameras ... even with training, they can still be subject to misdirection and social engineering. It's not like these cameras are hidden, and I imagine a terrorist could be trained to recognize the make/model of a camera to estimate its effective range.

If a subject registers as 'hinky,' what is our friend in the monitoring room doing? Is he focusing his attention on the subject, or is he handing surveilliance off to others, while continuing to monitor the other cameras on the grid?

How many 'false positives' occur before the behavior of the monitoring agent and the responding agents changes, and are they aware that this process can be managed to create a blind spot?

Better yet, take steps to make the person monitoring the cameras a malicious insider ...

@Bruce, "One was identified with the help of his mother, I recall"

The July 7 attack led police to set up a hotline, which the mother of one of the attackers is reported to have called enquiring after her son. After confirming that all four suspects were dead, the police released CCTV photos.

The July 21 attackers have not yet been (publicly) identified by name, only by CCTV photo. It is the July 21 attackers with whom the UK Police urgently wish to speak, and to that end their CCTV photos have been publicly released to enlist the public's aid.

What would you propose Police Commissioner Sir Ian Blair and Andy Hayman should have released this afternoon (7/22), if no CCTV photos were available? (See police.uk for the transcript of the press conference, and the photos.)

@Davi:

Part of my point may have been obscured. One useful way of evaluating utility is to evaluate the (absurd) extremes. In this case, the absurd extreme is to stop people randomly and only then compare them to a photo, rather than stopping people who bear a slight resemblance to a photo and comparing them to the photo. And it is ridiculous to assert that you're better off ignoring people who bear a (strong) resemblance but who don't hit on the random number generator.

So at least some degree of profiling is useful.

The discussion then (and I believe your comment) is about what degree of profiling retains a positive utility coefficient (and the concomitant circumstances), and how expert the "operator" has to be in order to effect that coefficient.

@Nick:

I believe your comment focuses on CCTV systems as preventative measures.

But I would suggest that the situations tend to develop so quickly, and are so similar to non-threatening situations ("false positives"), that the preventative nature is primarily to raise the stakes for the perpetrators. Knowing that you'll be quickly identified tends to dissuade some would-be attackers and tends to make others do more extensive planning (and risk exposing themselves by doing so).

Even in the ideal condition where a monitoring station detected illicit behaviour, the reaction time is still measured in minutes.

Traffic cameras don't prevent collisions. Speed/radar cameras don't prevent speeding. Store cameras don't prevent shoplifting. But each enhances the reaction to the event.

I still contend that the fact that CCTV systems do not prevent attacks (indeed, no single tool does) does not diminish their value.

@Davi, you said:

Cameras are just another tool that can be used for good or bad. What really matters is that "good security has people in charge". Having good people in charge is essential, and they undoubtedly are more effective with the right tools.

I believe that this is a dangerous attitude. Governments themselves are a great security risk to their people. The proper way to set up government agencies is not to give them overreaching power and then say, "we must make sure that only the best people are in charge of it", but rather to imagine that your worst enemy was the king of all and design things so that he still can't do much damage.

The American federal government is a great example of this. Despite the massive whining of Republicans during the Clinton years about how evil he was, he never did much damage from their perspective, because the system held him in check. And likewise, despite the massive whining of Democrats during the Bush years, his damage is limited as well, by the same system.

I'm not sure how to reconcile this idea with effective counterterrorism, at least while the good guys are in power. The temptation to brand your enemies as terrorists and then disappear them is going to be strong.

First off I'd like to I invite you all to look at the published photos of individuals wanted in connection with yesterdays bombing attempts here: http://www.met.police.uk/news/july_21_07_05/docs/suspects_poster.pdf

Two of these photos are useless to actually identify individuals by and so poor that I doubt they provide any more information than the sought individuals are young and male, one probably dark skinned, one of any complexion from white through to light skinned afro-carribean/dark skinned mediteranean. Two are more useful but are still vague enough that I wouldn't be comfortable using them as the sole source of identification.

None of them show a perpetrator caught in the act so they must have been selected on some other criteria. The most likely criteria is that they were selected from tapes by police officers working from a written description collected from eye witnesses. So we are probably back to having to rely on eye witness identification evidence with its proven flaws.

The truth is that CCTV is still mostly security theater. On occassions it works but most of the time it's of no use. It may discourage some criminals but clearly it doesn't discourage all or shop lifting would be a crime of the past. In England, where street CCTV is ubiquitous in town centres, street crime is still prevalent.

In London a vast number of CCTV cameras are linked to automatic number plate recognition (more than people generally realise). Records are kept of vehicle movements, in some cases for long periods or indefinately. My car is registered to me and I'm the only driver. Hence my movements are been tracked, even if only passively, so I have lost some privacy. Yet I still had a car stolen in London, it was never recovered and the perpetrator was not caught. So I have lost some privacy but I've gained no security - a lousy trade off.

"9/11, 3/11, and 7/7 have proven beyond a shadow of a doubt that being Muslim in the West makes you by definition hinky.' Until members of this subculture learn to assimilate into our populations and to root out the terrorists among them, they will be subject to greater scrutiny."

The hope is that we'll have better trained people than that, because that's a completely useless definition of "hinky," and one that will make us all less secure.

"Cameras are just another tool that can be used for good or bad. What really matters is that 'good security has people in charge.' Having good people in charge is essential, and they undoubtedly are more effective with the right tools."

Agree 100%.

"What would you propose Police Commissioner Sir Ian Blair and Andy Hayman should have released this afternoon (7/22), if no CCTV photos were available? (See police.uk for the transcript of the press conference, and the photos.)"

I have no idea.

But if someone knew, and we could evaluate the effectiveness of it, we would have something to compare the CCTV cameras to.

"I still contend that the fact that CCTV systems do not prevent attacks (indeed, no single tool does) does not diminish their value."

Cameras are primarily an audit tool; they're most useful after the fact. And there is some deterrence factor, because they are in place.

As to whether that diminishes their value or not, it depends on what you're comparing it to. It's definitely less valuable than another person, because all it can do it monitor and record. But like everything else: cameras are a trade-off. They have some value, and they have some cost (financial and etc). Are they worth it?

I contend that, most of the time, they are not.

"Governments themselves are a great security risk to their people. The proper way to set up government agencies is not to give them overreaching power and then say, 'we must make sure that only the best people are in charge of it', but rather to imagine that your worst enemy was the king of all and design things so that he still can't do much damage."

This is also true. What's going on here is that there are many different threats -- the threat of terrorism, the threat of rogue government, the threat of rogue policement, etc. -- that must all be balanced. Implementing security to defend against one threat at the expense of another can be an overall mistake.

So far, the best solution society has come up with for governments is to give them considerable power, but also exercise considerable oversight.

@Ian:

With respect, the PDF of the poster is not the best source for the images, which are available directly from the police.uk website.
http://www.police.uk/content/viewarticle.asp?xslfile=~xsl~transform_article.xsl&xmlfile=~_content~xml~wanted~MPS-BSW-23.xml

That being said, yes, the images are less than perfect even in JPEG form. (And clearly, most have been cropped, so one would presume the police have better quality originals.) But even imperfect, they are clear enough that some of their friends and neighbours are surely calling in. Compare that to what could be achieved if only a written description were available (which is what we would need to compare against without the cameras available), and that only from unreliable witnesses.

