Monday, August 21, 2006



A victory for terrorism

As if the British government's plans to reintroduce internment weren't enough of a foot-bullet, now it seems British rednecks want "flying while Asian" to become illegal. 150 British passangers refused to board a flight from Malaga to Manchester until two Asian men were removed from it. Worse, the craven airline - Monarch, for anyone wanting to express their displeasure - bowed to it, rather than saying "they've passed security, there's nothing suspicious about them, and we cannot victimise innocent people on the basis of your racism".

Its so egregious that even the Tories object. Their "homeland security" spokesperson is quoted as saying

"This is a victory for terrorists. These people on the flight have been terrorised into behaving irrationally.

"For those unfortunate two men to be victimised because of the colour of their skin is just nonsense."

He may be a Conservative, but on this, he's completely right.

16 comments:

A lot of "public service" advertising around terrorism encourages people to "report anything suspicious" and to *be suspicious* more generally. It seems that these two were considered suspicious by a significant number of those watching them, some of whom then acted on their suspicions.

The first question, then, is whether the suspicions were reasonable. Do "flying while scruffy", wearing a thick coat, and being a compulsive watch-checker at the airport doesn't really merit suspicion?

The second question is how the authorities and the airline should have responded, given that these guys had already cleared security. I guess to a certain extent they're obliged to act on the suspicions reported to them?

Posted by dc_red : 8/21/2006 03:43:00 PM

Oh dear god. What a pointless panic over nothing. Two asian guys who were scruffy and spoke in a language that some passenger "thought might be arabic".

I and my family fly to the US in a few weeks for a friend's wedding and to catch up with relatives. We're then flying around inside the US.

And what's the most dangerous part of our trip? It's the two weeks we'll spend driving around the East Coast. The USA's roads are far more dangerous than their skies are. The same's true of England.

Posted by Icehawk : 8/21/2006 04:22:00 PM

This is the logical outcome of the Clash of Terror, which is all about a War on Civilization.

We will stamp out all civilization and civilized people, and Western Terror shall prevail.

Posted by Id : 8/21/2006 04:28:00 PM

And what's the most dangerous part of our trip? It's the two weeks we'll spend driving around the East Coast. The USA's roads are far more dangerous than their skies are. The same's true of England.

Yes people (including at this blog) are crap at assessing risk: they won't stand for minute error rates and pesticides/toxins in food or the environment but they'll regularly expose themselves to much greater risks by *driving* a mile or so to get a slightly cheaper loaf of bread, or to a farmers' market, etc..

For what it's worth too... even though the false alarms are galling, all of the experts agree that having the whole population mobilized to be suspicious is one of the most important lines of defence against terrorism, one that becomes more important insofar as you don't do things like intern people preventatively and/or create a big brother state to gather intelligence and/or close your borders/restrict immigration etc.. If the weapon of choice is bombs in packages then any unattended bag is suspicious.... if it's suicide-bombers then it's people themselves who are the loci of suspicion and who may raise alarms on what objectively seem like the slightest of grounds, and who then have to bear the brunt of the false alarms. When the cost of getting a false-negative is incredibly high - essentially infinite as far as you're concerned if you're another passenger - then it doesn't take too much to make raising the alarm worth it. That unattended bag probably isn't a bomb either.

But, hey, if people bitch and moan enough about false alarms then eventually the price of false alarms goes up so that the alarm trigger is made less sensitive (to fit the new budget constraint), which increases the chances of a false negative .... Sooner or latter then somebody gets blown out of the sky thinking: "Damn, I knew there was something fishy about that guy, my inner alarm bells were ringing. I should have said something but I didn't want to open myself to being called a 'racist' and a 'redneck', and I rationalized that he must have been through security....I guess I embarrassed and rationalized myself and my family to death".

An apology to false alarmees in the Monarch Air case is appropriate, and general public discussion about how to improve criteria for suspiciousness (deference to cabin crew as a regulative principle?) and perhaps of what sort of herd behavior/mob psychology broke out in this particular case (150 people is remarkable!) might be important. But slamming people as racist and so on is unhelpful, question-begging, and fatal in the long run (but don't expect the people bitching about false positives now to apologize for the false negatives their actions will help cause later - that'll always be someone else's fault!).

