Face Analysis Face-Off: Part the First


If you are like me at all, you have tried to ignore the ceaseless array of online personality tests that infest Facebook like bedbugs in a Motel 6 mattress pad. You have also, if you are like me at all, absolutely failed in avoiding taking any more of those tests. My one saving grace is that I rarely encounter or even seek out one of these tests on my own. And so I take them very infrequently.

But just how do I take (or even find) such tests on those "infrequent" occasions? Obviously, because one of my friends or acquaintances on Facebook had already taken the test first. Of course, for me to even stop and click the "Find Out Now" tab (or other such shiny, candy-coated buttons that entice you to disrupt whatever flow you had going), the quiz's subject matter has to be just interesting enough where I wouldn't mind finding out what an entirely robotic system has to say about some aspect of my personality derived solely by the spare (and often intentionally) misleading information I have posted on Facebook. 

My profile picture is rarely of my own face, so that either negates or at least perverts or stunts the results of any tests based on photo usage, and I am too self-conscious or defensive to actually post the straight results I might obtain from personality quizzes based on multiple choice selections. There is also the scary proposition of exactly what the quiz sites are gathering from your Facebook page, and/or how they are going to use that information ultimately. Most such sites are now upfront with precisely which info they will be gathering once you approve the link between their site and your page, but it is still a daunting thought as to what the nefarious ends are as to why they would ever want it. However reticent I might be to take part in such entertainment after weighing the pros and cons, I will admit that it can be completely fun to take such tests, if only to kill a few minutes of the day in completely trivial and mind-numbing brain-rot. And so, should the mood strike me, I engage readily with them at a moment's notice without hesitation. 

A couple of weeks back, my friend Amber – who just so happens to be the wife of my writing partner and erstwhile pal who most closely shares my cinematic madness, Aaron – had taken an online game called "Which Nationality Do You Resemble Most?" The simple text for the test reads, "Let us analyze your face and discover which country suits you best! Click here!" The test takes a look at your profile picture and in just a handful of seconds, spits out the result, and if you thought such results might be fairly accurate, then you have never taken one of these stupid things before.

For Amber's go at determining her possible nationality, the testing site grabbed her current profile photo, which was a recent family portrait she and Aaron had taken to include the most recent addition to their little family of four, their newborn daughter roughly six months in age. (I am a baby-hater by instinct, but even I admit the little goof is pretty damn adorable.) Yes, Amber was in her own photo on her own page, but the test did not take into account the name or sex of the owner of the participating Facebook page, and instead latched on to the most prominent visage in the quartet: that of her husband Aaron (while still retaining her name in the reply)...



This result brought great joy to his wife, who posted the result with the only text being the word "Ha" followed by triple exclamation points. While Amber and Aaron have a good deal in common (as some married couples do), they most definitely do not look like each other at all. Nor are the pair even relatable in terms of height, Aaron being taller than even 6-foot me, and Amber being far shorter. As to the results, the test took a look at Aaron's face and declared, "Amber, according to our face analysis, your nationality is: Canadian." And to back up its findings, the Nametests site added a graphic that gave an accuracy rate of 85%.

Amber and Aaron are, like myself and my own wife, former lifelong residents of Anchorage, Alaska, which is as the crow flies, about 300 miles away from the Canadian border, and 500 miles if you are actually driving from our hometown to Dawson in the Yukon Territory, the first Canadian city after crossing said border. (I will not get into flight specifics here of traveling to other places to reach the Great White North in shorter distances; just trying to set a standard. And I use the word "city" to describe a place that shows up on the majority of maps without needing to use an electron microscope to find your idiotic town with a population of 46.) So, while we exist at roughly the same latitudes as Canadians, pretty much experiencing the same seasons, etc., in Alaska, we are still Americans. Still, I can see a measure of the stereotypical Canadian openness and politeness in Aaron's picture, and so the result is not as far off as it could have been.

So, let's move on now to those "could-have-beens," shall we?

Naturally, I just had to take this test, and for one very good reason: my current profile pic was none other than King Kong from his most recent film, Kong: Skull Island. I was dying to find out what the result would be, and honestly, I was ready for a wide variety of possible outcomes from this, including wondering just how racist the creators of the quiz might be. That I was itching to immediately publish the results is an understatement. And the answer?



Wow... Had the creators of this quiz recently seen the 1961 British horror film, Konga? That's the most prominent English ape flick of comparable sizing. (This is not taking into account international co-productions of recent vintage.) For real 'Mercans (aka the real snowflakes), the notion that Kong is not considered 100% American here might upset them greatly and get them reaching for their AKs. But, of course, a great cross-section of them also deny the American Civil War was primarily fought over slavery, so what the fuck do they know? They would also miss out on the fact that Kong in the original 1933 (and 1976... and 2005) version was straight up attacked and then kidnapped from his island home, transported in the hold of a massive ship to America, made to perform onstage in bound slave-like fashion for the delight of human beings, attempted to escape, and then was paid back for this mistreatment by being shot up by human weaponry, and finally, we are shown the image of his bloody, broken body strewn across the streets of New York. I am pretty certain Kong never would have considered himself an American, because being an ape (even a gigantic one with a serious jonesing for hot blonde women), he doesn't know from nationality. So, being called English instead of American is not going to upset him in the least.

This got me thinking, though, about how other monsters would come out nationality-wise via the same dopey test. To establish a base using a common image, I first jumped to Peter Jackson's version of King Kong from 2005. I thought the result might be roughly similar, but was told a very different origin story about this big guy:



Like any of these results, read into it what you will, but the 98% accuracy rating seems pretty definitive... doesn't it? It doesn't? Why a gorilla-like creature would be thought to have come from a continent entirely free from great apes is beyond me (though South America does possess a wide variety of what are termed New World monkeys), but then again, this test is not exactly giving us an actual biological analysis. In fact, I can't figure out what the hell it is using as a basis. The photo shows us three boxed points on each image as an example of how it determines the results, but the boxes are used in the images via a template, with said box trio configured in the same exact positions on each profile picture. If it is actually using those three points on each image, and if the selection is merely based on color points in the photo and not true physical characteristics, I do not know. (And, of course, this is all just very silly and never meant to be even mildly analyzed in reverse, like any test you might take on a social media platform...)

Aaron had seen his wife's post and taken the test himself, which he then posted on his own Facebook page. His profile pic, in contrast to mine, was of Godzilla. Aaron and I are both huge kaiju fans, so the fact that we were lined up with the two greatest figures of the genre at the same time in our profile pics is no real surprise. We often have header images of monster flicks and posters, or regularly post monster pictures for the heck of it on our Facebook pages. Myself, I was a bit jealous not having tried the Big G myself. While the 1933 Kong is my very favorite movie of all time, Godzilla is my absolute fave large-scale monster. (More on my fave human-sized "creature" is coming right up...) But Aaron had Godzilla in place already, and set down the comment point for the remainder of this exercise. Aaron's post had these words: "I am the only one this test has been accurate for."



So, why did Godzilla come out as Japanese? Is the coding for the test set to recognize certain pop cultures figures like Godzilla and know he is of Japanese origin? Or was there a more racial component to the result? One can't help but see one of the template red boxes in the image is directly over one of his eyes. And how is that portrayed? Is it reading his eyes as being slanted at a certain angle that certain citizens would not mistake for anything but of Asian origin? On which physical features does this test use to determine its results, if indeed it is doing so at all, and not just rolling the dice, so to speak? Whatever the algorithms are, the test seems more unsure of the Godzilla result than the last Kong one, giving it only 85% accuracy.

As I said, Aaron's post became ground zero for the rest of this discussion, especially since he had tagged me in the post itself so it appears on my own home page. I commented on his post with my first two Kong results back to back. Then, after a brief span, I switched out my profile pic once more. This required, it seemed, the deletion of the previous profile pic entirely, otherwise the Nametests site would continue to employ the old picture instead. After letting it stew for a while and cleaned out my cache as well, I finally got a result for my new profile pic, which just happened to be that more human-sized "creature" I mentioned a few paragraphs ago...



Once more, read into this what you will as far as how it determines the result. Or is it just swinging wildly, and letting the user read into the answer instead? It seems most likely. Here, I posted on my comment, "Boy, they are off across a whole ocean and a continent..." since the Creature from the Black Lagoon's lagoon is located deep down the Amazon river.

I next wanted to see what it would do with a completely non-humanoid face and figure, and so a few hours later (after clearing everything under the sun out of my browser again), I posted the result for another favorite monster of mine – this time of the prehistoric sea slug variety – a certain "Monster that Challenged the World". There was a small problem though...



The Nametests site could not make out the image I posted, even though I had been making all of my profile pics the same size. This time, it looks like they used a microscope to read the picture. For once and only once in this article, I have created a mockup of the result (though without wasting time in blending the images together in the middle). Here is the result for the "Monster"...



Since I am Swedish on my dad's side, this made me laugh uproariously, and especially because it is the most gratuitously creepy of the monster images I had used to that point. Later that evening, I ran into a bigger problem. The Nametests site was possibly getting tired of my shenanigans (though it is more likely to have been coincidence). Whatever happened, after I had switched my profile to an image of Lon Chaney, Jr.'s Wolfman character, the system kept recognizing the Creature from the Black Lagoon instead, even though I had deleted the image from Facebook and its history from my browser and cache. I even tried to get a Wolfman result on my phone, but wasn't able to until early the next day, after about 20 tries. And I was not displeased with the final version, though once the game let the Wolfman in, it was going to prove remarkably hard to keep him away...



Aaron, not to be outdone, went another route with his second picture. Being a dedicated fan of the Master, John Carpenter, he changed his profile pic to one of the aliens from They Live, after their hideous countenances have been revealed to Rowdy Roddy Piper with his special sunglasses. More possible tension is likewise revealed by the test result...



Aaron was worried what it said about the makers of the game and their possible contempt for Argentinians; I said it probably had something more to do with their contempt for aliens. I commented on Aaron's post that "I can see this is going to become a great, new way of life for us," but then the test kept doggedly hanging onto an old image after I deleted it. This time, it was the Wolfman, but instead of seeing the exact same picture, I started getting other nations as results. I had switched my profile to that of Bela Lugosi as Count Dracula, but the Wolfman returned, this time as...



At this juncture, I knew the game was rigged. First off, there is no way that a Wolfman could get trapped in a mineshaft. It was becoming more clear gradually that this test was just hit or miss. There couldn't be any reason to it. After several hours of waiting, I was able to finally get a hit on the Dracula picture...


Now this was cool... I was very accepting of the result (speaking for my own partial peeps), even though Romania is a good distance away from Sweden. (Lugosi was Hungarian, but was actually born quite near Transylvania in an area that was then part of the Kingdom of Hungary but is now part of Romania) They at least got the continent correct. I switched to yet another image, going the alien route this time, by using the crazy big-eyed and big-headed costumes from Invasion of the Saucer Men. However, while Dracula is most renowned for coming back from the dead over and over again, it was the Wolfman who was still getting in my way. And wouldn't you know it, he has now been revealed to be totally in it with Trump's cohorts...


In fact, he may have rigged the election for Cheeto the Toddler. I kept trying to get the Saucer Men to show up, but it wasn't until three days later that I finally got the Wolfman to disappear when I took the test. To my delight, I was to discover that my pal Aaron (in his human guise) and the Saucer Men were probably one and the same...



I rather gave up on the Monster Nationality sport at that time, mostly due to being distracted by health appointments and family activities. I did try the test once more on my phone a couple of days later to see if the result was the same, but the Wolfman reappeared yet again. He had gone back to his Italian roots for some reason (or no reason at all). Finally, this morning, I decided to give the game one last try, still using the Saucer Men profile pic. This time, the little buggers shifted across the Pacific and were now apparently related to Aaron's Godzilla...


This time, the test's algorithms were far more sure about the Saucer Men being from Japan than it was about Godzilla (96% to 85%), but truly, it was just all silliness down the line. It's really what you want from one of these games, especially when you are just killing time. But I would still really love to learn what the creators of this particular quiz did to get these results.

Meanwhile, something else happened when the Wolfman mysteriously returned for the fourth time...

[To be continued soon...]


Comments

Popular posts from this blog

Refilling the Flagon of Chuckles (or at Least an Extra Tall Improv Glass)...

Before We Take Off...

The Monster's on the Loose!!! Non-Chaney, Pt. 2: Werewolves Along the Wall

Ignoring the Ignoramus...

Guillermo Del Toro: At Home with Monsters at LACMA 2016, Pt. 2

Parallax