Schneier on Security
A blog covering security and security technology.
« Gel-Filled Bras |
| Friday Squid Blogging: 13-foot Tentacle Found off Santa Cruz Coast »
August 18, 2006
I've long been a fan of behavioral profiling, as opposed to racial profiling. The U.S. has been testing such a program. While there are legitimate fears that this could end up being racial profiling in disguise, I think this kind of thing is the right idea. (Although I am less impressed with this kind of thing.)
EDITED TO ADD (8/18): Funny cartoon on profiling.
There's a moral here. Profiling is something we all do, and we do it because -- for the most part -- it works. But when you're dealing with an intelligent adversary, as opposed to the cat, you invite that adversary to deliberately try to subvert your profiling system. The effectiveness of any profiling system is directly related to how likely it will be subverted.
Posted on August 18, 2006 at 1:21 PM
• 51 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
Yup. Box-cutters, knives, bombs, guns, gel-filled bras (apparently.)
Maybe it's about time we stopped wondering what the next attack method du jour will be and started looking for the one common element in all these threats: terrorists.
You know, it's just crazy enough to work.
I enjoyed the article (link: "this kind of thing") detailing the Israeli "truth pod" technology being tested by the TSA in Tennessee. (Sorry, but it reminds me of the pods in 2001: A Space Odyssey.)
To my mind, if the Israeli airports were interested in this device, that would be one thing. But, if they're not, I have to wonder about US airports being interested in it.
The technology is being developed by an Israeli company working with "former leading Israeli intelligence officials." (Read: Retirees looking to make a buck working as consultants.) My suspicion is that there are officials in the US who will pay money for any security device that can claim to have some kind of Israeli pedigree.
I worry that US officials will squander good tax dollars for any kind of snake oil, the same way Mid-Western yokels of a hundred years ago would fall for "professors" from Eastern universities. (I'm thinking the Music Man here, but you get my point.)
P.S. Please don't take my comments as a slur against Israelis, Mid-Westerners, or any group in general -- oh, except for government bureaucrats.
One problem is that people who are very nervous because they are afraid of flying will find their air travel experience even more unpleasant.
The URL you linked to "with this kind of thing" indicates the TSA want computers and software to do the screening. I can see advantages to a software solution in some areas of business (i.e cooking French fries) as it takes a relatively small team to develop it and can be up scaled very easily. Does the TSA have aversion to training people? Is the number of people requiring training too great? Finding suitable trainees too hard? Cost of paying them?
Take a reluctant traveller running late on a hot day with 2 heavy bags. Sit this person down in a machine and they might have the cops all over them... a well trained behavioural profiler will see the situation for what it really is.
I share concern with software misidentifying somebody’s traits. While its ‘false alarm rate’ might be low, the effect on the victim may be significant – say, if it adds them to a ‘watch list’ or has their phone tapped. Stick to training people in behavioural profiling, and train them well. Also, we should remember that a behavioural profiler is a member of the public and will probably pick up on potential ‘problem people’ outside the perceived area of threat. Now extend the training to all law enforcement to provide expanded coverage across the nation. This will cover unlimited public places, shopping centres, theme parks etc. Places where machine scanning is completely impractical.
The last link brings to mind the Catburglar episode of The Simpsons. A crime wave grips Springfield and everyone freaks out so much they start implementing extreme security measures such as Prof. Frink's runaway house.
For those who haven't seen it, the house detects when someone is breaking in and runs away on giant robotic legs. The demo version trips, catches fire and the "occupants" (wooden dolls) fall out a window, also on fire. Frink's response is "Of course the real people wouldn't burn so quickly."
My thought is that the only consideration should be 'is this approach useful against the forsee-able opponent'. I wouldn't let worries about racial-profiling derail security measures - so what if the system interrogates 5% of arab-looking people while only interrogating 1% of non-arabs? the percentage of arab-looking terrorists is higher than their weight in the population. And recruiting non-arabs is probably harder for terrorist organizations - with the added benefit that, if they go trying to recruit non-arab muslim converts, it would ease the work of inflitrating these operations by FBI operatives (who are, like the american population in general, mostly non-arab).
One of the articles was dismissive of the system since it only identified 100 'common criminals' but no terrorists. As there are very few terrorists (at any given day, less than 0.01 people in airports are terrorists attempting an attack) this is an expected result. and grabbing 100 drug dealers is a definite side benefit.
