Why Airport Security Can't Be Done FAST

Video thumbnail (Frame 0) Video thumbnail (Frame 1844) Video thumbnail (Frame 2933) Video thumbnail (Frame 4473) Video thumbnail (Frame 5474) Video thumbnail (Frame 6490) Video thumbnail (Frame 9193) Video thumbnail (Frame 10796) Video thumbnail (Frame 13017) Video thumbnail (Frame 14038) Video thumbnail (Frame 15034) Video thumbnail (Frame 16220) Video thumbnail (Frame 19158) Video thumbnail (Frame 20134) Video thumbnail (Frame 21926) Video thumbnail (Frame 23209) Video thumbnail (Frame 25045) Video thumbnail (Frame 28612) Video thumbnail (Frame 30873) Video thumbnail (Frame 32493) Video thumbnail (Frame 34091) Video thumbnail (Frame 38183) Video thumbnail (Frame 39401) Video thumbnail (Frame 42074) Video thumbnail (Frame 44471) Video thumbnail (Frame 46113) Video thumbnail (Frame 47913) Video thumbnail (Frame 48911) Video thumbnail (Frame 51028) Video thumbnail (Frame 52076) Video thumbnail (Frame 53324) Video thumbnail (Frame 54813) Video thumbnail (Frame 56038) Video thumbnail (Frame 57671) Video thumbnail (Frame 58979) Video thumbnail (Frame 59995) Video thumbnail (Frame 60986)
Video in TIB AV-Portal: Why Airport Security Can't Be Done FAST

Formal Metadata

Why Airport Security Can't Be Done FAST
Title of Series
CC Attribution 3.0 Unported:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Release Date

Content Metadata

Subject Area
Eight years after 9/11 TSA finally decided to fix their security system. But what has really changed? Homeland Security's science division has been busy lately, and is currently polishing up a project called FAST - Future Attribute Screening Technology. FAST, part of project MALINTENT, is a project of the Department of Homeland Security Behavioral Science Unit, which supposedly can detect whether you want to blow up the plane purely based off of biological indicators. While it was originally slated for completion this year, the project has been delayed due to many technical difficulties. Starting to smell snake oil? Basic statistics and common sense agree! Methodological flaws, numerous exploits and better uses of tax dollars will be discussed. Semon Rezchikov is an independent security researcher, programmer, and wetware hacker. He is spending the summer at MIT making cells do his bidding. In his free time, he can be found messing around with mathematics, playing clarinet, and furiously scribbling thoughts into his handy black notebook. Morgan Wang is an independent security researcher and analyst, history buff, classically liberal Libertarian, avid knitter and overall crafty person. Her hobbies include impersonating long-dead historical figures and obtaining suspicious amounts of fabric. Joshua Engelman is an independent security researcher and amateur mathematician. In his spare time, he volunteers for the ACLU and teaches fencing to small children. His current project is an autonomous beverage retrieval bot for his robotics team. He can be found lurking on the DEF CON forums, in school labs at odd hours, and directly behind law enforcement personnel wearing a monocle and trenchcoat.
Presentation of a group Software Computer hardware Electronic mailing list Information security Traffic reporting Computer programming
CAN bus Information security Traffic reporting Computer programming
Word Collaborationism Presentation of a group Googol Loop (music) Information Internetworking Collaborationism Bit Quicksort Mereology
Vulnerability (computing) Internetworking Projective plane Website Hidden Markov model Attribute grammar Automation Prototype Biostatistics Physical system
Reading (process) Thread (computing) Computer programming
Projective plane Video game Bit Lie group Information security Theory Resultant
Coefficient of determination Regulärer Ausdruck <Textverarbeitung> Confidence interval Multiplication sign Expression Mathematical analysis Flag Bit Quicksort Office suite Computer programming
Point (geometry) Slide rule Presentation of a group Copula (linguistics) Theory of relativity 1 (number) Price index Bit Time series Computer programming Coefficient of determination Robotics Statement (computer science)
System administrator Computer programming Chi-squared distribution
Reading (process) Slide rule Presentation of a group Multiplication sign Form factor (electronics) Computer programming Theory Hand fan
Presentation of a group Transportation theory (mathematics) Dependent and independent variables Zeitdilatation Multiplication sign Virtual machine Set (mathematics) Stress (mechanics) Mereology Wärmestrahlung Goodness of fit Bit rate Internetworking Bus (computing) Videoconferencing Energy level Series (mathematics) Logic gate Dependent and independent variables Algorithm Expression Stress (mechanics) Electronic mailing list Measurement Component-based software engineering Right angle Quicksort Videoconferencing
Direction (geometry) Software framework Value-added network
Area Observational study Information Bit rate Intrusion detection system Videoconferencing Software testing Field (computer science) Row (database)
Demo (music) Trusted Computing Lattice (order) Data conversion 2 (number)
Multiplication Different (Kate Ryan album) State of matter Multiplication sign Bit Right angle Biostatistics Physical system
Point (geometry) Link (knot theory) Observational study Decision theory Civil engineering 1 (number) Theory Line (geometry) Theory Numeral (linguistics) Order (biology) Cycle (graph theory) Identity management
Slide rule Estimator Presentation of a group Observational study State of matter State of matter Maxima and minima Video game Set (mathematics) Quicksort Laptop God
Slide rule Email Observational study State of matter State of matter Set (mathematics) Computer network Mathematical analysis Field (computer science) Word Uniform resource locator Set (mathematics) Software testing
Slide rule Context awareness Sample (statistics) State of matter Personal digital assistant Different (Kate Ryan album) Software testing Mereology Quicksort Control flow Position operator
Statistics Web portal Observational study State of matter View (database) Mathematical analysis Cyberspace Bit Mereology Biostatistics Mathematics Term (mathematics) Theorem Condition number Software testing Quicksort Series (mathematics) Bayes, Thomas Bayes' theorem Condition number
