2 ms·
The best we have are case studies of non-diverse teams reliably making products that aren't very good for people who look different, despite being highly educat
by joshuamorton 9y ago
The best we have are case studies of non-diverse teams reliably making products that aren't very good for people who look different, despite being highly educated and crossfunctional. The very common example is the Google Photos image search labelling dark-skinned people as gorillas (and more broadly facial recognition and machine learning biases towards light skinned people in a variety of ways). But there are others I've seen.
It comes down to needing an advocate. Unless you have someone who advocates for making your products inclusive, you won't end up with inclusive products, and the best advocate is almost always an engineer who says "hey team this product doesn't work for me". (Edit: I should add, this is true for pretty much any feature, not just inclusive ones, but its also true for them)
There's no reason a machine learning team needs physics PhDs. There is a reason that same machine learning team needs dark skinned machine learning PhDs.
>It is also common with people who are more focused about hiring the most qualified people rather than being politically correct or fulfilling quotas.
Right, and given that most people don't want to explicitly state that they're bigoted, often people who are racist and xenophobic claim to hold their views because they care about hiring the most qualified people rather than being politically correct. Perhaps you aren't, but its a worryingly common tactic.
- anon12345690 9y agothat's called user testing and focus groups, not forcing the developers to look different just to get a job
- joshuamorton 9y agoFocus groups are not an advocate. Focus groups are not a developer with a vested interest in the feature doing what they want. Focus groups cannot be as effective. I address this, but you seemed to ignore it.
- anon12345690 9y agoyou didnt address it, you proclaimed that products wont be inclusive (whatever that means) unless you have someone advocating for it, which you then describe as having someone who looks different. its not like people cant understand other perspectives, its not that hard, and focus/user groups would catch all the of examples you cite, just like any QA testing catches bugs and edgecases
- joshuamorton 9y ago>you didnt address it, you proclaimed that products wont be inclusive (whatever that means) unless you have someone advocating for it, which you then describe as having someone who looks different. And gave examples of that happening. Repeated, continued examples in the same context (facial processing). >its not like people cant understand other perspectives, its not that hard, and focus/user groups would catch all the of examples you cite But empirically they don't. Yes, they should. But they don't. This keeps happening, which would imply that the user groups aren't actually catching these examples, or maybe they are, but again without an advocate, a single complaint, or even some complaints, doesn't matter. So let me ask you, who is more likely to advocate for changes to fix a bug: an engineer who is directly affected by that bug, or one who isn't?