What happens when the behavioural research provides proof that certain races/cultures have certain behavioural factors in common?
I've found that too much 'research' proves what it wants to prove in the first place. For example: many members of the goth sub-culture self harm. Therefore it is the goth sub-culture that causes them to start self harming. Instead of the other side of the correct logical statement, which is OR: self harming people are drawn to the goth sub-culture. They are goth because they self harm, not they self harm because they are goth. (There is probably a term for this, but I can't find it!)
It's situations like this that caused the 'positive affirmation' (is that the correct term?) that you have/had in the US.
"so what if the system interrogates 5% of arab-looking people while only interrogating 1% of non-arabs? the percentage of arab-looking terrorists is higher than their weight in the population. And recruiting non-arabs is probably harder for terrorist organizations"
In the same way that not all Arabs are terrorists, not all terrorists are Arabs. Terrorists from Chechenya, (caucasian people!) have blown up themselves in suicide missions on planes. An Indonesian doesn't looke like an Arab; most Africans don't look like Arabs. Pakistanis don't look like Arabs. Bosnian Serbs don't look like Arabs. Kosovans and Albanians don't looke like Arabs. A lot of Israelis, however, *do* look like Arabs!
A large percentage of the world's population could be recruited by terrorist organisations and if you're only looking for Arabs you'd miss them.
Somebody might look like a Captain of Industry from Sweden or Finland and blow you up. The Sky Marshall sitting next to you might look like an Arab...
And we have House Homeland Security Chairman Peter King endorsing the idea that people of "Middle Eastern and South Asian" descent to undergo additional security checks because of their ethnicity and religion.
Sigh. Limitted rescources. Energy spent targeting one group is less energy on other groups. Beslan. Oklahoma. Ireland.
6 people out of several hundred. Forget reliability (results yielding what the test is supposed to ferret out) ok, not many terrorists. But 6 out of 200, (the *smallest* of what 'several hundred' could be) is 3%.
Where is the data on what a random sampling would yield. BET it is no worse.
Bad idea unless used by extreme experts.
That should have been validity, not reliability. So much for memory of stats.
Let's just give screeners Magic 8-Balls and have them ask, "Is this person a terrorist?"
I think one of the problems is that people (and 'officials') are looking for magic security - the one technique or piece of equipment that will make everything 'safe' so we can go back to Not Having To Pay Attention.
Would something as simple as a tranquilizer defeat the security booth? What about hypnosis? What makes this system different from a polygraph?
I'm surprise by your (sudden) strong endorsement of behavioral profiling. My understanding until now was that you found that "Profiling works better if the characteristics profiled are accurate." (http://www.schneier.com/blog/archives/2005/07/profiling.html)
Too often, unfortunately, this is not the case. See http://gizmonaut.net/bits/profiling.html for some extracts of some US & UK profiling policies.
Here are some home-grown non-arab terrorists.
"An explosive device destroyed a car owned by a high-ranking petroleum industry spokesperson on Aug. 3, the Quebec Provincial Police confirmed Thursday."
"It said oil companies were to blame for exploiting ''docile consumers'' by making huge profits while hurting the environment.
The same group also claimed credit for an explosion at a Hydro-Quebec transmission tower at Coaticook in the Eastern Townships in December 2004. That blast coincided with a visit to Canada by U.S. President George W. Bush."
I don't want to make assumptions, but this group is probably not chock full of arab terrorists, just home grown ones with their own beef. I'm fascinated at how an actual explosion by what in this day and age qualifies as terrorists gets next to no attention while a serious but dubious foiled plot by Islamic yahoos stops the airline world in its tracks.
You can find this on google news using the search "hydro quebec car bomb".
stop looking for the solution to how to identify and confine pissed-off people, and start looking for how to not piss people off, i say
asking a computer to do it for you is just one more step away from where we need to go
@kees: "A large percentage of the world's population could be recruited by terrorist organisations and if you're only looking for Arabs you'd miss them."
Indeed; two real and currently very relevant examples of this include Richard Reed (the "Shoe Bomber": UK subject, mixed race Caucasian/Afro-Caribbean), and one of the people currently under arrest in the UK for the current alleged terrorist plot, who is described in the media as a white indigenous Briton who converted to Islam within the last twelve months.