Execution unit Observational study Observational study Multiplication sign Sampling (statistics) Sound effect Port scanner Total S.A. Online help Number Sample (statistics) Bit rate Different (Kate Ryan album) Personal digital assistant Operator (mathematics) Normal (geometry) Right angle Quicksort Physical system
Mathematics Multiplication sign Electronic program guide Gradient Port scanner Zeitdilatation Instance (computer science) Mass Biostatistics
Biostatistics Context awareness State of matter Multiplication sign Decision theory Execution unit 1 (number) Biostatistics Category of being Bit rate Personal digital assistant Quicksort Physical system
Information Multiplication sign Physical law Mathematical analysis Student's t-test Computer programming Formal language
Coefficient of determination View (database) Personal digital assistant View (database) System administrator Touch typing Virtual machine Right angle Numbering scheme Exception handling Area Exception handling
System administrator Numbering scheme Exception handling Exception handling
Wechselseitige Information Information Different (Kate Ryan album) Projective plane Physical law Bit Data storage device Spiral Information privacy Curve fitting Biostatistics
Point (geometry) Pulse (signal processing) Standard deviation Freeware Observational study Multiplication sign Mathematical analysis Information privacy Information privacy Product (business) Bit rate System programming Website Information security Position operator Physical system God
Algorithm Presentation of a group Observational study Multiplication sign Moment (mathematics) Mathematical analysis Shared memory Port scanner Computer programming Biostatistics
Module (mathematics) Arithmetic mean Key (cryptography) Causality Execution unit Software suite Real-time operating system Software testing Price index
Module (mathematics) Demo (music) Computer network Cuboid Mereology
Point (geometry) Observational study System administrator Multiplication sign 1 (number) Bell and Howell Water vapor Online help Mereology Field (computer science) Hacker (term) Different (Kate Ryan album) Videoconferencing Traffic reporting YouTube Exception handling Dependent and independent variables Email Texture mapping Demo (music) Moment (mathematics) Tableau Bit Software Telecommunication Right angle Information security
I had lots of fun troubles with infrared cameras in preparation for this talk I had one that was like I don't know 10 years old I was going to use it it turned out they don't make the software for it anymore and you need this weird thing that they don't make anymore this hardware to actually get it to work so then that didn't works and we went to these other guys to try to rent one because apparently you can rent these things and that got that's there that thing is being quite silly now so I'm going to start so we were going to start with a minority report trailer
originally but that didn't work out because we called Fox and Fox was like no that's a bad thing to do but you know they're actually it's so fun fact that i discovered if you ever want to know every single advertisement in a trailer just ask them just ask the guys to let you use it in a presentation i send you back to list saying these are the clips from the movie you can never use because their advertisements so we have this trick we have we want to choose the Minority Report trailer because it parallels very closely the program that we did research on we're going to present to you guys about the program's
called fast because that's because everything else ever clever acronym and the program like Lake Minority Report
pretty much has precogs because it's supposed to go and stop murder by
looking into the future and it's supposed to be correct so we thought that so as we were doing research into this program we came up with a few questions because you know as in
Minority Report doubts in the end or arose so I'm same on Russia cough I'm
Joshua Engleman and we're going to present to you guys about fast so we'd
like to thank Morgan Wang who was our collaborator who helped us do a lot of the research for this turns out you can find all sorts of very interesting things on the internet by using google who knew unfortunately he she can't be here right now we'd also like to thank the US
Department of Homeland Security and Draper laboratories and Paul Ekman and the people it's pod and everyone involved in Fast and who developed these programs because with you guys we really wouldn't have a presentation to present on and you know dissect your methodologies on so we heart to us gee thank you very much so a little bit of a disclaimer before we start I'm not really part of Draper labs or I'm not really a Fed we're kind of all out of the loop on all this stuff everything we found we found by googling so if any of our information is not accurate that's because you guys don't publish anything so that's kind of unfortunate so fast we
were just the way we found out about fast which we were just googling around you know browsing the internet because we do that sometimes and we found and we went on the DHS website or he saw these two very interesting projects called you know the fast project in the hostile intent project and we read through them and our eyes focused on this one little
very interesting sentence the system will measure both physiological and behavioral signals to make probabilistic assessments of mal-intent hmm so when we
saw this we were like whoa that sounds really cool because just like biometrics
to fight terrorism really cool it's you know curing terrorism with science it sounds awesome and so the premise of fast is that you can tell people's malin you can tell the intention of people by using by scanning them with very advanced biometrics and if you do a little and so that's the whole idea so
we were like precogs they're real now apparently so we started doing research
and we found out that the genesis of the thread be the ideological threat behind the program started sometime around late 90s you know agent history it's really
old back