As for image selection criteria, clearly the authorities should want to show as much of the face as possible from the full set of imagery available to them. Showing them "in the act" accomplishes nothing, even if they have such video. In fact, the news conference yesterday included a description of each of their actions as determined so far, suggesting that witness statements were emphatically not the sole source of identification but rather there was more information available to authorities from that full set of CCTV tapes.
http://www.police.uk/content/viewarticle.asp?xslfile=~xsl~transform_article.xsl&xmlfile=~_content~xml~news~MPS-BSW-024.xml

Other "experts" have opined that the only reason for releasing these pictures is the suspicion that while at large these individuals pose an ongoing threat. And it is not an unreasonable extrapolation to suggest that these individuals do have a certain motivation to either reattempt murder or to escape the island. And under that extrapolation, having a picture (however imperfect) is beyond value in attempting to prevent death and destruction in the next few days.

I'm sorry to hear your car was stolen and not recovered. Clearly, it wasn't mere "joyriders" who took it. On the other hand, it is not at all clear that theft is a sufficiently severe crime that Scotland Yard would have collected and analyzed the full set of CCTV tapes to track it down. Now _that_ would represent a loss of privacy.

@ Bruce, "So far, the best solution society has come up with for governments is to give them considerable power, but also exercise considerable oversight."

I would suggest this may be the societal analogue of developing strong cryptographic procedures (power), and publishing the algorithms to ensure proper review (oversight). It may make for a weak analogy, but it has transparency of operation as a common theme.

Since I seem to have drifted somewhat from the original topic, I might as well keep going.

It occurs to me that a CCTV-centric audit procedure also affects the behaviours during the first minutes after an event. And possibly again in intermediate stages where broad profile-based descriptions can be replaced with narrow photographic profiles.

Without cameras, it is important to corral and isolate witnesses to the maximum extent circumstances allow, as they form the primary source for information.

With cameras, witness statements become less important ... and can be collected much later, as the presence of the witness can be confirmed from images if necessary. It seems reasonable to assume that the perpetrators would already have escaped before first responders could arrive, so the focus can be on removing all people to a safe distance, treating any injuries, and securing the site. There can be fewer distractions dealing with bystanders as they, too, are removed from the scene.

It would still be ideal to gather witness statements immediately and in a secure fashion; however, there can also be a competing goal of clearing people from the scene (without have to create a corral for potential witnesses some blocks away) and focusing first on anything which could make casualty counts more serious.

So it could be argued that releasing uninvolved bystanders from the scene in an immediate manner (and private - bystanders don't need to identify themselves) can be a further benefit to the use of CCTV systems as a security tool.

@Paul

"I'm sorry to hear your car was stolen and not recovered. Clearly, it wasn't mere "joyriders" who took it. On the other hand, it is not at all clear that theft is a sufficiently severe crime that Scotland Yard would have collected and analyzed the full set of CCTV tapes to track it down. Now _that_ would represent a loss of privacy."

No, you miss my point. I didn't expect tapes or any manual procedure to be of any use. What ought to have been of use is the mass deployed automatic number plate recognition - and it wasn't. This mass tracking of vehicles is in place but it's not tackling the one crime it is perfect for - vehicle theft. If it is not achieving that then the cost/risk of mass loss of privacy isn't worth it.

Viewed objectively CCTV achieves much less than its proponents claim. The UK now has the highest per capita number of CCTV cameras in the world but relevent crime rates have not fallen significantly.

ANPR has been rolled out in London and is now being added on top of this CCTV network nationwide and clearly has a much higher abuse potential than raw CCTV. Yet ANPR does not appear to be reducing crime or detecting crime - with one significant exception, that is it is successful at catching road tax evaders.

Locally (Newham in East London) facial recognition is also in use on the CCTV network. I appear to be living in a land half way to Orwell's nightmare all justified as a crime reduction measure that in fact is not proving successful. I don't expect to see this admitted by the government or police or the failed apparatus dismantled. It will be kept and it will be used for more and more invasions of privacy.

The UK government is seriously suggesting adding GPS based tracking to ALL vehicles in the UK to report their movements to the authorities for use in road pricing (in lieu of fuel and standing vehicle taxes).

Don't forget also the proposed UK ID card, complete with a £1000 fine if you fail to report a change of address to the authorities (something you are currently under no obligation to do) along with other onerous penalties.

Viewed individual these might be construed as plausable measures and proportionate to their objectives - if they are effective. Taken together they paint a scary picture.
All this is in the name of security and there may well be members of government who honestly believe this but it's poorly thought through. You don't need to be paranoid or have much of an imagination to see where and how it could all go horribly wrong.

@Paul

I think we're of a like mind regarding cameras. Are they useful? Without question.

However, what we're seeing in the wake of the London bombings is this 'cameras = good, more cameras = better' thinking.

People are embracing cameras as this security panacea, when their true value is only as a tool in the security arsenal.

But the effectiveness of a high-resolution camera manned by an expert observer can be
blocked by a crowd of people. We're also banking on 'suspicious behavior' - that profiling question. How does a terrorist behave, and is that behavior uniform among all terrorists or imposed by their methodology (i.e. they wear a certain kind of jacket because they're wearing a semtex vest)?

"Profile for young Arab males, and you'll get terrorists that are old non-Arab females."

It is much easier for a terrorist group to recruit from young Arab males than from old non-Arab females. Profile for young Arab males and you make it much harder for a terrorist group to recruit bombers.

"It is much easier for a terrorist group to recruit from young Arab males than from old non-Arab females. Profile for young Arab males and you make it much harder for a terrorist group to recruit bombers."

I think reality proves that wrong.

On the subject of profiling it looks like there was a particularly tragic example of it (not that profiling is de facto a bad thing, it can be quite useful).

I'm sure you are all aware of the events in London this week with the bomb attempt that didn't work (which now looks to be a case of over-dry, homemade explosives). On Friday there was a case of a man who was followed by plainclothes police officers after leaving a flat under surveiillance following Thursday's attempted attacks. They tried to stop him outside a tube station, but he decided not to speak to them and ran into the station, jumped the turnstyle and got onto a train. The police chased him and shot him dead in front of a large number of now traumatised witnesses on their way to work. He was apparently wearing a large coat unsuitable for the weather (it currently being abnormally pleasent in the UK).

From the descriptions in the media the man fit what most people would accept as a profile of someone to speak to, seen leaving a flat associated with terrorism, wearing a big coat and refusing to stop when asked to. The issue is that today the Metropolitan Police have said he was unconnected to the events of Thursday.

http://news.bbc.co.uk/2/hi/uk_news/4711021.stm

It has been described as a tragedy, which it is. After beleiving that they had just foiled another suicide attempt the police then discover that they have killed someone who was unconnected to the attacks. In the current tense atmosphere of London it is to be expected that everyone is jumpy.

Should the man have stopped and answered the questions? Undoubtadly. He might have been carrying drugs or been wanted for something else but he should have known that it was a terminally foolish thing to do. Did he know they were policemen? If he had known they were armed would he have spoken to them? Questions that can be asked but never answered.

There will now be an investigation into this (obviously). Tim Collins (Ex Army officer) said the following recently in relation to UK soldiers being tried for war crimes: "Frankly, in the modern Army, you take risks alone. You also have to take them in the full knowledge that the outcome may be examined long after the event in the comfort of an office in the UK by men who have never experienced combat and who are not on your side."

