Daniel's been thinking about deception detection again. He points out we've done episodes on body language, concealment, reading people — dodgy landlords, controlling your own tells in delicate situations. And we keep circling back to the same research finding: people are generally terrible at spotting micro-expressions. Those involuntary facial flickers that last a fraction of a second. The science says most of us can't read them reliably. But here's the tension Daniel wants us to sit with. Israel runs a completely different kind of airport security — the human screening model, where trained personnel interview every passenger individually. Almost nobody else does it this way. And Israel wouldn't pour money into training people to detect deception and hostile intent if it didn't work at some level. Meanwhile, we've talked about TSA screening accuracy producing some disturbing numbers, and Herman, you've used the phrase "security theater" to describe what most of the world does. So Daniel's question is: is Israel fooling itself? Attributing success to human screening that actually comes from other methods? Or have certain specialized people cracked something about reading bodily cues? And if they have — doesn't that mean anyone can learn it?
I have used "security theater" and I stand by it. But Daniel's framing here is sharp because it forces us to separate two things that get mashed together constantly. The academic literature on lie detection — the controlled lab studies where undergrads watch videos and press buttons — that stuff finds accuracy rates barely above chance. Fifty-four percent, sometimes. And that's the literature people cite to say nobody can do this. But what Israel does at Ben Gurion isn't a lab task. It's not "watch this face and tell me if they're lying." It's a completely different procedure with a completely different target.
So the lab studies might be measuring the wrong thing.
They're measuring something real — baseline human ability on a stripped-down task. But the Israeli method isn't trying to spot lies. That's the first thing most people get wrong. They're not sitting there thinking "this person is lying about their luggage." They're looking for indicators of hostile intent, which is a broader and in some ways more observable set of behaviors. The interview is designed to establish a baseline for each passenger — how they talk, how they react — and then look for deviations when certain questions land.
Deviations from their own baseline, not from some universal "lying face."
And that's a crucial distinction. The micro-expression research — Paul Ekman's work, the whole FACS coding system — that was about universal emotional expressions that leak out involuntarily. Anger, fear, disgust, surprise. Ekman trained people to spot these at exposures as short as one twenty-fifth of a second. And he did demonstrate that some people — a small subset — can get quite good at it. But the Israeli security model isn't primarily a micro-expression reading exercise.
So what is it?
It's a structured conversation that runs maybe three to five minutes, sometimes longer. The screener asks apparently mundane questions — where are you going, who packed your bags, did anyone give you anything to carry, what do you do for work, where did you stay. And they're not just listening to the answers. They're watching how the answers are delivered. Does the passenger get flustered when the question shifts from "where are you staying" to "who packed your bags"? Does their response time change? Does their body orientation shift? The screener is building a picture of what's normal for this person and then probing for anomalies.
That's a lot more nuanced than "his eyebrow twitched, he's lying."
Much more. And the screener has enormous discretion to escalate. If something feels off, they can pull the passenger aside for secondary screening — longer interview, bag search, the works. The whole system is built on the premise that a trained human, given enough time and the right conversational tools, can detect when someone is unusually stressed about something.
But here's the thing Daniel's getting at. If the academic literature says people are bad at this, and Israel claims to be good at it, one of those things has to give. Either the literature is measuring something too narrow, or Israel is attributing its success to the wrong thing.
Or both. Let's talk about what else is happening at Ben Gurion that has nothing to do with human intuition. The profiling starts long before anyone talks to you. Passenger data is screened in advance. There are behavioral detection officers watching the terminal — not interviewing, just observing movement patterns, clusters, loitering. There are plainclothes officers. There's explosive trace detection. There's the fact that El Al has armed sky marshals on every flight and reinforced cockpit doors that were installed well before nine eleven. The human interview is one layer in a very thick stack.
So when Israel points to its security record and says "the interview method works," they might be taking credit for a whole system where the interview is just the visible part.
That's the skeptical read, and I think there's something to it. But I also think it undersells what's actually happening in those interviews. The screeners aren't random people off the street. They're typically young, often university students or recent graduates, and they go through intensive training. They learn to conduct these interviews in multiple languages. They learn what normal travel patterns look like for different passenger profiles. They learn to ask questions in a specific sequence and to watch for specific types of behavioral disruption.
What's a behavioral disruption look like in practice?
So imagine you're asked "what do you do for work?" and you answer smoothly. Then the screener asks "and what does your company actually make?" and suddenly there's a pause, a glance away, a shift in posture, a change in vocal pitch. One of those alone means nothing. But if the screener sees a cluster — multiple channels shifting at the same moment — that's a flag. The key is that they're not comparing you to a universal standard. They're comparing you to you, thirty seconds ago.