with Paul Ekman you see he looks a little bit like Napoleon so dr. Poliquin is famous for his research on deception theory and figuring out how we can tell who's lying and who's not any of you read Malcolm Gladwell's book blink I'm so sorry well he's more famously known as the inspiration for dr. Cal Lightman in the now-canceled Fox show lie to me his entire thing is this theory called micro-expressions when you try to hide feelings there are certain involuntary emotions little twitches on the side of your mouth things like that that are the same across every human being he to find out a little more about this he did this thing called the Wizards project he took people from all kinds of walks of life cops lawyers high school teachers and he tested their ability to reliably detect who was lying and who wasn't and high school teachers can't tell and she found that basically the only people who were significantly better than everyone else were Secret Service agents because they have some kind of training or it's a self-selecting we don't know which one of those it is because he stopped publishing when he says he realized that other countries were watching his work very closely and she didn't want them to be used against the United States this also results in no one else reviewing his research so we don't know if he's just making it up or not security woohoo so the Paul Eggman's research in
micro-expressions was the genesis of spawn which has nothing to do with dogs and it's not very cute in spot you have all these TSA agents that are trained in the analysis of micro expressions so if say you know going through the airport and you set off some sort of flag like you're a young Muslim guy who is reading the Quran and interested in Middle Eastern politics or you book a ticket a flight ticket that's one way and you check no luggage you might get referred to I think their behavioral what is it if your analysis yeah it's a behavioral analysis officer whoo-hoo Abby d/b/a oh who will go and question you in the nice little room and while he questions you he will go and pay very close attention to your face and try to detect any micro expressions you know to tell if you're evil and and you know so that the beginning this
sounds like it sounds like a really cool idea if this research actually works it seems very interesting so at the beginning you know people were kind of excited about this people are saying we should use it everywhere it's going to be awesome and but over time confidence the program started to wane a little bit
so that's not actually my idea that's a clip from that's an actual headline I'm not that mean although I did put in the dead dog so the thing was spot is that it refers so in the preliminary trials because it started right around nine eleven so it's been going on for about five years so these preliminary trials they referred I think 40,000 people to further screening and 300 of them are arrested which sounds awesome because like dude 300 terrorists caught that's pretty that's not that bad for you know a few guys who can read faces but the were no actual enemy combatants arrested because the 300 were drugged I think drugged all entirely drug dealers and people with people who lack documentation which so you know that's like oh look so you're arrested 300 people by questioning 40,000 extra and you know that's not at that point that's not very surprising so the program has not been incredibly successful and it seems that the big problem with the program is that really like human beings
even when they're trained are not good at detecting deception maybe some people are as per Pollock ones research but it's not something that you can really do very well so this isn't my slide this is the slide from one of Judy burgins presentations GT berggruen is someone who we think is very closely related to fast and so out of you know the relative semi failure kind of a little bit of spot you have people who decide that human beings are sucky at this and that we should get robots to do it which is always a good idea that coin is from also from her presentation as you can see government presentations are classy so spot was indeed not very sick is also
very expensive it's too it was 212 million a year I think of the Obama administration just increased its allotment to 286 million something like that and so you know in doughnuts because that's a great currency that's 160 dentists / federal employee per year or so there any feds in here you guys want a hundred sixty donuts just talked to some people or one doughnut for everyone in the United States on Christmas so yeah it's not in crits not incredibly expensive but it's not very good for 300 illegal immigrants so you
know you have this program and it wasn't very successful so you try and you try again because we believe in fast so fast
is this new program that is going to go and stop the terrorists from destroying
America and um so my this is also slide from 2d recruits presentation and I really like it because two of those pictures are from conspiracy theory books one of them being about FDR I'm just a big fan of this thing so what is
fast we haven't really explained exactly what it's supposed to be well it's a trailer like it's literally supposed to be in the form factor of a trailer that's something that they've emphasized many times because it's supposed to be portable so you could put it in lots of places not just like airports but like
Congress you know to check if there are any tyrus getting in or like your local
better know sports stadium I think we saw around the internet various presentations some suggestions of putting it on like Border Patrol or I think even public transportation which would be interesting you know get your mind right every time you go to a bus huh so what is fast all right so fast is
a series of pretty modern sensors the first list is this thing called bio lidar lidar is basically a fancy laser rangefinder and someone over the Defense Department has figured out how to use that to measure heart rate because you can totally do that the next part the one will be demonstrating is thermal camera which you use to measure stress responses after that it's just a video a video camera died level which looks at your pupils how red your eyes are agitation maybe even the micro expressions if the algorithms hit good enough but it could also just be used as a really good detector of hangovers they've also mentioned in a few of the presentations something to measure gate which is unique to each person but we haven't actually seen any data on that but we assume they're working on it somehow to be honest we haven't really seen much data on anything but we haven't seen any videos gate at all so I'm good yeah so two of the bushes behind fast are not only you know actually getting this detection thing working but also making sure it's fast because we have polygraph machines polygraph machines are great they're not incredibly accurate but they do sort of work but it takes an hour by a trained professional to run an exam this thing is supposed to be like five minutes it's supposed to be high throughput so they could actually use it in portal settings so you go into this trailer and now in