(quote from http://www.telegraph.co.uk/news/main.jhtml?xml=/news/2005/07/21/narmy121.xml)

I hope this is not the case with this investigation. In different circumstances it might have been a suicide bomber, or the policeman might not have shot and people could have died. The police officers have been given a "shot to kill" policy in relation to suspected suicide bombers and were following this through with someone who "fit the bill".

Does merely acting suspiciously entitle an armed police officer to "take you down"? No, but according to witnesses on Thursday the would-be bombers did not act suspiciously until just before the detanators exploded. There would not have been time for someone to disarm/kill them. The officers beleived they had the opportunity to avoid a tragedy and took it, but ended up causing a different kind of tragedy.

The problem with suicide bombers is just that. By the time you know they are a suicide bomber (unless you have prior intelligence) then it is too late to do much about it.

My hope for this investigation is that a situation in which the police acted in everyone's best interest does not get their name dragged through the mud and ruin careers as similar investigations have in the past. I hope the circumstances are taken into consideration and the matter is resolved quickly so everyone can get on with the job.

Sorry for the long post.

Another possibility, which has been raised by a user on another site, is that this man the police shot yesterday was a plant by terrorists. While not a suicide bomber, per se, he still committed suicide for the cause and his death will now make it easier for future suicide bombers to make it to their targets. One or two more of these accidental killings and the police will be forced to abandon their "shoot to kill" approach. Back to square one.

@davi

"Take for example today's eye-witness account: "All described the man as wearing a bulky, winter coat, despite the warm weather, and at least one said he thought he spotted a belt with wires running from it."

It turns out that the man was not carrying explosives, was not armed, had nothing to do with the previous attacks.

Not only that, but practically every second person in London carries a backpack or something else "with wires coming out of it" - but it is far more likely to be an ipod than a bomb.

Personally, as someone who cycles through London most days with bulky bag on the front of my bike, and mp3 player headphones coming out of it, I'm a lot more worried about this latest development than I ever was about the bombings.

@Patrick

According to the best eye witness account - Mark Whitby - who was about five feet away when the man was shot: "One of them [the police] was carrying a black handgun - it looked like an automatic - they pushed him to the floor, bundled on top of him and unloaded five shots into him". Mark also said they they fired at point blank range. Other travelers on the train have reported various things shouted by the police, none of which identified them as police or sounded like a challenge to the running man. The police were in plain clothes. The man shot was actually South American. He was shot five times in the head.

If the police had physical hold of him, as it appears they did, then firing five shots into his head is nothing less than an execution. If he had a "dead man's trigger" then the shots would have been futile, if he didn't then the two officers reported to have hold of him could have prevented him firing a trigger. The shots were unnecessary and as such make this an unlawful killing under English law. The relevant wording relating to lawful use of force (by anyone) is "use of force which is no more than absolutely necessary". Note the wording "absolutely necessary" - if there was ANY other way then the force would be unlawful.

"My hope for this investigation is that a situation in which the police acted in everyone's best interest does not get their name dragged through the mud and ruin careers as similar investigations have in the past. I hope the circumstances are taken into consideration and the matter is resolved quickly so everyone can get on with the job."

The police did NOT act in everyone's best interests - an innocent man was publicly executed without benefit of trial. What happened is, on the basis of the facts available, murder under English law. If the powers that be want the public to trust them then the officers responsible need to serve a jail term for murder and a rapid review of police use of firearms to prevent a recurrence needs to happen in short order.

Aargh, this horrible commenting system just ate all my bon mots. Quickly, before it bites again...

@Ian

Little chance the guy could have run some distance, vaulted a 3'high barrier and bolted into a train while holding a dead man's trigger.

For obvious reasons, dealing with suicide bombers isn't part of traditional Brit police procedure. In the circumstances, head-shooting the guy made sense. A body shot could have detonated a bomb, killing scores.

So I think the police acted rationally. I agree it's a shame about the guy if he was innocent. And anyway, Blair will ship most of Scotland Yard off to the International War Crimes Tribunal as soon as he can, so you'll get satisfaction...

@Bruce

Richard Reid being a Brit is irrelevant to profiling based on appearance. His arrest pictures show him as typically Middle Eastern - with hindsight exactly like a terrorist!

Still, nice post.

@Ian, 10:22 "What ought to have been of use is the mass deployed automatic number plate recognition - and it wasn't."

You seem undecided as to what purposes these cameras should be allowed to be used.

On the one hand, you seem to expect a minimal invasion of your privacy. On the other, you seem to complain when the invasion to your privacy is minimized.

You acknowledge that the purpose of the ANPR (automatic number plate recognition) cameras is road tax collection and not actual crimes, and then expect those cameras to be used against the theft of your vehicle.

Those are two very different purposes, and the former is not particularly well suited for the latter as theives can easily defeat such a system (while mere "joyriders" wouldn't necessarily bother to defeat it). And that was my point earlier.

Do you have evidence that the thieves who took your car didn't simply attach false plates over your own (perhaps stolen from another vehicle; perhaps even registered to their own), thereby defeating the ANPR? Or remove the plates (although that would presumably cause an exception within the system). To protect against theft, much more information needs to be analyzed than that a particular number plate passed a particular point at a particular time.

Cameras are a tool, but the users need to agree on what they will and won't be used for. And the interests of privacy place severe restrictions on their utility for other purposes.

@ 4:44

As for the man shot dead, the terrible fact is that the shots would appear to have become necessary in this particular instance as soon as the man jumped onto the train carriage with other innocent people on board. That he was being followed by armed British police suggests that while he may not have been guilty of any major crimes, there are certainly facts which are not in the public domain and it's a tad premature to draw conclusions about what happened without a full set of facts. Let's let the investigation do its work.

@Bruce

I'm coming at this from a stochastic control background. For what it's worth it seems like the no-brainer optimal allocation of screening effort should mirror the a priori distribution of terrorists (so if 50% of terrorists are young arab males, 25% are old arab males, 10% are jamaican/black males, 15% other; then optimal screening would mean that 50% of those screened would be young arab males, etc.).

I don't think anyone is say that our screening sample should be 100% young arab males -- but surely you agree that we should sample *more* young arab males than e.g old caucasian females ... ?

@Frank
Thanks for the update about the man with the coat. My suggestion was that a security professional observing a suspect is better trained and empowered to make a call on the situation, especially via surveillance technology rather than only from eye-witness accounts provided minutes or even hours after the fact. As you point out "coat" and "wires" are words with many meanings, but a picture is worth a thousand...

@Paul

ANPR - What you're missing is that these are actively used to surveil all vehicles. If you drive into the City of London you will be recorded doing so using ANPR. These records are kept, as best as I can find out, indefinately. This will flash lights and ring bells if the vehicle is on a terrorist suspect list, does not have a record of a valid road tax disk or, soon, if valid insurance is not on record for the vehicle. There's no problem with extending this to vehicles reported stolen but at the moment this is not done. It's being used for intelligence, anti-terrorism, minor civil offences (that deprive the state of money), public safety (vehicle insurance) but not for criminal theft (which loses citizens money).

@Paul,@Gandalf
Shooting - The guy was physically already in the hands of two officers! He was THEN shot five times in the head by a third officer. If these facts are accurate - they have been widely reported from a direct independant witness and not denied by the police - then this is a prima facie case of excessive force. The rest of the context doesn't matter - he was already physically restrained but they THEN also killed him when it was not "absolutely necessary" because he was already restrained. If you won't take my (I hope reasonably informed opinion) then how about a senior ex-police officer: "John O'Connor, a former Scotland Yard commander, said it was likely that the investigation could lead to one or more officers being prosecuted" [BBC news].