Posted by stephen glaister : 8/22/2006 02:50:00 AM

Stephen: based on your nym, you fall into the demographic group ("males") which is responsible for committing rapes in this country. While false alarms are clearly galling, rape is a serious crime, and the consequences of a "false negative" are devastating (essentially infinite as far as you're concerned if you're a victim). Are we then justified in ensuring you are detained by security staff at every opportunity and questioned about your intentions and whether you are planning to rape anyone, just because you "fit the profile"?

Thought not.

Posted by Idiot/Savant : 8/22/2006 03:30:00 AM

It's a ticklish point. I do think that resources should follow the data... so in general males should be more tightly surveiled (which of course happens every time you pass through customs... ), black males still more so, and so on - the basic rule is you just have to go where the criminals and their victims are. If you don't do that - and insist say on equal surveillance/equal distributions of police resources then (since victims of criminals aren't randomly selected,e.g., most crime by maori is on other maori, most crime by african americans is on african americans) then you will be disriminating against specific classes of (e.g. maori, african-american) victims if you follow the equal-surveillance etc. model. This last point is widely overlooked and even denied but it's true. OK, now let's consider your case:

Are we then justified in ensuring you are detained by security staff at every opportunity and questioned about your intentions and whether you are planning to rape anyone, just because you "fit the profile"?

The example changes quite a lot...

1. The cost of a false negative drops from a 150 dead to 1 person raped (We charge rape less than we charge *1* murder - and it's very important that we do - we don't want to give rapists perverse incentives to kill their victims: if the sentences were the same then killing the rape victim would be rational removing of a witness....which we don't want...)
2. The profile you literally describe ("male") covers half the population so isn't especially useful (hard to think through your case consistently for that reason - hard to imagine enough women to do all the interviews etc.). But if you narrow the profile to unaccompanied males between 20-40 then maybe it could work.
3. The opportunity for the crime in question is unlimited - you can rape anywhere, but there aren't too many occasions on which you can spectacularly murder 150 people. So the "alarm regime" you mention would have to be utterly pervasive, and have simply massive false positive costs distributed the length and breadth of society.
4. And there's much more to say (I think the Air NZ and Qantas system of not seating unaccompanied chhildren next to males is defensible.)

But the big points are 1 and 3 - your example *shrinks* the costs of false negatives a couple of orders of magnitude, and *inflates* the costs of false positives a couple of orders of magnitude. And then you expect me to have any sort of logical or more broadly rational problem in rejecting alarms in your case but not in the airline case? ("Thought not") Think again, my friend. You're the one with the problem!

The issues about different sorts of errors and their costs are related to the classical "How many guilty get to walk free, to ensure that 1 innocent isn't imprisoned?" problem. Eugene Volokh's excellent review of this is here if anyone's interested...

Posted by stephen glaister : 8/22/2006 03:13:00 PM

In London during the IRA campaign, if an empty bag was left on a train people would move a few seats away from it - they certainly wouldn't raise an alarm, because to do so would bring the railway grinding to a halt and leave them waiting for hours on end. Which is perfectly rational - the chances of the bag containing anything more explosive than sandwiches was (and is) miniscule - they were just taking a reasonable risk.

Posted by Rich : 8/22/2006 08:32:00 PM

The way I read the article, some people exercised a decision not to fly with people wearing heavy clothing and checking watches etc.

They did not demand these people be removed, they just decided they didn't want to risk flying.

Whilst race and language was obviously part of the decision, it was probably triggered by the men's behaviour.

If you read the article carefully, it was the pilots who ultimately decided not to fly until these men were removed.

Making the claim that passengers refused to board "until these men were removed" may be wrong.

Should people have the "right" to decide to get off a plane flight for whatever reason, before it takes off? Other than forfeiting airfares, there would be expenses incurred by the plane having to unpack baggage and lose departure timeslots and several downstream costs.

This could be a new area for travel insurance.

Posted by ZenTiger : 8/22/2006 09:22:00 PM

Stephen: perhaps because you don't live here you don't seem to understand that your average white Mancunian can't tell the difference between a Hindu, Sikh or a Muslim of Arabic, Bengali, Afghani or Persian background etc. Actually, it seems to your mind that it doesn't matter. You seem to suggest that the public should raise the alarm when someone of a different hue acts "suspiciously".

Fart all planes get blown up, but people seem to have read too many Airport novels. I'd worry more about the dreadful health and safety in NZ and how you can't even sue companies that offer dangerous products. Take the ferry "Aratere"...a disaster waiting to happen.