Given the incredibly low incidence of *actual* terrorists amongst the population of air travellers, the fact that one confirmed and one strongly suspected terrorist do not fit the suggested racial profile of Arab/Semitic appearance simply blows any suggestion of viable racially based screening right out of the water.
These two people alone not only prove what Bruce and others have been saying for a long time about the dangers of the base rate fallacy, but, given the high proportion of actual terrorists that they represent, also show how conservative were the examples used by those who have pointed out the dangers in screening for terrorists based on factors such as race.
@Joe Buck: "One problem is that people who are very nervous because they are afraid of flying will find their air travel experience even more unpleasant."
If the TSA keeps this up, we'll all be afraid to fly.
W/r/t Goth and "self harm": one way is a correlation rather than a causation.
so we can't profile based on race as 2 terrorists, in the current climate, are not arab (but the rest were 9/11, 7/7 (they should really have added the years to the ends of these dates so nobody forgets them http+//www.newscloud.com/redir/72117/)) - we can't profile on religion because how would you: a) know their religion by looking at them (although Sikh males or Buddhist monks might give the game away) and b) that would involved profiling everyone Asians, Arabs, Isrealis, Slavs, Europeans, South Americans etc.
Counldn't we just profile them on being f*&king nutters!
Why not just make peace wilh the Arabs? It would be much simpler and much less expensive.
@James: "so we can't profile based on race as 2 terrorists, in the current climate, are not arab (but the rest were 9/11, 7/7"
Wrong, I'm afraid; None of the 7/7 suicide bombers were, to the best of my knowledge "Arab".
Three were of South Asian (Pakistani) descent and the fourth, Germain Lindsay, was of Afro-Caribbean descent.
That is why profiling by 'race' is so pointless in the current climate, as you have just so neatly demonstrated.
I have a friend who is not arab, not a terroist, and who has to travel lots for his job as a chemical plant consultant. He gets pulled out of line and harrassed nearly every time. The "profile" in this case seems to be simple self-confidence (he is also an ex rock star, and carries himself as though he owns the place). That's all we can figure. He's totally clean.
I myself have been harrassed due to profiling, though innocent. I run a business from my rural place, it does product development and professional inventing. As such we have just about every capability you could imagine. Think of it -- people come and go (employees), usually smiling (I'm a nice boss), I have evidence of plenty of money (product development pays well), and own amongst a lot of other things, some nice chemistry glassware (A hobby is pyrotech, another is electrochemistry).
Must be a meth lab, right?
That's what the DEA thought, and we were attacked (no other word really is as good) by them, unidentified, armed with auto weapons, employees handcuffed and pushed into the mud, the whole TV scene.
This was not fun. Nothing bad eventuated (other than the damage they did, ruining doors and so on) because of course we weren't cooking meth, so there was none to be found, although one of my best guys didn't want to come to work for awhile as the experience of having his neck stepped on wasn't pleasant or part of even imaginary experience up till then. BTW on finding out who it was (they didn't identify themselves willingly) we ASKED for the BATF, since that sort of thing was what we were doing. They were, in contrast, quite nice, knocked, and we had a pleasant conversation -- they didn't have a problem with what I do here, and I continue to consult for them. The opposite of the Ruby Ridge experience, if you will.
Profiling may not be bad. It's what you do about it with NO other evidence that can be really harmful. I lost one of my best guys over it -- perhaps a few millions of bucks worth of financial loss since I can't book him on jobs any more. Of course it also cost me serious legal fees to get all cleared again.
For this I pay taxes (enough to pay that whole squad every year)?
Guess this is missing the real message..
@David Mery: Interesting report.
"Why not just make peace wilh the Arabs? It would be much simpler and much less expensive."
First, correct that statement to say "Terrorists", not Arabs.
"stop looking for the solution to how to identify and confine pissed-off people, and start looking for how to not piss people off, i say" -neville chamberlain
Someone made a great point above that there is no magic bullet that will let us all stop being vigilant. The answer isn't taking ridiculous things away... it's figuring out who intends to do harm. The airport isn't the only place a potential terrorist of any stripe may strike.