the middle of an unmarked van or something and those sensors I mean how
looks at you and this guy asks you a few
questions and it looks a little like this your
versa in your pocket or just inside back and have you looked in my direction keep your hands to the side and try and refrain from removing Ryan or frame from Lou the hospital will begin soon
show your IDs the first start now they're going to start asking our questions are you attempting to smuggle an explosive device into the expos are you from the local area do you see that I think she's blushing are you attempting to smuggle a recording device into the estoque so this is from we think we're not sure because again none of this is really published well this is from one of the only field studies of the test that they've done definitely an explosive at the expo so if you see that's the thermal camera video I think that's heart rate you play that illegally record any information at the expo and as you can see she failed the exam so that's pretty much it so if he'll
doesn't like you so contour you to that video from everything we've seen it seems that if Hal doesn't like you you
get to go through it again because you know we don't really trust computers all that much so right you guys know all about this so if he'll decides to have a glitch you get a second chance but if al doesn't like you again you got a meeting
with this guy do you see the glove notice the glove it's important and you
get to go into this room and have a nice little conversation with him and if hell
does like you you get off scot-free and it's wonderful so we thought that all this stuff was really really cool so we did it get working you don't know oh right so theoretically we should have a demo around here because we thought that
fleurs them forward-looking infrared cameras are cool so we're going to show you guys how to how little of a difference there is between someone who is lying and someone who isn't and how much of a difference there is between someone who ran around the room multiple times and someone who isn't but what all right so we'll get back to this in a little bit hopefully so that was the
system right and the system sounds pretty great I mean if you do get something like this working if you can in fact tell people's internal mental states by biometrics Oh brilliant wonderful thank you very much sir Oh or not huh okay so if you get this thing called fast work it's pretty awesome because you do in fact have a very effective tool for deterring terrorist attacks because well you can maybe invent new ways to fight to make columns but it's very hard to control the way your biometrics actually work it's very hard to control all these very fine things this would theoretically be a full system if it worked now
there is no there's not really too much
published stuff on this so we don't know maybe these the guys who are working on fast that Draper really do have it all working and it's awesome and like they're going to deploy it tomorrow and there will be no more terrorism but well we as we looked through we found a few methodological shady points and we'd like to point them out and so one of the
things that we found was that what so one of these I we found was that the theoretical the theoretical basis of fast was on interpersonal deception theory that's supposed to be deception and it was developed by buller and burgoon we're good as the person who slides we've I've been showing you guys and well the theory is it's fairly complex I can't really summarize all of it here because the cycle theory so it's like 18 different point 18 different assumptions and this is how we work when we lie and it's interesting but there are a few little sketchy things about the papers in which it was presented all the papers are called interpersonal deception theory Roman numeral so and most of them referred to a study in which you have two people so you have an interrogator and you have a suspect the suspect first end of interrogator first asks questions for which the answer is known in the suspect lines are truthfully and then the suspect and the identity interrogator starts asking questions for which the suspect may or may not lie and the interrogator knows which ones are definitely truthful and which ones aren't and it's always in this end though truthful ones and then on non truthful ones are always in the same order so there are the this study may have trained the interrogators to see clues about lying that we're not actually there it may they may have started to guess things based on the order of the questions a better criticism of it is in that paper which if you're in which you're interested in is quite nice it's civil but it's the tone of the paper is something like moon you did really awesome work back in the day what the heck it's a little sad oh so there is this thing so another
assumption that the fast guys have been making is illustrated in this slide which is also from burgers presentation which is that you have you have people who are benign in people who are hostile and then I people are going to be in one set of internal States and the hostile are going to be in another set of internal states which is fine but the benign people are going to be something like calm or excited and maybe it worse agitated the hostiles are going to be fearful and deceptive and angry intense because good people are never tense and so this is like it's all right it seems sort of reasonable people who are going to be deceptive are going to have a be in a slightly different mental state but the and so this might maybe extend to lots of things like drug dealing or people trying to smuggle stuff through but it won't necessarily extend to someone who is doing something for the glory of their country or their ideology or their God because you see if you're about to do something that is you know you're almost your purpose in life you're if you've got to really fulfill yourself there's no reason for you to be tense or fearful because they're going to be fairly calm you're good but if you've ever seen people who are about to do something really important they're there they they're quite happy there can be calling even if they're doing something incredibly dangerous they there are various studies that support this if you guys read into it and you know so you have these terrorists are about to blood estimates blow this comes up and make themselves happy they're about to get the ultimate prize and