Posted by Anonymous : 8/23/2006 12:37:00 AM

Anon, I'll annotate:

you don't seem to understand that your average white Mancunian can't tell the difference between a Hindu, Sikh or a Muslim of Arabic, Bengali, Afghani or Persian background etc. Actually, it seems to your mind that it doesn't matter. You seem to suggest that the public should raise the alarm when someone of a different hue acts "suspiciously".

I'm sure you're right about people being very imperfect observers... They'll trigger many false alarms. You're right. And that's normally a price worth paying to avoid a devastating false negative. You exaggerate when you say that I want the public to raise the alarm *every* time someone acts suspiciously. Many thousands of Muslims etc. fly every day into and out of Britain without issue.... many of those doubtless strike someone somewhere as somehow worrisome and yet no alarm is raised, the initial cause of worry isn't backed up by anything more and so the matter is dropped and rightly so. Every now and then, however, if you have an energized public, then there's going to be a case in which initial fleeting concerns appear to be amplified by further anomalies, and the alarm will be raised. *That's* what I don't have a problem with... even though it'll normally be a false alarm. Apologize to the alarmees, and move on.

Fart all planes get blown up, but people seem to have read too many Airport novels.

Maybe so. But we *are* supposed to be taking air security issues seriously. But if that's so then irritating false alarms are just part of the deal (as are irritating long security lines). Do feel free to declare that there's no significant threat left and that people can return to their normal state of not giving a crap about security and of just assuming that all of that's taken care of by the relevant authorities...

I'd worry more about the dreadful health and safety in NZ and how you can't even sue companies that offer dangerous products. Take the ferry "Aratere"...a disaster waiting to happen.

The inter-island ferries do seem to have some issues and the other stuff you mention may be right too.... Not sure what the point of bringing this up is here though! No one gets a pass on external air security issues just because they've got some problems with their food supply or their underdeveloped tort law or whatever it is....

Posted by stephen glaister : 8/23/2006 06:09:00 AM

"The example changes quite a lot... "
Yes, it involves trying to prevent something that happens hundreds or thousands of times every _day_, rather than something that might happen every other year or so if traditional security stuffs up.

"3. The opportunity for the crime in question is unlimited"
Irrelevant - that applies to both rape and terrorist attacks, as any Israeli could tell you.

"But we *are* supposed to be taking air security issues seriously."
We are? Says who?

"Do feel free to declare that there's no significant threat left and that people can return to their normal state of not giving a crap about security"
Sounds good to me.

Posted by Commie Mutant Traitor : 8/23/2006 09:51:00 AM

Yes, it involves trying to prevent something that happens hundreds or thousands of times every _day_, rather than something that might happen every other year or so if traditional security stuffs up.

Yes, and funnily enough, individuals who have their inner alarm bells triggered that there's something worrisome about their date (or whatever it is) so that they are concerned about possibilities for sexual violence against their person are not normally patronized as lunatics or sexist for wanting to leave that party (or whatever it is).... And they're entitled to a few false alarms.

There are other issues I don't have time or energy to sort out for you. There's the question of what to do about a type of threat, each token/instance of which does relatively little damage but which, because it has many instances, cumulatively does great damage, and then there's the question of what to do about a type of threat each relatively rare token of which does great damage. I can't explain the whole calculation to you (I have work to do!), but essentially having lots of chances to separately make a correct detection of parts of the highly distributed threat effectively controls for the sort of cumulative scaling up of the threat problem you (fallaciously) raise.

"3. The opportunity for the crime in question is unlimited"
Irrelevant - that applies to both rape and terrorist attacks, as any Israeli could tell you.


Subtle equivocations here. There aren't too many ways to inflict genuinely massive damage... and they can be kept track of at least to some extent. (And to the extent you can't, you have to keep building walls both literal and metaphorical until that changes). Overall, though, the Israel case is a good one for me in that the Israelis take simply mammoth precautions about air-travel... extensive, intelligent screening/interviewing of passengers, armed guards on all El Al flights, and so on. It's also one of the sticking points for Israel to try to settle things with Palestinians: they are *very* reluctant to give Palestinians control over airspace... for excellent reasons.... but then Palestinians correctly claim it's only Potemkin sovereignty that's being granted to them, and round we go.....

"But we *are* supposed to be taking air security issues seriously."
We are? Says who?


Please don't come the raw prawn..... For example, if NZ unilaterally stopped screening baggage or ceased taking care of its passports, then NZ-ers would soon not be able to travel.