Terrorism is the lowest form of crime against humanity. Attacking innocent civilians because you have some problem with the world means you lose any human rights. You are no longer a human being.
I am really tired of people focusing on points of travel, as terrorist catch alls, not to mention terrorists are people who only do things to "people". While it is true that is the case, our own country has bred terrorists for years, these terrorists subvert our economic and international policies, in a way which makes us a leech upon the rest of the world. Attacks in our own country are merely symptoms of our political system being taken over, and used to further it's own goals, and not those of the people. Our system now is so completely hosed, that in 10 years when China says they want all the money we owe them, (apx. 12 trilliion dollars over 20 years), that our economy will so decline that, large businesses here will flee to other countries, and small businesses which we don't really have any more, will not be able to pickup the load. People will starve, rioting will break out everywhere, and yet there will still be a bunch of people worried about planes not having terrorists on them.
My point is we have far more troublesome problems to worry about, and security at the airports now, is quite sufficient. You won't be able to spend any more money, to make the planes any safer, by any degree which is worth the bother. Even with terrorists blowing up a plane every now and then, walking down a steep flight of stairs is still more dangerous. But you don't see any security at the top of those stairs, do you?
Rich Mogull has an interesting piece at securosis.com, comparing airport security screening to concert security screening "Concerts vs. Airports: The Role and Effectiveness of Security Screening in Public Spaces".
The term you are looking for is 'post hoc ergo propter hoc' (lit. "After the thing therefore because of the thing."). The shorthand dismissal of that error is "correlation is not causation".
The great work of Paul Ekman is coming to the UK as well. To better the TSA, apparently Heathrow and Gatwick passengers may get lie detector tests as well. All these great news courtesy of http://www.timesonline.co.uk/printFriendly/...
The great work of Paul Ekman is coming to the UK as well. To better the TSA, apparently Heathrow and Gatwick passengers may get lie detector tests as well. All these great news courtesy of http://www.timesonline.co.uk/printFriendly/...
"when you're dealing with an intelligent adversary, as opposed to the cat"
You've not met my cat, have you... ;-)
@Matt D & Kees
The presence of 'man bites dog' evidence does not make my argument less valid. Historically, most (=more than half, and I believe more like %90) of airline terrorists have fitted the ethnic description.
The argument I am presenting is that a. using ethnic evidence intelligently as part of the screening process would improve screening and b. If said improvement is sufficient, security should be the overruling consideration over any objections to this approach.
I would agree that the actual property to measure would be religion, as all of these terrorists are Muslim; but religion is not independently measurable. So, properties that statistically correspond to the risk property can be used instead - therefore, I would agree that adding ethnicities such as Pakistanies and/or Indonesians to the desired criteria may be desirable.
It seems clear that random search inside groups according to the relative conditional probability of their being terrorists brings the optimal search-to-capture ratio. For example, if group A has 100 members and 4 terrorists and group B has 500 members and 4 terrorists, searches in group A have higher capture rate.
Recalling Bruce's previous argument as to why searching based on ethnic group membership will only cause terrorists to shift members to other ethnic groups, I have responded with two claims -
a. The ability of terrorists to shift ethnic groups of members is not infinite. Indeed, they do not draw from an infinite group of volunteers; as envisioned by the fact that even today (knowing this exact argument) some or most of the terrorists attempting airline actions are ethnic arabs.Therefore, at best, the rate of sampling should be the rate of non-ethnic-arabs we believe the terrorists can achieve. That is, if you believe that terrorists can achieve a 30:70 blend of arabs/non-arabs, and if arabs constitute %10 of the flying population, the optimal search ratio for ethnic arabs would be 10*30/70 = about 4 times the rate of search for non-ethnic arabs
b. There is a cost to the terrorists in shifting ethnic groups, resulting in increased delays and increased risk of exposere. That phenomena is benefitial and should be encouraged.
and no, I do not claim that this would be a magic bullet. Just increase search efficiency. And decrease the probability that I will be blown to bits. Sounds like a good tradeoff here.
To the admin:
There's a misplaced or missing ancher end tag on the funny cartoon reference.
D'oh! And I can't spel "anchor", either.
Making peace with Arabs would help you make "war" on terrorists. Treating them as interchangable is a major cause of problems. Is it really so difficult to grasp?