If this had happened at any other time the world would be up in arms. But in the current atmosphere it is deemed acceptable by normally right thinking individuals, such as yourself, because the police just mutter "terrorism". Think Stephen Waldorf, and then by contrast think Rodney King.

"but surely you agree that we should sample *more* young arab males than e.g old caucasian females ... ?"

No, I certainly do not. And I suggest you also give up on the "false comfort" that terrorist threats could be more easily identified by race.

Again, I feel I must point out that even the US Secretary of Transportation has directly rebutted race-based sampling:

http://www.dot.gov/affairs/042002sp.htm

"we are very clear on one point -- routinely pulling passengers out of line and subjecting them to searches need not, and should not, be done on the basis of race. Establishing such a policy would be counterproductive to our efforts to build a solid basis for aviation security. [...] we cannot, we must not, and we will not assume that all future terrorists will fit that particular profile. Without more information, we simply cannot tell -- and it certainly has not been true in the past."

Paul wrote:

"Until members of this subculture learn to assimilate into our populations and to root out the terrorists among them, they will be subject to greater scrutiny."

Mark El-Wakil responded:

"Future terrorists cells will have *far* more of a reason to try and assimilate into western culture than standard Muslims, due to the sort of this sort of racial profiling that is going on."

Mark, note that the Al Queda training manual was posted by the US Government (http://www.usdoj.gov/ag/trainingmanual.htm) and it has very clear instructions for how to properly assimilate and avoid detection. But this is hardly any surprise for a "clandestine" terrorist group, no? In this particular case the manual basically explains how to "fit in" with a secular group without sacrificing adherence to a strict/extremist interpretation of religious law.

But I am more concerned with Paul's suggestion that the act of assimilation by moderates would help prevent extremism or turn in terrorists. This seems noble in principle, but it ends up being a kind of chicken and egg dilemma during hostilities.

Moderates can be prevented from assimilating into a culture due to the very presence of the extremists' and their dominance of the profile they share. So how can the moderates overcome this and be expected to help change the overall profile from the outside?

More precisely, how and when does the "perception of risk" associated with a profile truly diminish? Sometimes, after assimilating, vague differentation is raised and people are told to properly distinguish themselves and stop trying to hide/infiltrate. Assimilation is therefore not a means, but an end that comes after forms of intolerance and prejudice have been lessened so more effective profiling can take place.

And when we talk about the path to opening society for assimilation of the moderates, consider the May 16, 2003 decision by the Bush Administration to fire the entire Iraqi civil service without exception. While perhaps intended to send a strong message about the end of the Baathist party reign, some argue that it actually had the opposite effect -- it destroyed any chances for assimilation by moderates and professionals into the new system and exacerbated violent resistance and mass exodus by the very people needed to rebuild the country:

http://www.newyorker.com/printables/fact/041115fa_fact

A US special-forces soldier is quoted in the article as saying "The problem with the blanket ban is that you get rid of the infrastructure; I mean, after all, these guys ran the country, and you polarize them. So did these decisions contribute to the insurgency? Unequivocally, yes. And we have to ask ourselves: How well did we really know how to run Iraq? Zero."

[...]

"We had a lot of directors general of hospitals who were very good, and, with de-Baathification, we lost them and their expertise overnight...we were left dealing with what seemed like the fifth string.... Nobody who was left knew anything."

I know you referred to assimilation in "our" culture, but please consider this in terms of the greater picture including US occupation. There might have been a chance to assimilate good folks into a new society and ask them to help kick out the bad or at least report the "hinky" ones, but that appears to have been seriously damaged when Bush abruptly replaced Garner with Bremer in Iraq and then immediately kicked out ANY and ALL former Baathist party members including the majority of school teachers, professors, engineers, doctors, etc.. That has to make you think again about why we can not get people to join our society (assimilate and accept our values) and help us find the real threat (terrorists).

Finally, the article cited above suggests that Bremer's decisions led to a situation where "As long as American troops remain, most Iraqis will likely continue to see themselves only as victims." That is an aspect of this profiling discussion that makes assimilation even more difficult to imagine.

@Will

Perhaps if we turn away from discussing racial types for a minute to something more neutral we might be able to get somewhere.

The City of London as a response to a string of IRA bombings instituted the "ring of plastic" which has been there for a little over 10 years. This involved closing off most of the roads into the city and choking traffic through a number of manned police checkpoints. Even casual observation of these checkpoints showed that the police always stopped vans, often stopped cars with more than one male occupant and rarely stopped cars with a single occupant.

Vans had been a favourite IRA vehicle for bombs because of their large carrying capacity, especially in view of the inprovised explosives that the IRA used. You don't have to be a brilliant strategist to see that if you're going to car-bomb the City you need to swap to single occupancy cars, preferably expensive looking ones.

Any tactic or strategy for security searches that has a definable pattern is easy for your adversary to circumvent - just swap your tactics to ones that slip through the net.

Now, back to race and al qaeda. Assuming that the threat is al qaeda: These people are not idiots, they are not primitive in their approach. Don't forget that their leaders were trained by the CIA. They will have forseen that after a couple of attacks in London with ethnically identifiable individuals that there will be a lookout for exactly that. That's the time to roll out the other ethnic groups that you've been keeping in reserve.

Assuming that Islam (albeit a perverted form) plays a role in this it is a very widespread religion - there are 2.1 billion Christians in the world and 1.3 billion Moslems. Yes, a lot of Moslems are Arabic or Middle Eastern but a lot are not.

The UK Moslem population from the 2001 census breaks down by self-identified ethnicity thus:
White 6000
Mixed 150000
Asian 775000 (This means Indian, Pakistani or Bangladeshi)
Black 144000
Chinese 5000
Other 397000 (Presumably this is where arabs are tabulated)
(rounded to the nearest thousand. Source: http://www.statistics.gov.uk/statbase/ssdataset.asp?vlnk=6891&More=Y)

The LONDON population by self-identified ethnicity (This is residents - what you'd expect to see in the street will be modified by the fact that London is a tourist magnet):
White 70.7% (no seperate figures for mediteranean)
Mixed 2.4%
Asian 12.1%
Black 11.3% (Black African 5.7%)
Chinese 0.9%
Other 2.6%

These figures are rather a poor guide to appearance as, for example, a Turk might well self-identify as white but would be visually difficult to distinguish from some african/arabic persons who might self-identify as non-white. The London population is 7.1 million, the whole UK 59.6 million.

Remember too that the attackers can be drawn from the whole UK or even overseas but that the profile for searching would be based on what you'd see in London - assuming London remains the target. Although London bore the brunt of IRA mainland activity they also attacked other major cities and garrison towns.

What we need to complete the picture is a (normal adherent):(nutter) ratio for monotheist religions. All I can say from my experience of "born again" Christians is that it is higher than one might suppose. As it is you don't need many - the London bombings so far have only required 8 front line troops. If the sane:nutter ratio is 1000:1 that's enough to field one white and one Chinese 'assault' team.

@Davi following on from @Paul

I'd be interested in where Paul lives in the context of Moslems being assimilated into (western) society.

