Slashdot Mirror
Schneier Has Something Good To Say About Airport Security
Bruce Schneier points out on his blog a proposal to use electronic randomizers at airport security checkpoints. Schneier writes there:
"I've seen something like this at customs in, I think, India. Every passenger walks up to a kiosk and presses a button. If the green light turns on, he walks through. If the red light turns on, his bags get searched. Presumably the customs officials can set the search percentage.
Automatic randomized screening is a good idea. It's free from bias or profiling. It can't be gamed. These both make it more secure. Note that this is just an RFI from the TSA. An actual program might be years away, and it might not be implemented well. But it's certainly a start."
In this case, the proposal is for randomizers that direct passengers to particular conveyor-belt lines for screening.
Nothing new here.
Had the same experience in mexico a dozen years ago.
Red light or green light.
But back then, there was a guy standing on a switch could just flex his knee to make additional selections if you looks particularly shady.
Sig Battery depleted. Reverting to safe mode.
Yes but if it's random surely they would need a separate belt for the foreign looking people thats more random.. Right?
Issue the TSA some dice?
I propose a device for random selection, consisting of a circular round object minted by our very own Federal Government that generates binary decisions with 50% probability, I can deliver these devices to the TSA at 100 units a shipment for a small price of $340,000 per shipment. I can have them delivered to every airport in the country within 2 weeks and we can implement this program by the Fall. They require no maintenance other than a 10 year service contract that adjust their randomness factor every year.
Any VC's out there?
'We are trying to prove ourselves wrong as quickly as possible, because only in that way can we find progress.' RPF
Is there evidence that profiling passengers based on appearance and behavior is not more effective than randomized screening?
You stand on a mat and it directs you to one of three different security lines, presumably to randomize the screeners incase you have one on your payroll.,
Any terrorist with a simple grasp of binomial theory could work out the number of terrorists to send through the gate necessary to achieve a 90% confidence that one of them gets through with the bomb, given only the relative probability of red vs. green.
So we must prevent binomial theory getting in the hands of terrorists.
I should use this sig to advertise my book ISBN-13 : 978-1501515132.
Some private contractor (probably recommended by Chertoff) will deliver years late, over budget, and after a terrorist gets through, people will discover that the light always turns green.
And that private contractor will most likely be named Dr. Gaius Baltar.
Nothing new here.
Had the same experience in mexico a dozen years ago.
Red light or green light.
But back then, there was a guy standing on a switch could just flex his knee to make additional selections if you looks particularly shady.
We had (still have?) this in Brazil. But i think it was only in the customs area, not really for security screening.
I don't want to misrepresent Schneier's position, but I've read articles of his in the past which basically say a profile is bad because it gives a way to avoid screening: avoid matching the profile.
Randomized screening may allow a single terrorist through, but something like 9/11 which required 19 guys means almost certainly one of them will be caught. If one is caught, you know to look for others.
Of course, the real solution is locking the cockpit doors and passengers who will kill anyone who tries to hijack an airplane.
I worked for T.I. when they were making LED watches in the Dallas plant. Security asked me to design a random search generator hooked to a switch on a turnstile leaving the assembly area. They could select the search frequency by means of a DIP switch.
"Do the Right Thing. It will gratify some people and astound the rest." - Mark Twain
More importantly, I guarantee that were such a system to be used in the US, it would include an override that watching agent could trigger a red light if he saw something suspicious... if only to ensure the continued employment of said agents. And such an override would result in profiling, negating one of the major advantages of the system.
I got modded down for some reason, but wow, the first time you get groped at the airport, it all changes from abstract theory to miserable reality. To feel the soft caresses of the male security guard as he brushes by your balls......
That is something that affects you.
"First they came for the slanderers and i said nothing."
My guess is that this creates a psychological game of chance that a would-be attacker might not risk; and perhaps searches are more thorough when personnel isn't having to rifle through *everyone's* stuff.
There are two things we know have strengthened security:
1) reenforced cockpit doors
2) passengers who know the deal and won't put up with any shit
We could make further *real* changes to improve security, like having highly trained and skilled air marshalls on every flight, hiring actual officers with actual skills to patrol airports instead of hiring glorified assembly line monkeys, searching bags strategically based on behavior and questioning ... but those things are just too expensive in the "wrong" way (ie.: they don't line the coffers of porno-scan manufacturers and the bureaucrats who do then favors; it would kill the job creation program for unskilled, slack-jawed mouth breathers)
just functioning brain cells and a lack of bigotry.
It's not bigotry to pay more attention by behavior profiling and using a little common sense rather than blind rule following.
Behavior analysis is free of racial implications.
Meanwhile "The Randomizer" pulls aside a four year old while letting through some sweaty guy with the shakes and an oddly bulging coat.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Profiling inevitably produces more false (usually an order of magnitude more) positives than real positives, and generally produces as many false negatives as false positivves. In other words, you're a lot more likely to spend your time searching someone for no reason than catch an actual bad guy, and as likely to let a real bad guy through as not.