"The presence of 'man bites dog' evidence does not make my argument less valid. Historically, most (=more than half, and I believe more like %90) of airline terrorists have fitted the ethnic description."
"For example, if group A has 100 members and 4 terrorists and group B has 500 members and 4 terrorists, [random] searches in group A have higher capture rate."
OK, now redo your (incomplete) calculations for likelihood of random capture for the scenario where group A has 1,000,000 members and
24 terrorists and group B has 59,000,000 members and 1 terrorist.
Those figures are by no means accurate, but they're closer to reality (for e.g. the UK) by several orders of magnitude than those you have used above.
Now, factor in the disadvantages, the false negatives: the relative chances of the group A or B terrorists slipping through with and without the racial/ethnic/ancestral profiling.
Now, add appropriate margins of error and see if the before and after figures show one jot of difference when those margins of error are taken into consideration (hint: they won't, I promise).
Finally, consider the human element of, essentially, all non-white airline passengers being considered an increased terrorist threat (for this is what it will degenerate to when you attempt the conjunction of 'race' and 'ethnicity' and 'religious affiliation' and 'region of ancestral origin' that your 'random' screening arguments are rapidly approaching).
The downside of this is difficult to quantify in itself, but think of the incident in LA involving Rodney King (and its consequences), of the murder of Stephen Lawrence in London (and its handling by the police) and of the root-causes of the numerous race-riots in the UK (and elsewhere) over the last twenty-five years.
None of the above involved terrorist acts, but *all* involved, in one way or another, the kind of low-level, "Its all for the greater good" racism typified by proposals to modify random airport passenger screening to incorporate a crude racial or ethnic bias.
"Sounds like a good tradeoff here."
Sounds bloody awful, here...
Behavioural Profiling (the putative subject of this thread) can work well, if implemented intelligently. Racial Profiling is dangerous in and of itself, in addition to its lack of efficacy in situations such as these.
You make the racist assumption that only Arabs can inflict terror. Terror is something that can't be tolerated and is no means to achieve anything. If any peoples want peace, they must come to the table, not kill innocents. You kill innocents, you have no right to anything.
That's probably hard for Arab hating racists to understand.
"likelihood of random capture for the scenario where group A has 1,000,000 members and 24 terrorists and group B has 59,000,000 members and 1 terrorist."
OK, lets run some numbers - just to illustrate the point.
Assuming extra scruitny is relatively cheap (say, four times the normal screening time) and that it increases capture probability considerably, and taking into account that you are only dealing with the people at the airport (so you aren't screening 59M people, just the passengers for the day), and that you can further constrain the population by age and gender (similar argument to my previous one means that if %50 of the passengers are women but you only expect %25 of the terrorists to be women, you adjust accordingly), then screening %3 of the passengers more thoroughly is practical, which would end up with, say, over %50 of the more-suspect groups being screened. This makes it more likely that at least one terrorist will be apprehended, which would alert authorities and allow emergency measures to capture the rest (assuming that terrorists try for coordinated operations, which they do).
To run the numbers, if a 10-terrorist groups tries to pass unscreened, and 5 of them are in the highly-scrutinized group which gets %50 scrutinization, you get 0.5^5 = 0.03 -> %97 chance of screening at least one of these five, and
0.97^5 = 0.85, so %15 of screening one of the other five. 0.85*0.03 = 0.0255 resulting in %97.5 chance of screening at least one.
Compare this to the result of %3 totally-random screening, which is unlikely to screen any member of a 10-people terrorist group (0.97^10 = 0.74, so chance of screening of one or more terrorists would be %26). With 3.5% totally-random screening you get
(0.965^10 = 0.70) %30 of screening at least one terrorist member.
I do not see the Rodney King example as relevant. It is highly unlikely that anyone would be assaulted in public and in front of visible cameras.
I do agree that there will be some cost in 'good will' for specific screening. I just don't think that cost is in the order of another plane being blown up (and its 200 passengers being murdered).
And I am not going to even respond to the smearing of calling this approach 'racist'. Screening people based on all available information is not discriminatory, it is rational. Purposefully ignoring available information in deciding who to screen constitutes injustice towards the possible victims, who would expect us to do all we can, not less, to prevent future terrorist acts.