And that's trainable.
That part is absolutely trainable. There's decent evidence that people can learn to spot behavioral clusters and baseline shifts. It's not about reading micro-expressions in the Ekman sense — catching a one-twenty-fifth-second flash of contempt. It's about noticing that someone who was relaxed and chatty suddenly went rigid when you mentioned their hotel. That's a much coarser signal. You don't need special perceptual abilities to catch it. You need training, attention, and a structured framework.
So maybe the academic literature and the Israeli practice aren't actually in conflict. They're studying different things.
I think that's right. The lab studies typically isolate one channel — facial expression, or voice, or word choice — and test whether untrained or minimally trained subjects can detect deception from that channel alone. And the answer is: barely. But the Israeli method is multi-channel, baseline-referenced, and conducted by trained interviewers who control the conversational flow. It's almost a different species of task.
It's the difference between "can you identify this spice by smell alone" and "can you cook a meal."
That's not bad. And the screener gets to taste the soup as they go — they can ask follow-ups, circle back, change the order. The lab subject gets one sniff and has to name the spice.
But here's where Daniel's question gets uncomfortable. If the skills are trainable — if you can teach a twenty-two-year-old to spot behavioral clusters and baseline shifts — then doesn't that mean anyone can learn this? And if anyone can learn it, why are most airport security systems still so bad at it?
This is where the TSA comparison gets painful. The TSA tried to do something like this. They had a program called SPOT — Screening Passengers by Observation Techniques. Launched in 2007, trained thousands of behavior detection officers, put them in airports across the country. The idea was to spot suspicious behavior through observation and conversation. It cost something like nine hundred million dollars over the years.
And?
The Government Accountability Office tore it apart. Multiple times. They found no evidence that SPOT was effective. None. The behavioral indicators the TSA was using weren't scientifically validated. Officers were flagging people for things like "excessive fidgeting" or "avoiding eye contact" — which, as anyone who's been through a TSA checkpoint knows, could describe half the passengers in line who are just stressed about missing their flight.
Or people who are naturally fidgety. Or from cultures where eye contact norms are different.
The GAO said the indicators were not grounded in science and recommended the TSA stop using them. The program limped along for years and was eventually largely wound down. So you've got this stark contrast: Israel claims its human screening works, the US tried something similar and it was a documented failure. What's the difference?
I'm guessing you've got a theory.
Several. First, the Israeli screeners conduct actual interviews. They ask questions and listen to answers. The TSA's behavioral detection officers were mostly observing — watching people in queues, looking for predetermined indicators. There was no conversation, no baseline, no probing for deviations. It was checklist-based observation, which is much closer to the failed lab paradigm than to what Israel does.
So they were trying to spot the spice by looking at it.
Without even smelling it. Second, the Israeli system is selective. They're not interviewing every single passenger with the same depth — they're making rapid triage decisions about who gets a longer conversation. The TSA was trying to screen everyone, which dilutes attention and creates enormous false positive rates. Third, and this is the uncomfortable part, the Israeli system operates with a level of profiling that would be legally and politically impossible in the United States.
Say more about that.
The Israeli screeners are explicitly trained to consider ethnicity, religion, travel history, and political affiliations as part of their assessment. An American program that did that openly would be sued into oblivion before lunch. The TSA had to operate under strict anti-profiling rules, which meant their behavioral indicators had to be universal — things like "looking down" or "yawning excessively." And universal indicators are almost useless because the baseline variation in human behavior is enormous.
So one system gets to use the full picture, and the other has to work with one hand tied behind its back.
That's one way to put it. Another way is that one system is willing to accept the civil liberties trade-off and the other isn't. I'm not making a judgment about which is right — I'm saying the operational constraints are completely different, and that has to factor into any comparison of effectiveness.
But that still leaves Daniel's core question hanging. Is the Israeli method actually effective, or is it just theater with better PR?
The honest answer is that we don't have independent, rigorous public data on the effectiveness of the Israeli interview method. Israel points to its track record — no successful hijacking or bombing originating from Ben Gurion in decades — and attributes that success to the human screening. But as we said, there are so many layers in that system that isolating the interview's contribution is nearly impossible. The reinforced cockpit doors alone may have prevented incidents that the interview would have missed.
It's a correlation-causation problem wrapped in national security secrecy.
And wrapped in a strong institutional incentive to believe the method works. The people who run Ben Gurion security have built their careers on this approach. They train other countries' security services in it. There's an entire export industry around Israeli security methodology. That doesn't mean it doesn't work — but it means we should be cautious about taking self-reported success at face value.
