You'd even think that a high proportion of the human race have darker skin and that product developers who weren't racist might have thought of that....
The problem isn't the dark skin. It's the combination of dark skin and strong backlighting with limited foreground lighting - which is *not* a normal condition for computer+webcam use.
Really, this is a facepalm kind of moment, but not because they never tried it on black faces (HP says they *did*, and it *does* work in normal conditions, and I kinda believe them on that because lying about it would be SPESHUL even for My Former Beloved Corporate Masters). It's a facepalm moment because an understandable failure mode in a nonstandard environment gives a *deeply* embarassing entirely unexpected result.
Good God. If only the developers had known that their product might encounter the rare Non-White-Person and had designed accordingly! I mean, it's not as if digital imaging can't adjust contrast, pulling images out of what, to the human eye, is apparently a pure black image.
But who could possibly have foreseen that the product might have been called into use around people who weren't white?
It works based on relative shading differences. The lighter your skin, the more light it reflects and thus the darker, in comparison, that the shadowed bits are.
It's an UNDERSTANDABLE technical problem. It's just also a funny one.
I'm not sure that the developers having not bothered to take this into account - ie white people as default humans - is understandable (or funny, but y'know).
In their defense, HP says it tested out in lab conditions. It's just that lab conditions (and, really, most computer user conditions) aren't strongly backlit and they didn't test *for that*.
Honestly, it's a bug. It's pretty clearly a case where software that works in other cases (or so the manufacturer says - it would be easy to test, if I had one, but I don't) but when presented with a non-standard non-tested condition fails in a way that's entirely consistent with the kind of software setting that's intended to stop it from tracking on non-faces in standard or darkened front-lit lighting.
Shorter: They didn't test for that because it's a weird-ass situation to be using a camera in. In this weird-ass situation, otherwise sensible software fails in a consistent and predictable way.
This statement makes me think you don't know too many developers. In my experience they tend to be a little . . . focused. Little things like real world applications of their product don't tend to make as much of an impression on them as meeting the specs and deadline they were given does.
I can find the results of the mistake to be unfortunate and unacceptable, while finding the mistake itself to be both understandable and, in some ways, funny.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
(no subject)
Date: 2009-12-21 06:59 pm (UTC)(no subject)
Date: 2009-12-21 07:20 pm (UTC)(no subject)
Date: 2009-12-21 07:33 pm (UTC)(no subject)
Date: 2009-12-21 07:45 pm (UTC)(no subject)
Date: 2009-12-21 07:47 pm (UTC)(no subject)
Date: 2009-12-21 07:48 pm (UTC)(no subject)
Date: 2009-12-21 08:27 pm (UTC)Really, this is a facepalm kind of moment, but not because they never tried it on black faces (HP says they *did*, and it *does* work in normal conditions, and I kinda believe them on that because lying about it would be SPESHUL even for My Former Beloved Corporate Masters). It's a facepalm moment because an understandable failure mode in a nonstandard environment gives a *deeply* embarassing entirely unexpected result.
(no subject)
Date: 2009-12-21 07:55 pm (UTC)But who could possibly have foreseen that the product might have been called into use around people who weren't white?
(no subject)
Date: 2009-12-21 08:10 pm (UTC)(no subject)
Date: 2009-12-21 08:10 pm (UTC)It's an UNDERSTANDABLE technical problem. It's just also a funny one.
(no subject)
Date: 2009-12-21 08:11 pm (UTC)(no subject)
Date: 2009-12-21 08:25 pm (UTC)Honestly, it's a bug. It's pretty clearly a case where software that works in other cases (or so the manufacturer says - it would be easy to test, if I had one, but I don't) but when presented with a non-standard non-tested condition fails in a way that's entirely consistent with the kind of software setting that's intended to stop it from tracking on non-faces in standard or darkened front-lit lighting.
Shorter: They didn't test for that because it's a weird-ass situation to be using a camera in. In this weird-ass situation, otherwise sensible software fails in a consistent and predictable way.
(no subject)
Date: 2009-12-21 08:30 pm (UTC)I can find the results of the mistake to be unfortunate and unacceptable, while finding the mistake itself to be both understandable and, in some ways, funny.
(no subject)
Date: 2009-12-21 07:37 pm (UTC)(no subject)
Date: 2009-12-21 08:11 pm (UTC)Note to HP: Better Off Ted is neither a documentary nor an instructional film.
(no subject)
Date: 2009-12-22 12:35 am (UTC)(no subject)
Date: 2009-12-21 08:21 pm (UTC)(no subject)
Date: 2009-12-22 12:45 am (UTC)