I found this article on Gateway Pundit. GP doesn’t get enough attention simply because it is run by the dumbest man on the internet and publishes the dumbest articles, so people tend to ignore it, except for its audience of very dumb people, but I thought I’d dredge up some of the sludge because this one pretends to be Science™. Why, just look at this illustration!
Image 1: Sample of the facial analysis results for one video frame out of about 120K frames analyzed.
That looks like a color glossy with circles and arrows, sure enough, and it’s probably got a paragraph on the back explaining what each one was to be used as evidence against Omar. It’s also got polygons! And multiple graphs! And lists! And colors! It’s just densely packed with barely legible information, and kind of reminds me of computer screens from science fiction movies in its generally uninterpretable and confusing layout, designed to bewilder with the awesome technology of it all.
You will not be surprised to learn that this heap of pseudoscientific, meaningless noise is being used to accuse Ilhan Omar of being a liar, since this is Gateway Pundit, after all.
An study of Ilhan Omar through facial, speech and virtual polygraph analysis shows Ilhan Omar is a pathological liar or just not being honest about her stories related to her immigration to the US!!!
Huh. Real polygraphs are useless and inaccurate, I wonder how reliable a virtual polygraph would be? All this “analysis” comes from a crank conspiracy theorist named Yaacov Apelbaum who has software that goes through video clips frame by frame and assigns emotional values to every facial expression, that counts blinks, “umms”, and pauses, and arrives at an interpretation defined by the user, that is, Yaacov Apelbaum. It’s garbage science, mangled by a biased kook, presented as fact.
Does this even count as real data?
Impressive multi-colored squiggles all over the chart, but the end result is a smear. She’s talking about an encounter her family had in Somalia with armed militia who shot up her home, and her stress about that event is being interpreted as proof that she is lying about everything. Worse, though, is that every change in her expression is twisted to mean she’s faking her story, rather than that she is recounting a traumatic incident in her past.
Remember this, everyone. If anything awful has happened to you, the only trustworthy to tell it to anyone is with a stony impassive face, no blinking, and with no pauses or stumbles or stutters, or it’s going to be judged by crackpots like Apelbaum or the goons at Gateway Pundit.
He’s presumably using some variant of “computer voice stress analysis”, which is the currently fashionable pseudo-scientific bullshit meter that you can use to railroad people you’re biased against. It’s everywhere. It has the advantage of being much cheaper, simpler, and more portable than an old-fashioned polygraph (not to mention shinier!), plus you can use it over the phone.
gotta use the Star Trek: Next Generation LCARS graphics to make it impressive.
Excuse me for being such a Trekkie and getting so distracted by the reference
If anything awful has happened to you, the only trustworthy to tell it to anyone is with a stony impassive face, no blinking, and with no pauses or stumbles or stutters, or it’s going to be judged by crackpots like Apelbaum or the goons at Gateway Pundit.
Aha! But then how could she be recounting such a horrifying event without any emotion! Obviously a fake story!
That catch-22 is a hell of a catch.
I was impressed by the graph. Something Tufte would call chart-garbage?
Evidence that unicorns do actually exist.
No photoshopping out the “Engineering” mode indicator, or “Long Range Scan” label? They might as well just use a replica of Spock’s periscope to force us to believe their science god.
Why do I suspect a performance like that would be interperted as “evidence” of being a trained spy / intelligence operative?
Possibly because it could be: Yonks ago, my work included the occasional interaction with the NSA. In the one and only face-to-face I can now recall (I could be conflating together more than one), that is almost exactly what the two NSA gooks were like, with the addition they always hesitated before answering (without stumbles, &tc), clearly thinking over their answer, and then spoke slowly. (No recollection about blinking, or the lack thereof.) It was quite possibly the eeriest meeting I’ve ever attended.
(And let’s not even mention trying to phone the NSA… especially from a “foreign” country…!)
That Star Trek-like graphic is “impressive”: Loads of numbers, almost no captions / labels, and the few labels which do exist either being meaningless (), mysterious (), or wildly inappropriate (). About the only thing missing is “Wrap Speed”, albeit there is .
The graph — as others have noted — is also very hilarious. I note that among the (extremely likely totally made-up labels), there is nothing like “Impassioned”, “Annoyed”, or “Confident” — all the labels can be construed to be negatives (, &tc).
The wingnuts did the same thing with President Obama, breathlessly hyping crackpots who cited “neurolinguistics” or other quack science to prove he was lying, or hated white people, or was secretly communicating with his communist United Nations overlords about the upcoming invasion of America, et cetera, et cetera.
All I can say in response is holy cow these rubes are terrified of brown people.
If they want real-world and world-class examples of lying, all they have to do is look towards their idol. Every time his lips move, he is lying.
Every time he tweeters, he is lying.
I have no idea if he snores, but if he does, I’d be very surprised if that isn’t also all lies. I’ve long been at the point where I don’t believe anything he emits — not even his name — without multiple independent generally-reliable confirmation.
At least Alice’s Restaurant used real science. –
Too bad for that judge with the seing-dog :-)
Loved the reference.