I live in the heart of Asian London. My neighbours are mostly Sikhs, Hindus and Moslems - Indian, Pakistani, Bangladeshi and African. According to the local borough council the local ethnic breakdown is:
44% White (this includes a lot of non-British whites)
32% Asian
21% Black (8% African Caribbean, 11% African and 2% Black ‘other’)
3% ‘others’
Genetically I'm Celtic and I stick out like a sore thumb around here.

I can say with some authority that the Moslem population HAS assimilated into the local population/culture. They, like all London's ethnic groups, retain their identity and customs but not to the extent that it interferes with living alongside them. The ones who don't fit in yet are the recent influx of white, nominally Christian Lithuanians to the area.

Some sort of name may be more useful for attribution than just referring to several people as "anonymous", but it's also helpful if attribution is accurate in responses.

And there appears to be some confusion over some such attribution. I've said much, but the "assimilation" thread would, I believe, be better attributed to Chris Walters, as I never made such a statement.

As for the seemingly circular discussion around what I did post, I refer in a circular manner to my earlier statements on the matter; I don't see repeating my comments as helpful to reaching an understanding.

@Ian

I agree with you regarding the deadman's trigger but consider the following scenarios. They are simplistic and assume that the gentleman in question was a hard-core suicide bomber, not an innocent man:

1-If the guy had a deadman's trigger and the police shot him, much death would have occurred.

2-If the guy had a deadman's trigger and the police DIDN'T shoot him do you think that he would have not detonated and handed himself over? I don't. another explosion.

3-If the guy DIDN'T have a deadman's trigger and the police shot him there would NOT have been an explosion

4-If the guy DIDN'T have a deadman's trigger and the poilce DIDN'T shoot him then there would have been an explosion.

There is only one of these options that does not result in more death and destruction and that is to shoot the guy, unfortunate as it seems.

I agree that there should be a full in-depth investigation and I also agree that had this happened at any other time then the world would be up in arms about it. The Brazilian govt is demanding an explanation (I think it was originally reported as he was a South African). There are facts that are not yet in he public domain (and may never be). We need to have an investigation to find out what happened but my point is that this did not happen at any other time, it happened in the current tense atmosphere in a city in which the police are expecting further attacks.

Let's wait to hear the whole story. I hope that the circumstances are taken into account and they don't crucify the officers involved unless it can be shown that they were acting improperly.

Remember that the officers involved will have to live with what they have done, but also that someone has lost a son, brother, father, uncle in tragic circumstances.

It is a difficult situation.

@Patrick

In the prevailing circumstances if the police had not been able to restrain him shooting him could well have been the only option. His behaviour with regard to all the circumstances definately would have made the use of force justified. However, according to undisputed eye witness evidence he was *already under physical restraint when he was shot at point balnk range*. That is what makes this an execution, a murder and completely unacceptable.

The use of five rounds to the head at point blank range strongly suggests a police officer out of control. By chance, while researching something completely unrelated, I've already read a summary of every other incident where police fired weapons in recent years. Even when under fire the typical police action has been for each armed officer to fire a single round in response. Even in six year old case where two Metropolitan officers have been recently arrested by the Surrey police on suspicion of murder they only fired a single round each. That is, those officers were reckless as to the cause for firing their weapons but didn't fire them in an attempt to anihillate the target.

I'm one of those rare UK civilians who has firearms training and practical experience and appreciates the destructive power of these things. Five rounds fired into someone's head will have reduced it to pulp. When this officer fired his fifth round he will have already been looking at a shattered mess of bone, brain and blood. That's why I'm deeply concerned about this - an man, already under restraint, was deliberately anihilated.

Profiling is useful. Anybody who says otherwise is a fool. Our entire brains are geared around looking for patterns, and with the current Islamic terrorist battles engaged worldwide, it is clear that the terrorists belong to a group that can be profiled.

Yet the fact is the Brits just gunned down a Brazilian without adequate suspicion, the problem with any zero tolerance position. Zero tolerance is ignorant as well.

Clearly, if you are looking for Al Qaeda operatives, you are not looking for white people or women, though white women could be used. Clearly, though, attracting white women to the Al Qaeda cause is not easily accomplished, so Al Qaeda cannot simply switch to this tactic. Clearly, Islamic fundamentalism with a radical streak is required to motivate such cowardly attacks against unarmed people who are not bothering anybody.

So, profiling is going to happen. It has to happen as it's the ONLY WAY to catch someone before. But that certainly doesn't mean everyone who fits the profile is guilty, and apparently being Brazilian can be mistaken for being Middle Eastern, showing how poor most people are at racial profiling. The fault isn't in the profiling, but in the overreaction and over-reliance on it.

In a great nation, you are supposed to be innocent until proven guilty. You are supposed to be able to roam freely unless you are posing a threat or otherwise raise suspicions worthy of a search or arrest.

In a piss poor nation, you can just ID everyone; have cameras trained on everyone all the time; kill innocent people at point blank range; attack countries with trumped up charges; arrest people and hold them indefinitely without charges.

We are more like that Taliban today thanks to Al Qaeda than we are to the ideals of liberty and justice for all.

"Clearly, though, attracting white women to the Al Qaeda cause is not easily accomplished"

Not clear at all. Really, your point just shows again how profiling can run right off the rails...

1) History shows that you only need one-degree of separation, not necessarily the "white woman" herself to carry out the plot. Read the facts laid bare by the US Department of Tranportation.

2) Many "white" women could be symathetic or already active members of the extremist groups that are in or affiliated with Al-Qaeda. Skin color is not useless, but it also does not bring the kind of comfort you are falsely hoping for.

@Paul

"And there appears to be some confusion over some such attribution. I've said much, but the "assimilation" thread would, I believe, be better attributed to Chris Walters, as I never made such a statement."

Ooops. My bad. Sorry about that. I was attempting to manually "thread" the discussion, but accidentally swapped your name for Chris Walters.

"The hope is that we'll have better trained people than that, because that's a completely useless definition of "hinky," and one that will make us all less secure."

I respectfully disagree with this statement.

First, we must note the fact that most Muslims are either terrorists or terrorist sympathizers. Recent surveys have shown that the majority of them condone the use of suicide bombers to kill civilians. Their holy documents carry the same exact message. It is no coincidence that followers of Islam are the biggest terrorist threat to the West today.

Second, although there is no litmus test that allows us to determine 100% if a given individual in a train station or airport is a Muslim, there are some obvious signs that often give it away. In addition, the most effective security will be obtained by infiltrating the mosques (which are by definition extremist hangouts) and keeping a close eye on practicing Muslims, individuals from Muslim countries, and individuals whose pattern of behavior matches that of known jihadists.

While no security system is perfect, a multi-pronged approach aimed at identifying the threats from several angles is our best hope for neutralizing this problem.

A recent article in the Christian Science Monitor, "The sidewalks where terror breeds" (22 July 2005; http://www.csmonitor.com/2005/0722/p01s01-woeu.html ) examined some of the more radical "homegrown" UK Islamic groups. One of the things relevant to the profiling topic is how some of the people interviewed would not ethnically fit the superficial profiles.

A snippet from the article referring to Kelly, an Irish fellow who convert to Islam in Saudi Arabia and eventually gravitated to radical teachings of Abu Osama:
[quote]
His disillusionment with Britain became complete when he was sacked from his IT job "for telling a kafir [unbeliever, or non-Muslim] woman to cover up." Ironically, only Abu Osama dons religious garb. The others wear jeans and shirts. Kelly would look at home in an Irish pub.
[[end quote]

Profiling by primarily superficial traits can be a "two-way street". Just as defenders might resort to profiling in hopes of stopping the attackers, the attackers can use profiling to find profiles less likely to arouse suspicious or which would fit enormous numbers of people.

