Racism is a free software issue. I gave a talk that touched on this at CopyLeft Conf 2019. I also talked a little bit about it at All Things Open 2019 and FOSDEM 2020 in my talk The Ethics Behind Your IoT. I know statistics, theory, and free software. I don’t know about race and racism nearly as well. I might make mistakes – I have made some and I will make more. Please, when I do, help me do better.
I want to look at a few particular technologies and think about how they reinforce systemic racism. Worded another way: how is technology racist? How does technology hurt Black Indigenous People of Color (BIPOC)? How does technology keep us racist? How does technology make it easier to be racist?
In the United States, Latinx folks are less likely to drink than white people and, overall, less likely to be arrested for DUIs3,4. However, they are more likely to be stopped by police while driving5,6.
Who is being stopped by police is up to the police and they pull over a disproportionate number of Latinx drivers. After someone is pulled over for suspected drunk driving, they are given a breathalyzer test. Breathalyzers are so easy to (un)intentionally mis-calibrate that they have been banned as valid evidence in multiple states. The biases of the police are not canceled out by the technology that should, in theory, let us know whether someone is actually drunk.
I could talk about for quite some time and, in fact, have. So have others. Google’s image recognition software recognized black people as gorillas – and to fix the issue it removed gorillas from it’s image-labeling technology.
Facial recognition software does a bad job at recognizing black people. In fact, it’s also terrible at identifying indigenous people and other people of color. (Incidentally, it’s also not great at recognizing women, but let’s not talk about that right now.)
As we use facial recognition technology for more things, from automated store checkouts (even more relevant in the socially distanced age of Covid-19), airport ticketing, phone unlocking, police identification, and a number of other things, it becomes a bigger problem that this software cannot tell the difference between two Asian people.
Black kids see 70% more online ads for food than white kids, and twice as many ads for junk food. In general BIPOC youth are more likely to see junk food advertisements online. This is intentional, and happens after they are identified as BIPOC youth.
Technology Reinforces Racism; Racism Builds Technology
The technology we have developed reinforces racism on a society wide scale because it makes it harder for BIPOC people to interact with this world that is run by computers and software. It’s harder to not be racist when the technology around us is being used to perpetuate racist paradigms. For example, if a store implements facial recognition software for checkout, black women are less likely to be identified. They are then more likely to be targeted as trying to steal from the store. We are more likely to take this to mean that black women are more likely to steal. This is how technology builds racism,
People are being excluded largely because they are not building these technologies, because they are not welcome in our spaces. There simply are not enough Black and Hispanic technologists and that is a problem. We need to care about this because when software doesn’t work for everyone, it doesn’t work. We cannot build on the promise of free and open source software when we are excluding the majority of people.