Hmm, maybe you're right, but isn't it more likely that I just noticed that you said "correct that statement to say 'Terrorists', not Arabs." and found your assumption that 'Terrorists' is a better way of saying 'Arabs' was just a teeny teeny bit predjudiced... dare I say racist.
Since currently the vast majority of the active fighting by the US is against Arabs in Iraq, a place which was previously very anti-terrorist and is now believed by the US State department one of the primary recruiting grounds for terrorists (http://seattletimes.nwsource.com/html/nationworld/2002357529_iraqterror05.html) I think that it's pretty clear that the US making peace Arabs would be a good way of reducing terrorism.
You are assuming that terrorists are stupid robots. The problem with profiling is not that it can't predict who might want to be a terrorist. The problem is that it allows terrorists to predict who will be screened.
Lets assume that the terrorists know who is likely to be screened (they either know the rules or they just do trials where they send their people through with no weapons and see who gets screened). They only use low risk people for carrying weapons.
The chance of finding someone in the target group is 0% (remember they aren't carrying weapons since they know they are liable for screening). But the chance of finding the terrist with the weapons is now reduced direct in proportion to the screening you do on your target group. Based on your figures, you have increased the chance of a terrorist suceeding by about a quarter.
Security is not like engineering. You are dealing with an adversary which is actively trying to subvert your mechanisms.
"he problem with profiling is not that it can't predict who might want to be a terrorist. The problem is that it allows terrorists to predict who will be screened."
I answered this concern above. Your point would have been valid if terrorists' ability to shift recruitment was perfectly elastic. As this assumption is not correct, your conclusion is wrong.
For example, you assumed "They only use low risk people for carrying weapons.". But unarmed terrorists are not a concern, or have not been to date. To blow 7 planes, you need 7 terrorists with bombs; having 2 terrorists per plane (but only one bomb) will only contribute marginally, and having two bomb-less terrorists on a plane (esp. now that cookpit doors are somewhat armoured) is a non-threat. Complex strategies (2 handguns carried by innocuous terrorist, one given to the other terrorist) may help their cause, but not enough to compensate.
So, if the risk is a function of the number of armed terrorists, if you want to kidnap 7 planes, you need 7 innocuous-looking terrorists. Which may be harder to find.
Also, (using Matt's numbers of 1:60 ratio), reducing the screening of the general population enough to screen most of the suspected population would cause very little impact on general population random screening (since the difference between 3% and %3.05 is small).
Given that you have purposefully neglected to address issues of false positives/negatives etc in your figures, and that you have purposefully neglected to address the costs of your scheme's various failure modes, I'm not bothering to dissect your figures in any detail at this point, save to observe that, at the rates you suggest (somewhere between 35 and 50 percent 'enhanced' screening across all passengers), there will be little cost difference should you apply enhanced screening to /everyone/, regardless of their race, their age, their gender, or whether or not they won't eat bacon on religious grounds.
If nothing else, 100% enhanced screening would be more consistent with the 'safety over all' position you appear to be adopting - by your preferred methods of estimation, it would catch virtually all terrorists and prevent virtually all attacks on aircraft in flight, and at an overall cost very little higher than the selective screening you advocate. So why are you not advocating that everyone is checked to the same degree, if the threat of the loss of one plane to terrorists is as unacceptable as you would appear to belive?
"I do not see the Rodney King example as relevant. It is highly unlikely that anyone would be assaulted in public and in front of visible cameras."
Rodney King is highly relevant, if, as you propose, half of all non-white passengers are to be automatically subject to intimate searches etc, and at a rate more than three times that of the likelihood of similar screening being imposed upon white passengers.
The issue here is not whether a literal replay of Mr King's assault will occur, but rather the whys and wherefores of the race riot that was triggered by it, and, by extension, the likelihood of the implementation of proposals such as yours fomenting a sense of injustice similar to that which was the fuel for the riots following Mr King's assault and furthermore of them also providing the trigger event for similar civil unrest - anyone objecting strenuously to random anti-terrorist searches is placing themselves in significant danger in the current climate.
Remember, more 'thorough' searching is somewhat of an euphemism - it also means 'more invasive', 'more personal' and 'more publicly degrading and humiliating', especially to those who are unreasonably singled out to receive such treatment, for example by way of skin colour.