J.D. Abolins

@Chris

"First, we must note the fact that most Muslims are either terrorists or terrorist sympathizers."

That is nonsense. Absolute nonsense. Again, YOUR definition of one particular group that you fear is so flawed that it is totally self-defeating...you virutally guarantee worsening of the situation if you take such a foolish position.

What matters is that we can distinguish moderates from radical extremists in any group. Here is a good look at how that might be accomplished:

http://www.cnsnews.com/ViewCommentary.asp?Page=%5CCommentary%5Carchive%5C200311%5CCOM20031125a.html

"No single reply establishes a militant Islamic disposition (plenty of non-Muslim Europeans believe the Bush administration itself carried out the 9/11 attacks); and pretence is always a possibility, but these questions offer a good start to the vexing issue of separating enemy from friend. "

Further to my points above:

http://observer.guardian.co.uk/international/story/0,6903,1535100,00.html

"The urban middle classes may be critical. Almost every hardline Islamic movement in recent times has been led by frustrated pharmacists, engineers, teachers and businessmen."

This makes sense, and adds a chilling note to the US policy in Iraq. When you close the door in the face of moderates who are genuinely willing to help, they are forced to choose another path and radicalize.

Back to Bruce's point, if you profile based on innacurate characteristics, especially on a wide-scale (i.e. "most Muslems"), you actually play into the hands of the extremists and seriously undermine security.

oops..."Muslims"

"fact" "most" "Recent surveys have shown" "obvious signs" = bullshit

@ David ...

"Clearly, if you are looking for Al Qaeda operatives, you are not looking for white people or women, though white women could be used."

Although he is not indicative of Al Qaeda any more than 'Muslim' is indicative of 'terrorist,' John Walker Lindh is white. He is American born and raised. He looked like a scraggly bum when he was picked off the battlefield, but when he was arraigned, he'd been given a haircut, shaved, and cleaned up. Looked just like any other average citizen.

And, Lord knows former Attorney General John Ashcroft thought liberals are the next best thing when he derided critics of USA-PATRIOT as, '... scaring freedom-loving peoples with phantoms of lost liberties, I say to you, you are only aiding terrorists.'

An ideology may be more common to a given population than others, but there's nothing that says only Muslims, or only people who look like they come from the Middle East, can subscribe to that ideology.

[quote]
Clearly, though, attracting white women to the Al Qaeda cause is not easily accomplished, so Al Qaeda cannot simply switch to this tactic.
[/quote]

Unlikely, improbable ... but it should be examined on the basis of whether or not it is practicable through indoctrination or duress.

It could be the wife/girlfriend of a coworker, and she need not be aware that she has a bomb in the car - she's only helping out her friend by delivering it to the post office, etc.

[quote]
Clearly, Islamic fundamentalism with a radical streak is required to motivate such cowardly attacks against unarmed people who are not bothering anybody.
[/quote]

Fundamentalism, period. We just got done sentencing abortion clinic bomber Eric Rudolph, and he's white and non-Islamic as can be.

Bruce,
while you raised the possibility of terrorists masquerading any profiled attributes, this does not render profiling ineffective. Actually, since masquerading bears a cost to a would-be terrorist (perhaps depleting the supply of volunteers), it seems an effective approach, from a game theory POV, would be to randomly screen -everyone-, but increase the probability of screening based on some profiled attribute.
Which is in reality what common sense would say.
I wrote a slightly longer post on this issue on http://yaniverse.blogspot.com/2005/07/on-profiling-and-game-theory.html

@Yaniverse:

The cost to a terrorist organization of finding and sending a member who doesn't match the profile is negligible. In fact, the paper Bruce linked shows how it can be folded into the "scouting" that any non-fool terrorists will do.

If you use any type of static profiling (race, predetermined ticket patterns, etc.) you are automatically barking up the wrong tree. Why? Because you're always barking up the same tree, so the terrorists will move to different tree. They've paid a minor cost in switching trees and gained a huge advantage. You've paid a bigger cost and put yourself at a disadvantage. How's that for common sense?

There's never any shortage of people trying to justify a jerking knee. Several comments on this blog try to justify the London shooting: an innocent man shot five times in the head, apparently because of his skin, clothes, and apartment building.

"...it seems an effective approach, from a game theory POV, would be to randomly screen -everyone-, but increase the probability of screening based on some profiled attribute."

Yes; this is likely to be more effective -- assuming there is a good profile.

I have been getting several comments about trying to have things both way. It probably looks like that, but remember that there are two distinct decision points here. One, should we screen a percentage of people. And two, if we do screen them, what kind of selection algorithm should we use. For a lot of security situations -- checking bags on the New York City subways, for example -- I think screening people is a bad use of security resources. But if we're forced to screen people, random screening works better than profiling. If we have a good profile -- which I don't think we do in this case -- the a combination of random and profile-based screening is best.

@Bruce

You should re-read that "Carnival Booth" paper you linked. They were modelling a combined system with a "good" profile (actually they varied the accuracy of the profile and found that it had little effect). It turned out to be worse than a purely random system.

It might seem strange that the accuracy of the profile doesn't make much difference, but it makes sense when you think about it. All that really matters is the percentage of terrorists who *don't* match the profile, because it's just a matter of the terrorists identifying and sending them. Once the profile is mostly accurate, you have diminishing returns from improving it. All the statistical rules of thumb are actually working against you because you're using a statistical profile to race against people who can send their outliers.

@peachpuff:
I disagree that the cost to a terrorist organization of finding and sending a member who doesn't match the profile is negligible. There is a finite (small)number of people willing to be suicide bombers [to support this: if the number of people willing to be suicide-whatever and currently in the US were thausands, an opponent could apply other strategies such as a suicide driving, or suicide shooting, causing a cumulative casualties dwarfing 9-11; we have not seen that] ;and by attempted recruitement the opponents are exposing themselves to the risk of recruiting a double agent, or otherwise tipping their hand.
Using static profile is used -as part of the equation-, say attribute such as age or gender.
If we tend to scrutinize pregnant women less than the general population, the opponent might opt to send more pregnant-women-suicide-bombers. But this choice might limit his supply of suicide bombers enough to prevent a co-ordinate attack, or to force him to over-extend his recruitment efforts and be discovered. The fact that masquarading or selection of cell members is limited was not assumed in the carnival-booth analysis article.
Look at the situation from the POV of an opponent having N bombs and M would-be-suicide-killers. Encouraging the opponent to to opt not to use a specific would-be-suicide-killer would prevent an attempted bombing.

@Bruce:
"assuming there is a good profile" is always a good question. However is seems that you, me, and the readership are ill equipped to judge the quality of the profiles we have and the analysis those profiles comes up with, since this is not (and cannot be) public information. I'd agree that relying on non-predictive profiles would not do much good. It's a separate discussion whether the current profiles are a good idea, but this logic actually supports letting the TSA use additional information (hoping that would improve the profiles)

@Yaniverse

When it only takes one person to carry a bomb, recruiting that one person from a pool of 2 million instead of a pool of 100 million is a negligible cost, especially considering the payoff. Look how far people go to evade detection while killing a single victim.