that's um so that's a problem here's
another problem this is another slide and it's interesting because it's showing the way they've been constructing some of the studies the mock theft experiment is in which is one of those tests it was one of their experiments which we found some information on although no specific methodology on which you have people who steal something people who don't and they base they both try to say that they're innocent and they run through sets of detectors and you know they see what happens but one of the strange thing is as you see these words like two states innocent and deceptive and actors in screening location scenarios that's weird because that sounds like they're getting people to pretend to be innocent or guilty to pretend to be relaxed or agitated or over controlled and we thought you know that would be really strange because if you're calibrating your instruments to something like you know you go act suspicious you're not going to get anything that works too well in the field but in fact it seems like this is
actually what they have been doing as you can see it's a little troublesome yeah it's very early in the research but it looks very promising I tried to contact John Verico but he never answered any of my emails it was cut unfortunate so you have so you know we thought we were trying to really think that like they can't they can't be doing this I've got to be doing tons of field tests they've got to be designing their experiments is such a way that people are actually you know lying and caring about it but apparently we're in many
experiments that has not been the case if you see here this is another slide this is I think from the mock theft experiment and you have the same guy in three different states so the only way you would get something in that context is either if you're asking the same guy three different questions and he naturally tenses up like that or if you're telling the guy to pretend to be relaxed versus pretend to be controlled frankly by the exaggerated pneus of those positions i think is the latter so that's another problem and even Paul
Ekman the guy that no originally started spot and has been pushing through a lot of this a lot of these ideas thinks that this is just a bad idea because it makes sense if you're going to calibrate your instruments to detect something as fine as the difference between someone who's angry at their spouse and someone who's angry at America you need to be doing really really really good research and calibrating your stuff you're calibrating your instruments are some things that are actually relevant to that sort of question so you know we
don't want to criticize entirely want to be a little constructive so there are two things that they have to be that they should be doing in their studies to actually make these sorts of studies meaningful where the first of all the subject must be actually lying they can't pretend to be someone evil they have to actually lie about something and the second part is that the subject was actually care about getting through the subject must actually there must be some sort of stake in the situation because if there's no steak that's a very different situation from if there is a serious steak and getting through or not getting through so these conditions are it is non-trivial to replicate to get them to work in the lab in a psychology lab but it is possible you could get someone who strongly cares about some sort of political view or religiously or whatever to try to get through this series of the screening in which they are pretending to believe the opposite so I don't know get an atheist to pretend to be religious and stick them through this thing see what happens see if it can tell whether they're lying or not you can get someone who has issues that they personally care things that say they're embarrassed about like medical conditions to try to pretend that they don't have it that's not directly the same situation but that is closer than getting actors to pretend to the evil you also could do field tests they should be doing lots of fuel test on all this with field tests what they could do is they could just you know you have a lot of people going through the airport just have someone flashed someone flashed someone who's going through an FBI badge and say hi would you like to participate in the study have a knife you have to go through that door and say that you don't have a knife it's okay you won't get in trouble I'm a real I'm a real fed but and just go through so that would be a very realistic situation because not only would the person be agitated but the person would be trying to smuggle something through that they think is ok but they're kind of not really sure about because you're so you know smuggling something through so if they did lots of tests like that and that sort of test you know you could there's enough space in airports to stick a little trailer on the side they might actually be start getting some pretty good data so but those are all
methodological criticisms those are all things you can fix with a little bit of tweaking a little bit money there are other things that are fundamental to the problem of portals research and biometric biometric state analysis that you can't really get around so to about 200 years ago there was this guy named Thomas Bayes use a priest he's essentially the patron saint of statistics you're seeing now probably on some in some university math department where you're given a few probabilities you can figure out if one thing has happened what's the chance of something else happening so because this is all math stuff and it doesn't actually mean anything we thought to put it into a little bit more concrete terms so let's
say worst case scenario there's a hundred terrorists in an airport of a million people this is five times the number that were involved in the 911 operation so point zero one percent of the total or one ten-thousandth of the total are in fact pateros so if this is several years the future we have something called super fast which is eventually it can tell the difference between terrorists and non terrace ninety-nine percent of the time so sounds pretty good right sounds like we would catch almost all the terrorists unfortunately not quite because we would catch 99 terrace and while we're get through because it's ninety-nine percent accuracy on that side but then 9999 normal people non terrorists would also be set as Terrence so we'd catch over 10,000 people only 99 of which will be so if the system says someone is a terrace there's actually a really low chance that it's true there's actually a much much greater chance that they're not a terrorist and think this is with 99 percent we've nowhere near that right now yeah so the issue with this is that you can't really rely on the data that this thing gives you until it actually has an incredibly low base right until it actually has incredibly low false positive rate it would sure you know help narrow things down but nonetheless because it takes up so much I mean it decreases its effectiveness compared to say normal humans or just lots of secondary screening furthermore we don't
even know the actual you know accuracy rate of this thing because we've in a paper we saw 78% we've seen everything up to ninety nine percent although actually I spent all morning looking through for aware that ninety-nine percent is and I have no idea where I saw that I just remember it but because the thing is they really haven't published anything at all and that's kind of a problem so here's a study that they did this is I think the only field test of this stuff that we heard of they had 144 people who had no idea what was happening go through a fast scanner on the way to the edge of some sort of electronics expo I'm not sure if the era so don't expect it was an actual expert or whether it was staged I don't remember but you have these hundred forty four people and they go through and 23 of them were given a disruptive device that's all we know and yeah that was the study it was a field test does anyone see any interesting things about this no yeah him that guy yeah that's one thing but another slightly different issue is just you know the sample sizes you're going to if you're going to implement this sort of biometric scanning system industrially and scan everyone with it you need to be doing very large field tests on a very large numbers of people to make sure it actually works reliably the sort of thing it's not incredibly good because i
think is every human being a special in their own little way I told you guys in grade school it's true you know some people for instance have like heart
arrhythmia where you know they have these little attacks where all of a sudden they don't even get you know shorts short breath they're working perfectly fine but their heart beats a little faster in fact like that many times faster and you know so suppose someone who's going through a fast scanner all that happens that might be something interesting and there are lots of other things that produce such great massive changes in biometrics like I mean take fevers or hangovers or what else can you take that seriously almost any disease is going to change a lot of these things it might not change things like people dilations but what possibly yes indeed yeah actually three cups of coffee might do that too too yeah if you ever get our French cafe they serve you like cappuccinos and bowls so if you three of those yeah so that's a problem and that's not something you can really
get around well you can sort of you can collect and store the biometrics of every single human being in the United States and then individual to calibrate everything to everyone which would be nice you're probably gonna have to do a big chunk of that any way to get enough training data for the system to make sure it works reliably but and you know then you'd have a nice little ID system for every person United States unfortunately having the government having biometrics on all of you especially very detailed ones like your heart rate and your how fast how much you sweat is not an incredibly popular decision at the time so here's another
thing do you know what these things are I know any ideas yeah valium valium is a beta-blocker valium makes you calmer and makes your base and makes your excited states go closer to a basal state well that might help some people wouldn't it lots of other drugs have these sorts of properties and they're not even incredibly legal it's you know not actually a problem for someone to take some valium before they go through there have been cases of people who use Valley and regularly to try to you know calm themselves down and make themselves less suspicious in other contexts this is one that they could definitely use it in so
one question probably you're probably having how can this much collection of information be legal there's thing called the Fourth Amendment no unreasonable search and seizure they need to be able to say oh I saw blood coming out of that guy's truck so we
found this paper called the gill paper by a Lindsay Gill who was who was a law student at BU at the time who wrote about this entire program and a lot of the analysis that we did it was in rather better language in there so the
Fourth Amendment actually has a few exceptions one of them is called the plain view doctrine and has like things you can see one of them is the administrative search reception where it applies to everyone so the plain view
doctrine is basically everything you can smell see touch without any particular equipment something that anyone can do if they have the training even even the sniffer dogs count in this way however
most of the most people besides us don't have thermal cameras thus it's still technically breaking into their house without a warrant yeah there was an interesting case I think California where the Fed with the cops did a huge drug bust of a giant pot farm and the way they found it was by getting very expensive infrared cameras and just seeing the heat of the high phonics through the concrete and so you know they busted them and the case went to court and the court completely threw out everything because they were like this is not cool these guys couldn't have bought this camera on the consumer market so no so you know you think that TSA searches would also be illegal right because not everyone can buy metal detectors and not everyone can buy giant x-ray machines but there's this thing
called the administrative search exception which pretty much which was originally invented frankly for like people going into court houses to be searched and it pretty much says that if you're if you know about the search beforehand and it's not too too bad it's all so let me explain imagine you're a
bunny and you go into a park yeah it's nice and then this guy just runs out you
need like stop the terrorists I'm going to search you that's not cool that's not
a thing you're allowed to do but if you
stick metal the secretary the front of the park and you say all bunnies must remove the shoes before they go into the park or because otherwise the park may blow up that's all right that's perfectly fine the thing is you start getting a sketchy territory when you're
getting more and more and more intrusive searches like how getting a little bit difficult here because you know you're
collecting lots and lots of biometric information and people care about that people care about their health privacy there are fairly strong privacy laws in here and this thing is essentially a medical exam so we're not even sure like exactly how they're going to store the data that they collect a few different spokes people have claimed that they're going to take they're going to erase all the data but that can't be entirely true because I mean obviously if you get like arrested they're going to keep your data and if you and at the very beginning when they're ramping up the project they're going to have to collect a huge amount of data to make sure it works reliably on most people and yeah that's a little bit iffy we're thinking you
know so this is a privacy issue right this is pretty serious but on the other hand it could have also bees the solution to our health care problems because you see it's essentially a medical exam every time you go through the airport that's not too bad so like you know you go through and and how scans you he's like hey you have cancer go to this doctor that'd be pretty nice I think it's not too bad i mean i wouldn't mind like getting a throat checkup every time i went through and if you're getting it to an incredible reliability thing at that point you're probably doing enough analysis to be able to get things like this so in the end in our incredibly unprofessional opinion well we think that they could probably push push fast through because most people are entirely okay with giving up security for at least those are giving up privacy for at least a semblance of security but there are lots of interesting and fairly serious potential legal challenges that anyone with a crack team of lawyers could put up against the implementation of fast so there's another interesting
thing that we found while we were looking which is that the Israelis are doing the same thing but privately this is company called suspect detection systems or SDS which is essentially made entirely of X Mossad agents and they're making that booth which is called cogito it's like fast but it's smaller and more portable and you go in and ask you some questions while you're listening to go you ask the question System headphones it listens to your voice stick a hand in this standard the whole time and it's like taking pulse and temperature and it's entirely automated it's another product that's trying to do the same thing they claim on their website that they have a four percent false positive rate we've savory we've seen everything between four and eight and we still haven't seen any actual no peer-reviewed studies for any of this which is kind of unfortunate because god damn it guys can
you please share your awesome research with us all like this whole thing all this you know reading people's minds by biometrics and all this incredibly interesting analysis could be could very well be very valuable research the scientific community people I mean it's interesting I'm interested in having a scanner that tells how what people think that's pretty cool I'm interesting how the heck they pull that off I want to know that you know the research behind that is reliable before we stick one into every Airport and yet there really hasn't been very much published about all this kind of unfortunate it would be nice if the guys from draper published all this just like a paper or two saying hey yeah we ran this study this is how we ran it this is not exactly how our algorithms work but this is at least the things that we look for and so other people could test it out and see hey yes indeed this program works and then this would be fine because at the moment you know if if you're an outsider you're doing research on at all seems a little sketchy that's a problem so fast it was
fun doing research on everything and it took a while it was an exciting time we discovered a lot of stuff in the end like we honestly just really want to know what the heck they're doing with this thing because they're not being very public about it and so before we ended we wanted to we wanted to end the presentation on a lighter note because there was this well as we were well there was this trailer that they had for the fast program like not a trailer like the fast trailer but like a literal almost like a movie trailer and it was a little weird because it seemed like people well how do I say this it seemed like the you know we know we know where they're doing this at draper but it also seemed like they might be doing this in another research laboratory that we follow very closely so here it is
inside the screening mobile module a suite of real-time non-invasive sensors measuring the basal and physiological indications of mal intent or the intent
or desire to cause harm fact the key to
any successful cooperative testing trust and as our data clearly shows humans cannot be trusted ceiling at floors