Let me push on the other side, though. Daniel's point about investment is real. Israel is not a country that throws money away on symbolism. The security establishment here is pragmatic to the point of ruthlessness. If the interviews were actually useless — if they were purely theater — someone would have noticed by now and the method would have been scrapped.
I think that's fair. My guess — and it is a guess — is that the interviews are contributing something. Not because screeners are human lie detectors in the Ekman sense, but because a structured, adaptive conversation with a trained interviewer generates information that passive screening doesn't. Even if the screener can't articulate exactly what made them suspicious, the process of probing and escalating catches things.
It's a filter, not a detector.
Yes. And filters don't need to be perfect to be useful. If the interview process flags even a modest percentage of suspicious passengers for secondary screening, and the secondary screening catches things the interview missed, the whole pipeline works. The interview doesn't have to be accurate in isolation — it just has to enrich the pool that gets deeper scrutiny.
Which brings us back to Daniel's last question. If this is trainable — if the skills exist — can anyone learn them?
I think the answer is yes, with major caveats. The core skill — conducting a structured conversation while monitoring for behavioral deviations — is absolutely learnable. Law enforcement does versions of this. Clinical psychologists do versions of this. Salespeople do versions of this, though for different reasons. The training exists and it produces measurable improvement.
But?
But the context matters enormously. A trained screener at Ben Gurion has institutional backing. They can escalate. They have time. They're operating in a system designed around their method. If I teach you the same interview techniques and you try to use them on your landlord, you're missing most of what makes the method work. You don't have the information asymmetry. You don't have the escalation path. You don't have the backup.
And you've got a relationship with the landlord that complicates everything.
Right. The screener has no prior relationship with the passenger. They're not trying to maintain a cordial ongoing interaction. They can be direct in ways that would be socially inappropriate in daily life. The whole frame is different.
So Daniel's "anybody can learn these skills" — yes in principle, but the skills are only part of the equation.
The skills are maybe forty percent of it. The rest is context, authority, structure, and the freedom to act on your judgments without having to justify them socially.
There's also the question of what happens when people try to learn this stuff badly. The TSA SPOT program is a cautionary tale. Nine hundred million dollars and they ended up training people to be suspicious of fidgeting.
The bad-training problem is real. There's a whole industry of "body language experts" who sell courses to corporations and law enforcement agencies, and a lot of what they teach is pseudoscience. The "liars look up and to the left" stuff. The crossed-arms-means-defensiveness stuff. None of that holds up to scrutiny. But it's easy to teach, it feels intuitive, and it sells.
So there's a gap between what the Israeli screeners are actually doing and what the body-language-industrial-complex is selling.
A chasm. And I think Daniel's instinct to question the "people can't detect deception" narrative is correct — not because the narrative is wrong, but because it's too blunt. What the research actually shows is that untrained people, in artificial settings, using single-channel information, are bad at spotting lies. That's a very specific finding. It doesn't tell us much about what trained people can do in rich, interactive settings with the right institutional scaffolding.
And the Israeli case is interesting precisely because it sits in that gap between the lab finding and the real-world claim.
It's the most prominent real-world test of the idea that human screening can work. And we still can't evaluate it properly because the data isn't public and the system is too complex to isolate variables. That's frustrating.
It's also worth noting that even the Israeli system has had failures. The 1972 Lod Airport massacre happened at Ben Gurion. More recently, there have been security breaches and close calls that didn't make international news. No system is perfect.
And the system has evolved enormously since Lod. That attack is part of why the interview method exists in the first place — it was developed in response to specific failures. The Israelis learned that technology alone wasn't enough and built the human layer on top.
Let me try to synthesize what we're saying for Daniel. The academic finding that people are bad at detecting micro-expressions and deception is real, but it describes a narrow task. The Israeli interview method isn't primarily about micro-expression detection — it's about structured conversation, baseline establishment, and deviation spotting. Those are different skills, and they're trainable. Whether the method actually works at the level Israel claims is unclear because we lack independent data, but there are good reasons to think it contributes something real within a multi-layered system.
And to Daniel's last question — can anyone learn this — the skills are learnable, but the effectiveness depends heavily on context, authority, and institutional support. Learning the interview technique doesn't make you a human lie detector. It makes you better at structured observation, which is useful but limited.
Useful but limited. That might be the most honest summary of the whole field.
The other thing I'd add — and this connects back to the security theater point — is that even if the interviews are only partially effective at catching actual threats, they may serve other functions. Deterrence. Intelligence gathering. Creating a visible security presence that shapes behavior. A screener who asks detailed questions about your trip is also collecting information that can be cross-referenced later. The interview is a data collection tool as much as a screening tool.