The Carnival Booth analysis assumed no recruiting at all. It assumed that already recruited terrorists fall into a bell curve centered tightly around the profile. The algorithm is a way to check existing members for the inevitable "leaking out" around the edges of the profile. That's why it's so powerful. The profile will always miss a few, and Carnival Booth lets them figure out who. It only takes one.

"Back to Bruce's point, if you profile based on innacurate characteristics, especially on a wide-scale (i.e. "most Muslems"), you actually play into the hands of the extremists and seriously undermine security."

Yes; exactly. It's a bad security trade-off.

@ peachpuff

I think of it as a signal-to-noise problem.

@Chris Walters

"First, we must note the fact that most Muslims are either terrorists or terrorist sympathizers. Recent surveys have shown that the majority of them condone the use of suicide bombers to kill civilians. Their holy documents carry the same exact message. It is no coincidence that followers of Islam are the biggest terrorist threat to the West today."

If Islam's mantra were truly so violent, then the 6-7 million muslims in this country (http://www.soundvision.com/info/yearinreview/2001/profile.asp) would be militant against this country, instead of a handful of terrorist cells.

"In addition, the most effective security will be obtained by infiltrating the mosques (which are by definition extremist hangouts) and keeping a close eye on practicing Muslims, individuals from Muslim countries, and individuals whose pattern of behavior matches that of known jihadists."

Any sort of infiltration of muslim organizations here will do the same as what happened in Iraq.

1) Drive moderates to extremism and create terrorists born in the US.
2) Drive any terrorists away from worshipping at mosques and communicate in a more covert manner.
3) Persecute based on religion and violate first amendment rights.

I know it's tempting to pigeonhole entire races or entire religions, but in reality it doesn't make much sense. It screams "Japanese Internment" and "Nazi Germany" a little too much.

Richard Reid wasn't an Arab, or even Middle eastern. Ethically, he's actually white (although of mixed race origin).

And none of the 7/7 bombers were Arab either, one was Afro-Carribean, three were Pakistani.

@Nick B,

You're dealing with statements about statistical distributions and populations, which can't be disproved by counterexamples.

Example:

"Men are, on average, taller than women."

Nick: "Nonsense! Here is a list of tall women."

"Young Arab males are more likely to be terrorists than are 70 year old grandmothers from St. Paul."

Nick: "Nonsense! Here is a list of terrorists who don't fit the Arab male profile."

To be sure, each terrorist (and non-terrorist) adds to the set of available data. But unfortunately, there are and have been enough terrorist incidents that adding a few non-Arab data points to the set of terrorists isn't going to move the distribution substantially.

>>3) Persecute based on religion and violate first amendment rights.<<

Looking at the discussion, where profiling, contintuous cameras, etc are being argued almost exclusively by whether they have *demonstrable function* ..

hell why not throw more value into the mix. Get rid of that pesky 4th amendment, after all if police could search on a whim, all kinds of crime could be nipped in the bud (sure they'll be false positives, but, hell we can demonstrate that more crimes will be solved), get rid of the 5th too, we have ways of exracting information from people, why not use them, get rid of the 1st, too people are hiding behind religion anyhow.

When 'success' is the only criteria, things get pretty ugly pretty fast.

@Nick. Reid had West Indian (African)and European ancestry. He sure doesn't look white. Actually, he looks rather "hinky".

http://www.time.com/time/world/article/0,8599,203478,00.html

@Mark El-Wakil. You've seen the internet multiple choice quiz that lists terrorist attacks and asks you who did it. The answer is always "Muslim male extremists mostly between the ages of 17 and 40". These extremists can be Arabs, Chechens, South Asians and so forth. You can't easily tag them by their ethnic origin. But their religious views are a dead give-away. The British born bombers of 7/7 were described as having "become more religious" in the months leading up to the attacks. Two of them went to Pakistan to further their "religious studies".

If profiling is to of any use, it needs to focus on "muslim extremists". While it is possible to identify such people in an airport -- El Al does it very successfully -- it is next to impossible to screen for such people in a mass-transit system.

We do know, after the fact, that many of the 9/11 hijackers raised serious security concerns, but the PC climate of the day, meant that these concerns were not properly followed up. This story about Atta's behaviour would be funny if the consequences had not been so tragic:

http://www.frontpagemag.com/Articles/ReadArticle.asp?ID=190.

The latest surveys on Muslim attitudes in Britain are not very encouraging. 6% said they were ready to support terrorist attacks like that carried out on 7/7, 24% sympathized with the motives of the attackers, and 56% can at least understand why some people might want to behave in this way. This is in line with a survey around 1990 that showed a majority of British muslims thought that the most important issue facing the world was not poverty, or war, or disease, but killing Salman Rushdie. Given such attitudes, the British people would be stupid not to pigeon-hole their Muslim communities as breeding grounds for terrorism. The 7/7 bombers were not the first British Muslim terrorists and they won't be the last.

@Mark, again. The internment of Japaneses during WW2 had a legitimate rationale. It was intended to stop sabotage on the west coast while not compromising intelligence sources. Had the US just rounded up the known Japanese spies they would have revealed their intelligence sources to the enemy. The interned Japanese were treated humanely. That is rather different from the treatment of Jews, Christians, Gypsies and Gays by the Nazis. Equating the two is a "Durbinism".

A minor correction for your next edition.

"by ferryboat from Victoria Island, British Columbia."

Should read
"by ferryboat from Vancouver Island, British Columbia."

The city of Victoria (where Ressam boarded the boat) is located on Vancouver Island.

I'm worried that we're treating profiling and CAPS-type systems as a "silver bullet" that would stop all terrorist attacks. They're not. Alone, they are not effective. But as part of a nested, networked system of various security measures (technological and human) they add a great deal.

By forcing an Islamic terrorist organization to branch out for non-traditional members in order to carry out terrorist attacks, you force them to expend resources and expose themselves. Indeed, the reason that Al-Qaeda is such a tough nut to crack is because western and allied intelligence services have very few operatives who fit the standard profile of an Al-Qaeda recruit (young, Muslim, male, Arab/South Asian/Northern African ethnicity). Using profiling or CAPS-type systems may indeed have the perverse effect of making Al-Qaeda and similar organizations more vulnerable to penetration by intelligence services by forcing them to recruit in "hostile" territory, as it were, especially if word got out that such recruiting was going on. If Al-Qaeda's new goal is to recruit average looking white people, my guess is they will be picking up a few more CIA, FBI, MI6 or MI5 moles than they used to be able to.

I guess you could call it a "Thermopylae" problem. (After the battle of Thermopylae in ancient Greece, where a contingent of 300 Spartan soldiers held a narrow pass against a million-man Persian Army, resulting in huge casualties for the Persians, until the Persians learned of a narrow pass, barely big enough to hold a single file of soldiers, that allowed them to send a small force around the Spartan line and get behind them.) Yes, you always want to keep an eye on every possible avenue of attack, but by closing off the largest and most obvious avenues first, you force the enemy onto avenues with lower capacity and easier surveilance. Of course, the problem then becomes keeping those less obvious avenues under surveilance, which was the Spartans' problem, and ours as well.