It’s easier to do this kind of nonsense with Rep. Omar because English is not her first language, so she makes atypical facial expressions when pronouncing certain sounds. I’m guessing the person who made these nonsense graphs doesn’t speak other languages, or they’d know the uncomfortable contortions necessary to make sounds (or produce reasonable facsimiles) that your brain never learned to make when you were first learning to speak.
Is this science-talk for “Mom! She looked at me funny!”
If I had a student hand in a graph made of hair like that — meaningless squiggles — I would fail them so hard they’d bounce. I had a memorable manuscript to referee in a respectable journal in which the authors submitted something like that. This is indicative of people who are not only not on the same page as others, they’re not reading the same book. They’re reading discarded gum wrappers. Or the bible. Or just watching teevee. They don’t share the same worldview. They failed something early on in life and are on a parallel track. Different sheet of the brane.
Yes. One of the exercises in our cell biology class is to simply make a clear, readable graph with all axes labeled, a good legend, and a text narrative that explains what we’re seeing. That graph would get shredded and used as a bad example.
If I were to guess, I’d say that facial expression was caused by being close enough to Trump to smell him.
What’s next? Craniometry?
The graph isn’t for proof. It’s a cherry on top. The proof is supposed to be the unusual expression in the top paragraph.
Presumably it’s pitched to people who are not familiar with hitting pause on videos. If you do that enough you start to see that just about any video of someone speaking has frames that show them with a goofy look on their face. It doesn’t matter how serious the conversation is, how sincere they’re being, or anything else really. If they’re speaking, there are frames that look goofy when you freeze them, remove them from the normal flow and present them as a stand-alone expression.
mnb0 wrote:
I’m counting on the IDW types bringing back the of phrenology.
lanir wrote:
I dunno, I imagine that’s just the icing on top. The people this was written for are looking for any reason to hate “The Squad”, and this is just red meat for them.
In other news, Mr. Apelbaum is the CTO and co-founder of the company that makes the software and also the person reporting that Ilhan Omar is a lying liar who lies all the time. I find little else about the company except some generic press releases (even their website is basically a placeholder). I wonder if they actually have any actual software, or if these are all hand generated by Mr. Apelbaum. And if the software does exist, how well did they work to minimize racial, political, and social bias into their data set. I wonder if all black people are designated as liars in the software, and if women are classified as stressed more often.
For anyone who is familiar with IT “architectural diagrams” that salespeople love to throw on slides, I suggest checking out the image that describes what “XRVision Sentinel AI” is supposed to be. It’s highly amusing in that it looks like something generated by someone who has seen those kinds of sales graphics masquerading as architectural diagrams, but doesn’t quite understand what belongs in them.
It’s the bible codes with modern machine learning! Looking for something in noise? You will always find it.
Is it just me or is that graph almost pure noise? I don’t have any of the mental or mathematical tools to understand these things, but I’ve graphed stuff at work to try to detect patterns in the data and outside of a periodicity that I assume is imposed by external factors (aka you can only speak so quickly or move the muscles of your face so quickly) I am coming up completely dry on this one.
Nothing corresponds to anything. There’s no pattern in it. Oh, yeah, and the axis might as well be unlabeled. Why should we care if there’s a peak at one spot if we don’t know when that peak occurs in relation to something else?
What’s next? Scary ways to create more fake news is my guess: https://www.youtube.com/watch?v=0ybLCfVeFL4
Looks like some of that data may have been found under a half a ton of garbage.
I’ve seen “data” like that in my work. It’s called random noise.
Random noise where the “Evasive” variable has been bolded and colored red so that it stands out to make it look impressive, but is no more amplified than the rest, where the “friendlier” variables have been removed, and on a scale that looks nothing more than arbitrary. It’s not just a bad graphic, it’s horribly manipulated and misleading. And will be swallowed hook line and sinker by anyone who want’s to agree with it. Ugh.
@ck:
I think they’d do even less work than you suggest. Why bother with building an actual program to generate flawed results when your whole program can just be a bunch of print statements? Or better yet your “program” is 10 minutes of private time with photoshop and a bunch of made-up labels?
lanir wrote:
Honestly, that is my impression of the screenshot. I can tell he certainly used some face tracking software on the image, but everything else is easily generated by common off the shelf software and an image editor. It does not look like actual software doing something as complicated as he describes, despite the supposed playback buttons on the image, and I can only assume it exists only to fool the credulous and venture capitalists (alas, I repeat myself).
@7
Yeah, been there, done that. The gist of the telephonic conversation was, “Monitor high side, got a message coming in for you”.
Laughably, a note of thanks for an alert that malware traffic was detected on one of our installation networks, as that rapid alert allowed mitigation before the malware spread and the report on the incident, to add to their knowledge base.
As for conversations with such folks, yeah, thinking of a response and avoiding mention of things that shouldn’t be discussed.
@PZ, why polygraphs are good for something! They’re exceptionally good at keeping polygraphers employed.
Despite missing Walker and Ames – repeatedly.