"I do agree that there will be some cost in 'good will' for specific screening. I just don't think that cost is in the order of another plane being blown up (and its 200 passengers being murdered)."
As noted above, if your assessment of costs in this way is consistent, then you would remove the stipulation that only certain groups are screened and screen everyone, rigorously - the increase in screening costs (probably less than 50%) would presumably still be less than whatever cost it is you are assigning to 200 or so lives plus the capital cost of an airliner.
"And I am not going to even respond to the smearing of calling this approach 'racist'. Screening people based on all available information is not discriminatory, it is rational."
As was pointed out previously by myself and others, terrorist != Arab. As was further pointed, your scheme will invariably degenerate to considering all 'non-white' airline passengers as a higher terrorism risk than 'white' passengers, due to the inherent difficulty of identifying, by racial and ethnic origin, adherents of a particular extreme-minority belief system (i.e. Islamic fundamentalist terrorists).
Whether you like it or not, and however you attempt to dress it up in good intentions and "one life is too much" style rhetoric, the position which you advocate is one of racial and ethnic discrimination for some perceived greater good.
"Purposefully ignoring available information in deciding who to screen constitutes injustice towards the possible victims, who would expect us to do all we can, not less, to prevent future terrorist acts."
Then screen everyone. Otherwise you will, by default, be doing "an injustice to the possible victims", because sooner or later some terrorist group of whatever colour, creed or motivation will attempt to get another bomb aboard another plane, and these "possible victims", of course, "expect us to do all we can, not less to prevent" this putative terrorist attack.
That cloud of smoke and debris which you can see rapidly receding below you, "reasonable", is your own petard - an apt metaphor under the circumstances, if ever I saw one.
Feel free to try for the last word, if you so wish, "reasonable" - I for one am tired of repeating myself over and again in the face of your willful incomprehension, and consider the subject closed.
Previous comment should have begun with the line "@reasonable:"
"There will be little cost difference should you apply enhanced screening to /everyone/, regardless of"
Your entire argument is based on misunderstanding my figured above.
Specifically, in the numbers above, 1/60 was in a 'more suspect group' (in reality, as I stated, the more-suspect group would be reduced further by other criteria such as age).
For every 1000 passengers, total searches (random searches) were %3
, or 30 searches.
As the more-suspect group (1/60) contains 15 people, focused searches came out as (985 passengers at %2.23) + (15 passengers at 50% = 8) would also result at 30 searches.
Subjecting everyone to a %50 random search would result in 500 searches, however.
If you can only do N=30 searches (and with current technology, it's impossible to screen everyone, or even %50, whichin acceptable time), method B seems more efficient.
"Feel free to try for the last word, if you so wish"
I'd have been quite happy to stop the argument as our positions are quite clear; however as your numbers don't add up, that would not make sense.
How about we narrow the discussion to the issue of whether focusing a larger portion of screening on ethnicity-identified groups would result in higher detection rates?
[you stated there will be other social costs to such a policy, I agreed. We differ on how high these costs are - and it is unlikely this second point can be settled. ]
The first point however (higher detection rate) seems more likely to be resolve-able, esp. in the context of this website which is security-related. In specific, I am interested to see if anyone has any specific arguments to my point of "since the terrorists cannot easilly shift the ethnicity of their operators, they cannot take advantage of non-random scrutinization".
i am a pakistani i need visa for U K
Jesus F'ing Christ.
The solution is obvious.
2nd Amendment Airlines.
Anybody stupid enough to pull something dies.
That's a solution that is both consistent with liberty and with security.
Grow up and stop being sheep. Get away from this screening/papersplease/ mindset. Be men.
"The technology is being developed by an Israeli company working with "former leading Israeli intelligence officials." (Read: Retirees looking to make a buck working as consultants.) My suspicion is that there are officials in the US who will pay money for any security device that can claim to have some kind of Israeli pedigree."
Israel has a good scam going. They've learned that the best way to spy on the world is to be the country making all the spy gadgets every country needs to spy on their own citizens and everyone else.
This lets Israel spy on everybody.
This is why an Israeli company got caught selling CALEA wiretap information to a drug gang in Los Angeles. The FBI was highly upset that Israelis had too much contact with the CALEA hardware and software. They felt it was a serious security risk.
Schneier.com is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc.