Part of it is the sheer hassle of repeatedly identifying objects — traffic lights, staircases, palm trees and buses — just so I can finish a web search. I also don’t like being forced to donate free labor to AI companies to help train their visual-recognition systems.
But a while ago, while numbly clicking on grainy images of fire hydrants, I was struck by another reason:
The images are deeply, overwhelmingly depressing.
CAPTCHA images are never joyful vistas of human activity, full of Whitmanesque vigor. No, they’re blurry, anonymous landscapes that possess a positively Soviet anomie. Here’s a typical one …
You can feel your spirits deflate just beholding this picture, can’t you? Even worse, really, are the jumble of bleak images …
Each cube here is a tone poem in melancholia. Looking at these leaden vistas of America makes you, slightly but noticeably, feel worse than you did before.
Why are these photos so depressing? What is it about their composition that is so enervating? I’ve been musing on this for a few months now, and had some terrific exchanges about it with Todd Pruzan and Emily Gordon, two friends of mine, on Twitter.
I think I’ve figured it out, and so now I present — The Six Reasons CAPTCHA Pictures Make You Feel Like Crap:
1) They’re devoid of humans
CAPTCHA images are pictures of the outside world, but it’s a world that is unsettlingly bare of people. This is likely for privacy reasons, which is a laudable motive on Google’s part. But it winds up making the pictures look totally postapocalyptic. Each CAPTCHA depicts a world blasted by a neutron bomb, where the objects survive but none of the people do.
When the photos actually do include humans, it’s somehow even creepier, because they are (again, for privacy reasons) shot from afar, and unidentifiable. So the only people you see are these weird replicants devoid of individuality. The guy in the red shirt in this next photo, bottom center: Where is he going? Why are we watching him go there?
2) The angles are all wrong
CAPTCHA pictures are often shot from extremely awkward angles — angles that we humans would never pick.
When you or I take a picture, we typically shoot from the view from our eyes. We’re usually holding the camera; our body is inherently involved in framing the view. But CAPTCHA photos were taken to train self-driving cars, so they were shot from the vantage-point of cameras mounted on Google’s Waymo experimental vehicles. That’s why the angles are just so oddly off.
As Emily put it …
It’s a subtle thing, but once Emily points this out, you can’t unsee it! Look at the angles on these picture here …
The top three images looks sort of normal. But the bottom six are just a bit strange. The center picture is not the view from which you or I would typically behold a car. Each crosswalk image is taken from a position that’s close to that a human-eye view, but off by like 10 degrees or something.
It’s like they fall into some sort of uncanny valley of camera-perspective.
3) They’re voyeuristic
These pictures all give off the icky vibe that comes from a lack of consent. Nobody in the photos expected to be photographed. These images are thus all inherently voyeuristic. Worse, because they’re also incredibly bland, the voyeurism feels all the more sordid and vapid. As Todd put it …
4) They look like crime-scene footage
Google’s CAPTCHA images are frequently grainy and badly focused. This is likely because, as Vox points out, Google has gone through most of the easy visual-recognition training cases, where the pictures were clear and sharp. Now they’re stuck with the hard stuff, which tend to be pictures of terrible quality.
This gives CAPTCHA images the low-res feel of a crime-scene video, as Todd noted in a tweet: “They remind me a lot of CCTV or dashcam video footage.” When you look at the pictures, it feels like you’re about to see some terrible incident.
These pictures! My god. They’re simply ashudder with suspense and dread. That taxi in the middle frame: What dread cargo does it carry, to what wretched appointment? The bottom right photo looks like something you’d see plastered on a WANTED poster. And good grief, in the bottom left: What looming terror is casting that fuzzy, Lovecraftian shadow?
5) The grids on the photos are an alien’s-eye view of the world
When you get those CAPTCHAs that chop up a single photo into sixteen squares, the imposition of those crisp white lines feels so disconcerting. It’s an alien view of the world: Behold the riddle of human existence. What could it possibly mean? By asking us to identify elements of an image that are sliced into pieces — “Select all squares with traffic lights” — CAPTCHAs turn everyday reality into a puzzle that no normal human would ever think of as a puzzle.
This is what’s so disquieting about the exercise. We’re being asked to parse the world in the visual-scanning style of an AI. Which, in turn, makes you feel like an AI, hunting for meaning in a baffling world.
6) There’s very little nature
Self-driving AI is worried about recognizing things in the built environment — stoplights, taxis, cyclists, fire hydrants, crosswalks. It doesn’t care about trees, or the fractal beauty of leaves, or flowers or birds or creeks or really any of the objects to which the human eye gravitates with delight.
This is why CAPTCHA photos are a nonstop brutalist slideshow of metal and concrete. It is as if someone took you on a tour of a lovely scenic town, but strapped you into horse blinders and forced you to stare only at fire hydrants.
(The one exception is, improbably, palm trees. I keep on getting CAPTCHAs asking me to recognize palm trees. I have no idea what’s going on there! Did, like, a Waymo car once slam into a palm tree?)
Here’s the thing, ultimately, about Google’s CAPTCHA images:
They weren’t taken by humans, and they weren’t taken for humans. They are by AI, for AI. They thus lack any sense of human composition or human audience. They are creations of utterly bloodless industrial logic. Google’s CAPTCHA images demand you to look at the world the way an AI does.
It’s no wonder we wind up feeling so numbed and depressed as we click through them, day in and day out.
Sometimes I try to put a playful spin on this, and imagine Google’s CAPTCHA system as a massive, surreal Art Prank the corporation is playing on the public. I mean, you could squint at this and see it as art, right? That’s what good art does: It shakes us out of normal perceptions and forces us to look at life in a new, slightly alien fashion. And hey, plenty of art uses automation and randomness to produce combinations that sidestep normal human meaning-making.
The problem with CAPTCHAS, though, is that we’re forced to look at these images, by a $1.8 trillion firm. It makes it a lot harder to find any sort of aesthetic intrigue here.
What has it done to humanity, to be forced us to regard these images, for years on end? Cloudfare recently calculated that we humans collectively spend 500 years, every single day, looking at CAPTCHAs. I’ve often joked on Twitter about the strange philosophical ratholes that CAPTCHAs lead me into. (“Wait, wait, is the pole that supports a stop sign part of the sign?”) And it’s now common to chuckle over the deep irony that we’re all forced by robots to prove that we’re human.
But these pictures? These pictures erode the soul.