![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Entry tags:
Sometimes the racism is accidental: the problem is the system
So, there's a kerfuffle.
HP's automatic face-tracking software tends to suck at tracking dark-skinned faces. There's a video. The camera tracks the white person, breaks if the black person comes in.
Here's how I see it:
This is, ultimately, because of racism.
But it's only a rather indirect symptom, in this case.
See, the camera is controlled by the software that drives the face recognition algorithm.
The odds are, the software was tested by the team who were writing it. And they would almost certainly have tested it on themselves, because why bother getting other people in when you yourself have a face to practice on?
The first reason racism is to blame is because it's less likely that the team included a dark-skinned person, because of institutional racism.
The reason this can show up so easily in this situation is that you are dealing with a camera - and a low-grade webcam, at that. The Angry Black Woman points out that adjusting the settings can make the difference to the camera successfully tracking dark-skinned faces, and that different shades of skin tone can make a difference.
For those of you who haven't spent much time playing with cameras or thinking about how they operate: cameras function by the detection of light. Dark-skinned faces reflect less light. That's why they're darker. This means they are more challenging from the perspective of automatic camera controls (and good photography in general).
So here we come to the second reason racism is to blame.
Let's imagine, for a second, that we lived in an alternate universe where black and white roles in history were reversed. The modern world in this alternate universe would have systemic advantage to black people and white people would be second-class citizens.
Now, in this alternate universe, someone is writing the same face-tracking software. They come up against that same issue - darker-skinned faces are harder to track. They realise this instantly, and they adjust the program to compensate automatically. Because they immediately factored dark skin into their plans, instead of just not thinking about it and casually overlooking the substantial segment of the population for whom their program would be rubbish.
This HP webcam thing is not about HP computers being racist (the guy is mostly kidding, but still), or about HP programmers at any point being deliberately racist; this is about a systemic problem with overlooking the existence and importance of people whose skin is darker than beige.
(In the original video, it's also, to an extent, a problem with the guy being backlit, but the camera manages to compensate pretty well for that on the white woman, so it's not actually an excuse.)
HP are not the problem. The problem is in society. This is a symptom.
(Meanwhile, yesterday a Maori who is also a total world history buff told me to go see the movie Avatar. I said I wasn't sure I wanted to, because I'd heard it was hideously colonialist and offensive. He said not to think about that stuff, and just enjoy the movie.)
HP's automatic face-tracking software tends to suck at tracking dark-skinned faces. There's a video. The camera tracks the white person, breaks if the black person comes in.
Here's how I see it:
This is, ultimately, because of racism.
But it's only a rather indirect symptom, in this case.
See, the camera is controlled by the software that drives the face recognition algorithm.
The odds are, the software was tested by the team who were writing it. And they would almost certainly have tested it on themselves, because why bother getting other people in when you yourself have a face to practice on?
The first reason racism is to blame is because it's less likely that the team included a dark-skinned person, because of institutional racism.
The reason this can show up so easily in this situation is that you are dealing with a camera - and a low-grade webcam, at that. The Angry Black Woman points out that adjusting the settings can make the difference to the camera successfully tracking dark-skinned faces, and that different shades of skin tone can make a difference.
For those of you who haven't spent much time playing with cameras or thinking about how they operate: cameras function by the detection of light. Dark-skinned faces reflect less light. That's why they're darker. This means they are more challenging from the perspective of automatic camera controls (and good photography in general).
So here we come to the second reason racism is to blame.
Let's imagine, for a second, that we lived in an alternate universe where black and white roles in history were reversed. The modern world in this alternate universe would have systemic advantage to black people and white people would be second-class citizens.
Now, in this alternate universe, someone is writing the same face-tracking software. They come up against that same issue - darker-skinned faces are harder to track. They realise this instantly, and they adjust the program to compensate automatically. Because they immediately factored dark skin into their plans, instead of just not thinking about it and casually overlooking the substantial segment of the population for whom their program would be rubbish.
This HP webcam thing is not about HP computers being racist (the guy is mostly kidding, but still), or about HP programmers at any point being deliberately racist; this is about a systemic problem with overlooking the existence and importance of people whose skin is darker than beige.
(In the original video, it's also, to an extent, a problem with the guy being backlit, but the camera manages to compensate pretty well for that on the white woman, so it's not actually an excuse.)
HP are not the problem. The problem is in society. This is a symptom.
(Meanwhile, yesterday a Maori who is also a total world history buff told me to go see the movie Avatar. I said I wasn't sure I wanted to, because I'd heard it was hideously colonialist and offensive. He said not to think about that stuff, and just enjoy the movie.)
no subject
It's possible! And sometimes worthwhile. I'm desperately trying not to deconstruct Skins/i> at all (not that it's particularly in need of it), and also to not think about the possibility that I wouldn't like it so much if I hadn't started watching it right after rewatching some episodes of Torchwood S2 and catching up on S3 bc that show has just enough good things to make me irritated about all the bad things (in terms of writing more than -isms, actually).
no subject
It's quite interesting to visit a multicultural borough of London, for example, and see the white hairdresser (with its products) leasing alongside the salons providing for people of African or subcontinental complexions (with quite distinct products, advertising and clientele).
In this instance, I think there's a slight blurring of the boundaries between different types of equality going on. Demanding that consumer items will work the same way for all people isn't a normal part of the politics of equity.
That said, you're right in your observations. The fact that this issue was only observed after the device went to market suggests structural inequality in many different areas: among camera users, technologists, product marketers and those with disposable income for starters. That's where the true concern should lie.
no subject
Isn't that the kind of issue that ought to be addressed in beta testing? You know, before you release your product into the wild and discover that it doesn't actually work properly for an entire set of customers. (I know you address that in your post, but seriously, overlooking a technical detail that major is a huge marketing and PR embarassment for a company the size of HP, on par with Amazon.com's infamous "glitch" -- even if any bias was accidental, they better get out there and start apologizing now).
Avatar is, indeed, hideously saturated with Noble Savage stereoypes and "These quaint native people need a white guy to save them!". It's very, very pretty, with some nice military sci-fi tech details and beautiful alien flora, but only if you can get past the impulse to bang your head against the back of the seat in front of you every fifteen minutes or so (it's not just the race issues -- the plot is also generally cliche-laden and most of the characters are two-dimensional. And the alien culture are apparently the only four-limbed species in an otherwise hexapedal world, which just bothered me).