And that assumes the profiling is done in an objective, unbiased manner. When human decisions are made as to who gets profiled, there will be bias, whether the humans doing it realize it or not. This, at least, eliminates that.
I'll bet, though, without reading TFA, that there is no thought whatsoever of this replacing any current profile based screening, only being used in addition to everything done now.
The oldest suicide bomber I can find was a 64 year old woman. And an youngest person arrested for trying to be one is 11.
Not quite your 4 to 80 range, but close enough that you look pretty silly and uninformed.
Frankly, I don't care how effective either is; just get rid of the TSA and stop harassing people, even if at random or by profiling.
Filthy, filthy copyrapists!
Is there evidence that profiling passengers based on appearance and behavior is not more effective than randomized screening?
Yes. MIT published a paper entitled "Carnival Booth" that demonstrated that random screening is more secure than profiling, essentially due to the latter's vulnerability to probing:
Carnival Booth: An Algorithm for Defeating the Computer-Assisted Passenger Screening System
A Lay Explanation of the MIT Research Paper [Carnival Booth]
Schneier on Security: Profiling
Proxy bombs are also difficult to screen for with profiles.
Thank you, Edward Snowden.
"Arguments from authority are worthless." —Carl Sagan
I recently listened to a Scientific American podcast where they did some "lab" tests to figure out whether or not someone was carrying a contraban package. They had five people walk through a room. One of the five had a contraban package. Random selection would produce a 20% rate of success. I believe they had a "hit" ratio of 30% using ordinary college students - which is slightly higher than random. (They also did a test with college students who tested high on the "psychopath test" and they were actually 70% accurate.) My main point, though, was that people do slightly better than random.
Here's the podcast (jump to 2 minutes in): http://www.scientificamerican.com/podcast/episode.cfm?id=psychopathys-bright-side-kevin-dutt-12-12-29
Wrong. Now that we secure cockpit doors and passengers are willing to fight back (neither of which violate anyone's freedoms), such hijackings are simply not going to happen.
That said, even if we didn't have either of those things, I believe freedom is more important than security, so toss your "happy medium" right in the garbage.
Filthy, filthy copyrapists!
With a monolithic culture, a purely random process makes sense... Could you please direct me to that imaginary monolithic culture? I want to move there and F*ck it all up...
Imagine wasting 70 percent of your time searching grandmothers, children, and the handicapped instead of searching the more likely demographic. It's pure idiocy to think profiling is a bad thing. If you are profiling to harass then yes it is bad but if you are profiling because the profiled group is doing all of the bad things then profiling is not bad. Only an idiot can't see such an obvious truth...
If you have a building with four entrances, and you have twelve guards to cover them, do you put three at each entrance, covering each as best you can, or do you put nine on one entrance you think is most likely to see an attacker, and only one on each of the other three?
If you're an idiot, you do the latter. If you're not an idiot, you realize the former yields maximum security, because as soon as you put all your guards on one entrance, it becomes far easier for an attacker to get in, they just use one of the other three.
If you can understand that, you should be able to comprehend why searching any particular demographic more (and thus, by diverting resources, means you search others less) makes you less secure, not more. As soon as your move resources into an uneven distribution mode, you open up exploitable holes, and you're a moron if you think your enemy won't exploit that.
Your "obvious truth" is the kind of thing uneducated people who don't really understand the problem say. Answers always seem obvious when you don't understand the problem -- but you could actually try educating yourself before spouting off idiotic nonsense...
"Convictions are more dangerous enemies of truth than lies."
It might not be security theatre but it is still security circus. Anyone think a dedicated terrorist of the muslim kind would care? If the light turns red just blow up then and there.
Killing a dozen people at the terminal, instead of three hundred people on board an airplane. You just described the system working quite well at its job (keeping the airplane safe). Was that supposed to be an argument against it? It's supposed to be the airplane safe. It's not supposed to stop all terrorism. The people proposing this are well aware of the fact that no matter how much security you have at the airport, people can still blow themselves up somewhere (and frequently do -- busy markets are actually more frequent targets than aircraft). Not a pleasant fact, but beside the point here...
"Convictions are more dangerous enemies of truth than lies."
it would include an override that watching agent could trigger a red light if he saw something suspicious... if only to ensure the continued employment of said agents. And such an override would result in profiling, negating one of the major advantages of the system
Wait, *how* is not allowing an agent who saw something suspicious to stop someone an *advantage*!?
"Sir, I just saw this guy typing a text message 'almost through - they'll never find it before I get on' - should I stop him?"
"No, that would be profiling. Just make sure he pushes the button."