Because of the shift I worked and the location my main expertise in profiling was taking down drunk drivers. After five years on the job I was hardly ever wrong with a non-productive stop, the few times I was, involved non-drunks falling asleep or people on medications and, rarely, somebody fiddling with radios, trying to eat messy sammiches, things like that.
Over the years, though, I caught my share of other criminals. There is a thing that is hard to define, called "the Policeman's eye". Any experienced LEO will develop the ability to spot someone 'wrong'. We may not be able to put it into words and sometimes the 'wrong' doesn't involve the breaking of laws. More often than not we will spot someone 'wrong' and due to the strictures of probable cause be completely unable to stop that person.
The value of the 'random' searches is, IMO, that that Officers will be able to act on their hunches.
Personally I suspect these random searches will involve a whole lot of 'wrong' having nothing to do with splodeydopes. My, admittedly cloudy, crystal ball shows a bunch of bluesuits disgustedly flushing small amounts of drugs down transit station toilets because an arrest won't be worth the paperwork.
Like the customs agent of 1999 that was wrong about what was wrong with that attempted bomber, they'll spot a lot of things other than what we're actually looking for. I'm kind of curious about how the public will react to that. There will be drugs found, there will be people trying to skate on warrants. It's going to be interesting.
Just as an aside, let me comment that if I were laying on top of someone I thought might explode I, too would shoot until all motion ceased. Right? Wrong? Sorry, job number one is going home after my shift is over. I'll take my chances with the after the fact second guessers.
I'm glad I'm retired.

Regarding profiling, and its general lack of utility:

Regardless of the efficacy of the process of profiling, there are inevitable consequences to searching based upon a profile rather than random searching.

The first consequences is that your process will, by its very nature, identify more false positives in the superset of people who fall under part of the profile than in the general populace. In other words, if you're profiling "young muslim arab males", people who fit under the categories of "young" OR "muslim" OR "arab" OR "male" are more likely to be falsely identified by the search than people who are "old", or "non-arab", or "female". People who are "young and arab" or "arab and male" or "young and male" are much more likely, etc.

When you're elderly, non-arab, or female, you're much less likely to regard this as an annoyance, more likely to support profiling. You're contributing to the what America's Founding Fathers called "tyrrany of the majority".

The second inevitable consequence of profiling is that you encourage people to accept unreasonable actions as reasonable, or to take unreasonable courses of action under stressful situations. I've seen people on this thread defend the public execution of an innocent man because it appeared to be a "reasonable response under the circumstances." Yet, if the police had shot and killed a 60 year old, white, non-english speaking grandmother carrying a grocery bag under the same circumstances, *everyone* would be outraged, in spite of the fact that it is entirely possible that she could have a bomb in her bag. This man wasn't killed because he was wearing a heavy coat -> there were undoubtedly several others on the same train that were wearing unseasonably warm clothes. He wasn't killed because he didn't respond to commands -> not everyone on the train would be able to comprehend (or respond properly to) commands shouted at them by a squad of armed men in civvies charging at them. He wasn't killed because there was a cable coming out of his jacket, as someone else pointed out iPods are common. Any one of those individual points might make you worthy of additional scrutiny, but none would get you shot and killed.

He was killed because he was young and male and dark, and the officers were panicked and trigger-happy (not that I blame them.) Period. A woman under the same circumstances would probably not have been shot. An older person in the same circumstances would probably not have been shot. A white man in the same circumstances would probably not have been shot. I am not condemning the officers here, I am condemning the command and training methods that make this scenario occur.

The second inevitable consequence of profiling is that you are providing a strong incentive for your target to change his delivery method. People keep bringing up Richard Reid or other non-arab or "non-terrorist-looking" examples as to why this is a bad idea, but people are forgetting that the delivery method (the "suicide bomber") DOES NOT NEED TO BE AWARE that they are a suicide bomber.

People argue that profiling is okay, because "it's harder for terrorists to recruit non-arab females" or some such nonsense. I don't know if this is true or not (from a statistical standpoint) but it is largely irrelevent, because it is hardly difficult for someone to trick, bribe, blackmail, or threaten someone who doesn't fit the profile to become the bomb carrier. This would be a minor adjustment in tactics for a terrorist cell (I mentioned this on the other thread). In fact, this isn't an *adjustment*, per se... the 9/11 terrorists likely didn't all know that they were going to die.

So, what's the tradeoff here? You've impacted the civil liberties of a class of your citizens and you've encouraged your law enforcement officials to take extreme action in likely scenarios. Your gain for this is a minor irritant to your targets, the terrorists. This is a horrible, horrible tradeoff.

Bruce, you made AP:

http://news.yahoo.com/news?tmpl=story&cid=528&e=7&u=/ap/20050724/ap_on_re_us/us_security_cameras

@Peter:

"The value of the 'random' searches is, IMO, that that Officers will be able to act on their hunches."

This is not random at all. It is profiling; just not necessarily racial profiling.

If the intention behind preventing terrorism is to prevent deaths to non-combatant citizens, then the shooting of the South American man has be regarded as a failure. Call it collateral damage or death by friendly fire. Even in combat, the incident would be investigated and unlawful actions, i.e. not accordance with rules of engagement, would be punished accordingly. Its a tough situation, though it is a good reminder that most people who do not look like the majority population are usually not terrorists.

@Don

What dataset are we talking about here though? If we're talking about terrorist incidents globally, as in all of them ever, then you're just plain wrong. If we're talking about a smaller subset of terrorists, say just al Quaeda related terrorism over the last five years, then yes - most of the terrorists have been Middle Eastern Arabs, but this dataset is so small that every non-Arab data point *does* move the distribution substantially.

You could, of course, skew the dataset to favour your argument by including say, all incidents of Palestinean terrorism, but as they relate to a specific regional conflict, I don't see why the resulting statistics would be especially relevant (and I could easily argue as a Londoner and a European that Irish and Basque terrorists should be included in that dataset).

The Bali bombers were Indonesian. The Madrid bombers were Arabs, but not Middle Eastern. The 7/7 bombers have mainly been British born Pakistanis (and it currently sounds as if the 21/7 would-be bombers are another four non-Arab data points to add to the list).

It's worth mentioning that here in the UK the vast majority of Muslims aren't Arab at all, and are generally of Pakistani/Bangladeshi descent. Profiling for young Arab males wouldn't be much help at all.

There is a good opinion piece on this topic in today's New York Post: http://www.nypost.com/postopinion/opedcolumnists/50716.htm (registration required, but bugmenot.com registration worked). It's authored by Yishai Ha'etzni, director of the Shalem Center in Jerusalem, and he writes on the Israeli experience with profiling. One of his points is that ethnicity is a factor (underscore "a") in effective sociological profiling, along with other factors such as demeanor, dress, and hair. This sort of intelligent profiling takes time and discipline to do correctly, but makes a lot more sense than random searches. He uses the example of the pregnant Irish woman that Bruce cites above as an example of intelligent profiling. It's not just about ethnicity.

We're being Pollyanna-ish if we try to ignore the role that ethnicity plays in creating a profile. It should be a factor, along with a lot of other things. Abandoning intelligent profiling for random searches is simply irresponsible and lazy and is going to lead to more loss of privacy when the inevitable tragedy happens and the public realizes that the illusion of security created by the random searches was just that.

This post is garbage.

One of the major premises, purportedly to establish that the author is only against 'bad' profiling, tries to establish a officer questioning someone who is 'acting hinky' as 'good' profiling. But it has exactly the same weaknesses as 'bad' profiling.

Based on overly general criteria? Check.. People who are nervous for any reason would fit this profile. Smuggling perscription drugs(not exactly a national securtiy issue) - Previous bad experience with police in general or border watch - for chistmas's sake, a person nervous about getting strip-searched because their underwear was dirty would fit this.

Can be easily evaded? Check. In fact