provide a secure and easily reconfigurable means of installing equipment throughout the screening unit I give you panel lights up tomorrow
fully configurable the uniqlo floor
trailer design ensures accessibility for all their equal opportunity so now pay
attention to this part we might find this one interesting it can be wirelessly network to a single command
module that might be an interesting thing to look into then we box them up
and ship them straight to your doorstep
so you can protect the things that matter most Cave Johnson we're done here
thank you what Oh wonderful so our demo
is actually going to work now we think this is going to be cool so we'd like three volunteers one of which is winning willing to run around a thing for one or two times no no one yes sir yes sir yes ma'am come up so let me find Josh birdie at the water bottle I didn't put all right so what I'm going to do is I'm going to give one of them an object like we're going to give them water Bell a water bowl and a water bottle and pretend it's nitroglycerin unfortunately we can't find it so I'm going to give one of them a pen ten this pen is dangerous all right this pen is a dangerous weapon I can stab your eye out with it what so one of them is going to have so first of all I'd like one of you which everyone is willing to just run around a few times all right yeah sure just run around a few times that's all right this is fine guys so now these two are going to come up and I'm going to give one of them to spend none of you guys are going to know which one it is but they will and then we're going to go film them and I'm going to be the TSA agent asking them questions that infrared camera which apparently is going to work is going to show you guys heat maps or their faces you're going to see just how little of a difference there is between the ones that are lying in the ones that aren't about having a pen so when you guys come over here we're going to do this behind the table because the tables black and you can't see through black well I can with my camera yeah I was asking you to write a few laps not one I mean it's healthy come on guys this is def con we're not really being healthy at the moment so help is good so one of these guys or girls as the pen sorry um did I ask you to come I will need it to two volunteers and the third one I apologized I may have made myself a little unclear I'm sorry I feel really bad I'll I'm you can sit up to one of these three volunteers to pen the other one does not more realistic because there are a few terrorists so you guys all come up actually um so you guys all come up somewhere over there in front of the screen yeah go up there that no come on Josh seriously hahaha I hate technology what's up all right I'm gonna distract you guys for a bit so let's study guys fun things that I found out about fast well this whole thing was going on let's see um so hmm they were dead they're definitely Jesus Josh what are you doing not really all right so there are a few things that were amusing like the fact that fact that fox news was one of the guys that did the best coverage of fast it was really weird they had they were one of the few guys that mentioned the study that they or the actual field study there one of the few guys that um dude I don't have a software for this fine okay do you talk I Blaine every I blame everything on him oh yes all right what's up all right I'm going to keep talking so Fox News that this really awesome coverage like they had a those the place where I think all the copies of the trailer got up on the online that was the place where you had this study wow this is interesting ooh holy crap I got around the whole thing that's really cool i like this so so what else was there that was interesting um I think every single I think there were at least 20 videos on YouTube comparing it to minority report but we didn't really realize that until we had the idea ourselves very very original with that also yeah I told you all right so another thing that was interesting Fox asked us to pay them a thousand dollars a minute of trailer to show this just to show you guys the trailer so I mean it would probably as fair use but we were like screw it that else Fox is much easier to contact for copyright claims in valve is because like valve doesn't respond to anything while Fox you just call up News Corp's communications guy and he calls you back in five minutes saying yeah sure send an email to these guys you send an email to them and you're like yeah sure pay us a thousand dollars a second no you can't use it even though you're not making any money what else is there um I was going to use a camera that i borrowed from MIT but it was like from 1999 and that turned out to be a problem it's also much bigger than this one it was like this big what's up oh yeah so this is this is intense hacking right here if any if this doesn't work if anyone wants to come see us after all that the Q&A room will show you guys the camera and at least it will show you guys how cool it is and like will actually do a demo with a screen of this fig that's the big problem with screens they're just not big enough oh man keep talking all right so does that has anyone thought of anything else oh yeah question why should you talk to them well I mean you don't have to talk to them it this thing doesn't have to talk to you either it just scans you and says you're evil are you really actually not legally allowed to talk to TSA that's really cool not requiring yes not required that is a difference indeed yeah alright so that's the whole thing did the administrative search exception is act pretty much they're allowed to bar you from doing things and they can the government can force you to do things if you know about ahead of time even if it violates things like the Fourth Amendment and they're not too horrible anyone else yes oh man actually um I don't think they'd be very depends on how big the beard is like if it covers this part if it covers your cheeks at which point I think you need to shave might be pretty effective if it's just yes oh hey Josh can they go sit down you guys can go sit down yeah as long as you come to the speaker room afterwards so we can show some people pretty pictures of infrared things all right there's any one more question that i can answer here yes died all right thank you thank you everybody you