"Do feel free to declare that there's no significant threat left and that people can return to their normal state of not giving a crap about security"
Sounds good to me.


Well... the experts in these sorts of matters disagree with you, and think that a public that isn't treating security as always Someone Else's Problem is a huge part of combatting terrorist activities perhaps especially of the highly-coordinated 9/11 variety. There are issues about how to manage a public of this kind but denying it has a role to play as you do appears to be a non-starter.

Posted by stephen glaister : 8/23/2006 05:52:00 PM

Stephen - two points:
1. I'm curious about your nebulous "experts"? Who are they? Do they come from the same group as our own illustrious SIS? Do they include the consultants and companies within the security industry? Some might conclude many so-called "experts" have a vested interest in heightening security.
2. Your risk analysis ignores the perverse motivations of both the political class (who are getting re-elected on the back of talking tough on security) and the media (which thrive on playing up any hint of a crisis). Therefore the "inner alarm bells" of significant parts of the population are not to be trusted, as they've been deliberately ratcheted up to frequent false-alarm. So if frequent false-alarms occur, the public will naturally start to disgreard them. There's clearly no viable long-term increase in security in this scenario.
My point is you're attempting to treat this as if it were a rational scenario. It's nothing even vaguely rational.

Posted by Huskynut : 8/23/2006 06:40:00 PM

Huskynut: I guess the most recent thing I've read on this topic which made the basic point I need was the piece by James Fallows in _The Atlantic Monthly_ (Russell Brown posted some links to this but I don't think they still work) where Richard Clarke was quoted to this effect. Fred Kaplan at slate.com whom I take to be a very fair dealer but also very insider-y (and with whom - full disclosure - I've had the occasional piece of correspondence about democratization) has often quoted bits of reports to this effect.... So the experts I have in mind are American, but surely the basic point has to apply everywhere - it's an extension of the principles of neighborhood watch after all.

Your risk analysis ignores the perverse motivations of both the political class.... and the media.... Therefore the "inner alarm bells" of significant parts of the population are not to be trusted, as they've been deliberately ratcheted up to frequent false-alarm.

Yes I think there are worries of this sort and my own view - expressed near the beginning of this thread - is that a productive use of the somewhat extraordinary false alarm raised by the very large group of passengers would be for a public discussion to develop about how to screen ones suspicions (perhaps instruct people to run though a check-list including "Am I just being a racist?"), how to pool infomation with others (e.g., don't lead them, don't say, "Doesn't that guy look suspicious to you?"), and so on. This might help reduce the false positive rate which is certainly desirable without increasing the false negative rate which flagellating them as racist ninnies (with the aim of getting others not to raise similar alarms in future) will almost certainly do.

So if frequent false-alarms occur, the public will naturally start to disgreard them. There's clearly no viable long-term increase in security in this scenario. My point is you're attempting to treat this as if it were a rational scenario. It's nothing even vaguely rational.

These are worries I agree...and there are evidently issues with the Muslim communities in both England and France s.t. they are poorer, more ticked off and alienated than the Muslim communites in America in particular so that false alarm costs are apparently higher there than in the US - each false alarm is taken as a grievous insult by someone and maybe another couple of people are radicalized by it, which is a lunatic situation to be in when you think about it. So it's not easy to run the regime we've actually got right now - i.e., you've crossed the Rubicon of having everyone who flies or who rents trucks to people or whatever it is contemplating their mortality, and thinking about security as something they might have to personally contribute to - I absolutely agree.... Still, it's important not to try to mock and browbeat this stuff out of people: false alarm =/ cry wolf, false alarm =/ a pointless panic over nothing. False alarmees deserve sympathy and an apology but people raising the alarms in my view also deserve some sympathy: it's always easiest to do/say nothing and those who do/say something shouldn't be damned for doing so (even the friggin' Tories scored points off them - charming).

Posted by stephen glaister : 8/24/2006 02:18:00 AM

"Many thousands of Muslims etc. fly every day into and out of Britain without issue.... many of those doubtless strike someone somewhere as somehow worrisome and yet no alarm is raised, the initial cause of worry isn't backed up by anything more and so the matter is dropped and rightly so. "

Have you been flying in and out of the UK recently, or even observed the stop and searches at the tube stations? Many young Asians are getting stopped for little reason other than being darker-skinned. I get the strong sense from what you are intimating that it is perfectly reasonable for people to raise the alarm if an Asian person acts "suspiciously".