Profiling gives criminals a way to game the system; if you don't look like the profile then you don't get tagged as a potential criminal (it also allows some unfortunate biases to come into play by the profiler). The solution, Schneier suggests, is a system that by its simple randomness, does not allow profiling or gaming.
Whether you agree with his logic or not, I strongly doubt that any such system would be allowed into common usage in US airports without an override. This negates the very advantages Schneier advocates. Whether this addition strengthens the overall system is up to debate (Schneier would argue that it does not), but the addition of a human override weakens those aspects that Schneier looks upon favorably.
Myself, I think all such methods are extreme overkill and that its far more likely that criminals interested in damaging the US with such attacks will strike at our practically undefended infrastructure, be it the huge AV fuel tanks at the airport, or any of the bridges or tunnels in a major city, or some toxic chemical depot in an urban area. Most of these are protected by little more than rusty chain-link fencing and an underpaid security guard and could cause far more harm than a simple plane crash. It's these weaknesses that terrify me far more than the presumed risk of some schmuck with a razor blade hijacking a plane. I'd rather they stop wasting money frisking passengers for penknives and spend it shoring up those vulnerabilities instead.
Or alternately, we could stop pissing off three-quarters of the world so they all don't want to blow us up. It's just whacky enough an idea to work!
If someone is trying to use force to hijack a plane, I don't think it is unreasonable for people to try to stop them, especially after 9/11. I don't see your point.
Filthy, filthy copyrapists!
http://en.wikipedia.org/wiki/Category:Terrorist_attacks_on_airports
There is a whole list and the most recent one was in Russia I believe.
But there is a problem, Dirty Harry justice. We don't mind how many people the criminal in a Dirty Harry story kills, just as long as they are blown away at the end. What hurts about 9/11 is that they got away with it. Had they been gunned down in a fight with the NYPD, they would have been considered losers.
You could taste some of that with the Russian school/theather hostage situation, lots of Russians killed but the "movie" ended with Muslims lying dead and that means case closed. Justice was served.
Go tell Hollywood if you don't agree. Humans are easily pleased.
MMO Quests are like orgasms:
You may solo them, I prefer them in a group.
Guess what, there are other nations, also well-off that are not hated by some many other people. There's a reason the US is targeted and it's not your aircondition - your posting is just a perfect example of why the US is perceived to be a "problem" in large parts of the world. If it wasn't so sad, your posting would be funny.
Windows 2000 - from the guys who brought us edlin
Passengers on flights coming from Willemstad into Amsterdam get checked 100%, because of the lax checks at Willemstad and the proportionally high amount of drug trafficking on this route.
Doing random checks on people not selected because they trigger certain alerts that make them suspicious makes it hard for customs/safety to get bribed and increases the chance the bad guys get caught. Once the bad guys figure out how not to stand out or bribe the guards, it's hard to catch them otherwise. This is why the random selector is better than having people do the random part of the selection. You want to check the poor African guy travelling alone to a rich country with a stop of one day in central America, because that's suspicious. But that doesn't mean that the mom and pop with a kid coming back from a 2 week holiday in Mexico can't be smuggling in a few Ks of cocaine as well. Having them press the button will make them think twice about the risk and it will probably even have a preventative effect in itself.
I was promised a flying car. Where is my flying car?
I am sick of this "we cannot profile" crap. The israelis profile based on "behaviour" as their main focus. I don't care how "trained" a terrorist is, they will always have a "tell" and they will always be nervous. Stop irradiating people and stop with this random violation of human rights in the name of political correctness. I am sorry but I am not interested in playing russian roulette with my life. That is exactly what this is because it is based on the premise that they can randomly find the terrorist.
Enough is enough. We have to start profiling based on behaviour and background checks and allow law abiding citizens and visitors to travel relatively unmolested. If we continue not profiling then the terrorists have won. Find and prosecute the terrorists and attempted terrorists and leave people who want to visit peacefully and spend money in your countries alone.
Doing evidence based profiling would probably be a good idea, but it is difficult, and to do it well would require a level of background checking and data mining that I think many would be uncomfortable with. Heck, even determining with high confidence just the identity of the 300 people to board the plane is a non-trivial task. Having a purely random component to determining who gets increased scrutiny is a very good way of dealing with the limitations of our screening resources.
From a game-theory perspective, if your attacker can figure out your profiling system, they of course would work hard to get their attack vector into the profile with the lowest chance of being subject to extra scrutiny. Since the defender knows this, it makes sense to pay more attention to those who's profile would generally indicate low risk. When you iterate this process enough, you come out at the end with the idea that equal chance for everyone is a good strategy. If one percent of all travelers had their IDs, tickets, luggage and backgrounds thoroughly checked, it could have quite a deterrence effect.
Of course the whole thing is pretty much pointless - the cost in terms of money, time, and hassle for air-security is way way out of proportion for the level of risk. We would get much better bang-for-the-buck dumping all that money into perfecting driver-less cars and cut down on highway deaths.