And a would-be attacker who knows they'll face a probing interview might choose a different target.
The deterrent effect is real even if the detection rate is modest. Security theater isn't always useless — sometimes the performance itself does work, just not the work it claims to be doing.
That's a more nuanced version of "security theater" than the one you usually deploy.
I'm growing as a donkey.
Don't get carried away.
Fair. But the point stands. The TSA's body scanners and shoe removal rituals are theater in the bad sense — costly, invasive, and disconnected from actual threat vectors. A trained human having a conversation with you is theater in a different sense — it's performative, but the performance might actually generate useful information.
And it's harder to game than a metal detector.
Much harder. You can learn to walk through a metal detector without setting it off. Gaming a conversation with a trained interviewer who's watching your verbal and non-verbal responses in real time — that's a much taller order. Not impossible, but it requires a level of self-control and preparation that most people don't have.
Which circles back to something Daniel mentioned in passing — the idea of controlling your own tells in delicate situations. If the screeners are good enough, that becomes very difficult.
There's a whole literature on countermeasures — things people can do to beat deception detection. Controlling your blink rate, maintaining steady eye contact, rehearsing your story until it's automatic. Some of these work in lab settings. But in a real interview where the screener controls the conversational flow and can throw unexpected questions at you, countermeasures get harder. You can rehearse your cover story, but can you rehearse your reaction when the screener suddenly asks about the bus you took three days ago?
The unexpected probe.
That's the heart of the method. And it's why I think the Israeli approach, whatever its actual detection rate, is at least built on a sound psychological premise. People are bad at lying under cognitive load. The interview increases cognitive load.
So where does that leave someone like Daniel, who's trying to figure out whether any of this is learnable for everyday life?
I'd say: learn the principles, but be humble about what they can do for you. Understanding baselines, watching for deviations, paying attention to clusters rather than single cues — these are useful observational skills. They'll make you more attentive to the people around you. They might help you notice when something feels off in a conversation. But they won't turn you into a portable polygraph, and anyone who claims they can teach you to be one is selling something.
The modest version is probably the useful version.
Almost always.
Hilbert: The X-ray machine at Ben Gurion Terminal 3 is a Smiths Detection HI-SCAN 6040i. Dual-view. Runs about eighty thousand dollars per unit.
That's... specific.
Hilbert: The interview method gets all the press. The machines do most of the work. I installed four of them in 2014.
You installed airport security equipment.
Hilbert: Subcontractor. The Israeli company that won the tender needed people who could read the German documentation. I'd done two years at Siemens in Munich. Nineteen ninety-one, ninety-two. They paid me in shekels and the exchange rate was terrible.
The machines do most of the work.
Hilbert: Dual-view means you get two angles simultaneously. Operator sees the bag from the side and from below in the same pass. The single-view units they had before, you'd have to run the bag twice or flip it. Slows everything down. The 6040i cut secondary inspections by about thirty percent the first year. Nobody wrote a magazine article about it.
Because "better X-ray machine reduces bag checks" isn't as compelling as "trained interrogators read your soul."
Hilbert: The interviewers are fine. They catch people who act strange. The machine catches the thing in the lining of the bag that the interviewer never would have found because the passenger was perfectly calm. Calm doesn't show up on an X-ray. Density does.
So you're saying the credit is misallocated.
Hilbert: I'm saying the interview is the part you remember. Nobody goes home and tells their family "the X-ray machine at Ben Gurion was really thorough." They tell them about the kid who asked them twenty questions. The machine is better at its job and nobody thinks about it.
That's actually a really important point. The visible security creates the narrative. The invisible security does the work.
Hilbert: Always has. The screeners at Ben Gurion are good, I'm not saying they're not. But if you swapped the interview method for a different interview method tomorrow, the machines would still catch the same things. The interview is the garnish.
The garnish.
Hilbert: I don't know how else to put it.
I think you put it fine.
Daniel's question was whether Israel is deluding itself about the value of human screening. Hilbert's answer seems to be: partly, because the machines don't get enough credit.
Hilbert: The machines don't get any credit. They sit there humming and nobody writes a think piece about them. I read three different articles about the interview method while I was doing the install. Not one mentioned the scanner upgrade. The scanner upgrade was the whole reason I was there.
That's the sexiness problem in security. Human judgment is narratively satisfying. Technology is boring until it fails.
Hilbert: The 6040i doesn't fail. That's the point of it.
This has been My Weird Prompts. Thanks to Hilbert Flumingtop for producing and apparently for installing half the security infrastructure in the Middle East. Find us at my weird prompts dot com. If you enjoyed this, leave a review wherever you listen — it helps. We'll be back soon.