You also use the argument that the police and other arms of the government have been using to justify all manner of anti-demoncratic measures here: that the risk of harm is so great we should start being more paranoid and, in the governments view, repressive. It's exhibited every time John Reid, that odious little ex-Stalinist, gets on TV. Chip away at the judges decisions in following the law, extend detention, introduce "terror" laws and stop and search policies targetting Asian men, all with the excuse that we have never faced such a pernicious threat before...

How many ex-UK or UK bound planes planes were bombed by terrorists before this latest bout of fear? Pan Am Lockerbie is the only one I can think of. More planes have crashed due to bad weather, poor maintenance and pilot error, yet people do not question pilot's judgement, nor give planes a once over before flying, nor really worry about the weather.

Crime is everywhere, but I'm not going to cross the road because a black guy is walking towards me, even though more black guys are arrested for street violence in my area than other guys.

Quote: "But we *are* supposed to be taking air security issues seriously."

Why? The police just allegedly foiled a plot here. Seems like they are doing their job in catching criminals. After taking my shoes off, being on a zillion cameras, questioned and checked multiple times, I'm pretty happy with things once I get on the place.

Posted by Anonymous : 8/24/2006 03:10:00 AM

Anon, I'll just have a go at one of your points (but I don't hold any special brief for what Blair and Co. are doing and you may be right that there's really odious stuff going on - but it can be difficult to tell I think...see below):

Crime is everywhere, but I'm not going to cross the road because a black guy is walking towards me, even though more black guys are arrested for street violence in my area than other guys.

You might want to rethink that policy in general. Here's how I explain the basic point to my students:

U(crossroad&suspicousguy bad) = U(crossroad&suspicousguy bad) = -1 [just the cost of a slightly longer trip home]
U(don't cross& guy bad) = -10000 [mugged with possible brain damage]
U(don't cross&guy good) = 0 [you get shortest trip home]

Suppose Prob(Suspicious Guy Bad) = .01, that is you agree that the overwhelming majority of vaguely suspicious guys aren't up to no good....

then:

EU(You Cross) = -1

EU(You Don't Cross) = .01 x (-10000) + .99 x (0) = -100

How sure do you have to be that the unknown guy (say it's night so you really can't tell much - it's just some guy coming towards you vaguely suspiciously) is non-bad to make the expected utility of not crossing is higher than crossing?

EU(You Cross) < EU(You Don't Cross)
if Prob(Suspicious Guy Bad) < .0001

But we can *never* drive down the probability of an an arbitrary vaguely suspicious guy on a darkened street that low. Your mom was right - cross the frigging street. [pick any reasonable numbers to get a similar result]

Almost all people who are suspicious-looking are perfectly harmless, thus most of the time it's in fact in some sense a waste of energy/effort to take the precaution of crossing the street. Still it's rational to cross the street because of the asymmetrical costs of making a mistake: the cost of under-crossing (of not crossing the street if that guy is actually going to try to hurt you) is momentous, whereas the only cost of over-crossing (of crossing when in fact you didn't need to because the guy is harmless) is negligible - you have a slightly longer walk home. You don't think all men or all black men (or whetever it is) are muggers etc., you agree that almost all aren't.

[It's a mistake to try to read your beliefs/probabilities directly off your behavior, e.g. "You crossed the street, what a racist/sexist etc! You must think that perspon more likely than not to be a mugger etc" - is amistake since your behavior will reflect your values/utilities as much as your beliefs. The *same* is true at the level of institutional response. One of the most odious sides of "political correctness" is that, when it suits people, they'll throw away the basic moral about the dual dependence of behavior on both belief and value and claim to be able to read noxious beliefs directly off behavior which you can't do without lots of assumptions... I suspect that at least some of the bitching in england right now about security overkill reflects this point...]

Still, you should cross the street. Seriously. Your security isn't one good among many, it's the condition of all your other goods. (English people used to know this - the lines's from JS Mill!) I urge you to reflect upon your own irrationality at a personal level (your mom will thank me for telling you this, I'm sure of it). Until you think clearly about that simple case I'm afraid the rational calculus of the *much* more complicated institutionally mediated security case, will remain a closed book to you. You'll just be bitching in the dark and seeing beliefs that aren't there.

Posted by stephen glaister : 8/24/2006 02:59:00 PM