Diversity-friendly software at SXSW 2017

Shireen Mitchell and I talked about Diversity-Friendly Software at SXSW 2017. Here’s the video

And here’s the slides.

We covered a lot! Here a list of references, along with some notes.

Anti-patterns

• Machine Bias, Julia Angwin et. al., Pro Publica Software issue: algorithmic bias. Update, April 10: the Pro Publica Machine Bias series was a Pulitzer Prize finalist
• Digital redlining after Trump: Real names and fake news on Facebook, Tressie McMillan Cottom, Medium. Software issues: algorithms manipulable to favor “fake news”; mandatory automated race and gender identification (as opposed to optional self-identification) allows affinity targeting to penalize people identified by the algorithms as “black” and “woman”; reporting mechanism provides tool to harassers. Process issue: “real names” violations treated more seriously than violation of racism and sexism community standards. Policy issue: “real names” policies harmful to women and other marginalizd groups; see Geek Feminism’s Who is Harmed by a Real Names Policy
• Facebook Lets Advertisers Exclude Users by Race, Julia Angwin and Terry Paris, Jr., Pro Publica. Software issue: mandatory automated race and gender identification (as opposed to optional self-identification)
• Leslie Mac’s Facebook Ban Is The Latest Development In Racially Biased Censorship, Elizabeth Adetiba, Black Youth Project. Software issue: reporting mechanism provides tool to harassers.
  Updates: The DiDi Delgado’s Mark Zuckerberg Hates Black People has several more examples of this. Julia Angwin et. al.’s Facebook’s Secret Censorship Rules Protect White Men from Hate Speech But Not Black Children in Pro Publica documents Facebook’s (leaked) policy documents. Aaron Sankin’s How activists of color lose battles against Facebook’s moderator army, in Reveal. Jessica Guynn’s Facebook racism? Black users say racism convos blocked as hate speech from April 2019 updates the story, including perspectives from Carolyn Wysinger, Tanya Faison, Natasha Marin, DiDi Delgado, Ayo Henry, Brandi Collins-Dexter of Color Of Change, Steven Renderos of Center for Media Justice, Andy Marra, executive director of the Transgender Legal Defense & Education Fund and more.
• How a racist, sexist hate mob forced Leslie Jones off Twitter, Kristen V. Brown, Fusion. Software issue: functionality gives tools to attackers; lack of tools for people to defend themselves. Policy/process issue: Twitter didn’t enforce their terms of service against attackers.
• Another Six Weeks: Muting vs. Blocking and the Wolf Whistles of the Internet, Leigh Honeywell, Hypatia.ca, from January 2014, describes Twitter’s disastrously bad first attempt at implementing muting. Software issues: requirements badly specified, so problem not solved. Software process issues: not working with the people targeted . by harassment means that attempts to deal with the problem haven’t worked; lack of prioritization and investment in a key business problem.
• “A Honeypot For Assholes”: Inside Twitter’s 10-Year Failure To Stop Harassment, Charlie Warzel, Buzzfeed, describes many repetitions of this same pattern. Software issues: functionality gives tools to attackers; lack of tools for people to defend themselves. Software process issues: not working with the people targeted by harassment means that attempts to deal with the problem haven’t worked; lack of prioritization and investment in a key business problem.
• Airbnb Isn’t Really Confronting Its Racism Problem, Jamie Condliffe, MIT Technology Review.
• Preventing Discrimination at Airbnb, Ben Edelman, benedelman.org. Software issues: unnecessary information (names and photos) enables discrimination; no mechanism for people to test (lack of transparency)

Best Practices

The Open Source Bridge wiki has additional links on many of these topics.

Diverse representation, inclusive culture, equitable policies

• Project Include has recommendations on culture, employee lifecycle , metrics, and more.
• Diversity, Equity, and Inclusion in Science and Technology Action Grid, the White House Office of Science and Technology Policy, November 2016.
• How to build for a diverse and inclusive company, Jon Pincus: a summary of key takeaways from Tech Inclusion 2016
• HOW TO recruit and retain women in tech workplaces, Geek Feminism

Setting intention

• Dreamwidth Diversity Statement
• Django Diversity Statement
• Citizen Code of Conduct from the Stumptown Syndicate
• Adopting a code of conduct is an adaptive challenge not a technical one, Christie Koehler
• Microagressions in Everyday Life, University of Missouri handout
• Alex.js: catch insensitive, inconsiderate writing
• Django primary/replica patch dispute

Accessibility

• The W3C’s Web Accessibility Initiative has an Introduction to Web Accessibility, tips on Designing, Writing, and Developing for web accessibility, a summary of the Web Content Accessibility Guidelines as well as the full spec, Authoring Practices, and a lot more info.
• The A11y project: a community-driven effort to make web accessibility easier. Digestible, up-to-date, and forgiving.
• Web accessibility basics by Marco Zehe packs a lot of information into a 50-minute video

Flexible, optional, self-identification

• Best Practices for Collecting Names, Gender and Pronouns, TJ Warfield, TRANSform Tech 2017, is a very good short overview.
• Male/Female/Othered: Implementing Gender-Inclusiveness in User Data Collection, Finn Harker and Jonathan Ellis, Open Source Bridge 2015
• Designing forms for gender diversity and inclusion, Sabrina Fonseca, uxdesign.cc (see also Yonatan Zunger’s detailed reply on multi-lingual challenges)
• Disalienation: Why Gender is a Text Field on Diaspora, Sarah Mei
• Genders and Drop-down Menus and Designing a Better Drop-Down Menu for Gender, Sarah Dopp on Dopp Juice.
• Stop Mapping Names to Gender, by Os Keyes on Ironholds, is a short guide to why name-gender mappings produce incorrect results, are morally horrifying, and are usually unnecessary. Their The Misgendering Machines: Trans/HCI Implications of Automatic Gender Recognition looks at problems in facial-recoginition systems.

Emerging techniques

Gender HCI, Feminist HCI, Post-Colonial Computing, Anti-Oppressive Design, and Design Justice has a more detailed, and more recent, reference list.

Gender HCI

• The GenderMag Project’s site includes a downloadable kit and instructions for a gender-specialized cognitive walkthrough and a set of four GenderMag personas.
• GenderMag: A method for evaluating software’s gender inclusiveness and Finding Gender-Inclusiveness Software Issues with GenderMag: A Field Investigation, both by Margaret Burnett et. al., describe this work from a research perspective.
• Are you sure your software is gender-neutral?, Gayna Williams, ACM Interactions
• Gender HCI, Feminist HCI, and Post-Colonial Computing, Jon Pincus, Medium, summarizes research in these areas, and includes several videos
• Most work to date on Gender HCI has used a simple binary gender model. Gopinaath Kannabiran’s Where are all the queers? looks at some of the implications of this. Morgen Brommell’s 2016 AlterConf talk Imagining Radical Queer Futures Through Tech considers the possibilities of online spaces created by queer and trans people of color.

Threat modeling and harassment

• SXSW canceled panels: Here is what happened, caroline sinders, Slate. “Thinking about threat modeling and where I am “placed” or “sit” in a digital world helps me evaluate situations and figure out where I fall in the landscape.”
• A threat model approach to attacks and countermeasures in on-line social networks, Borja Sanz et al., in Proceedings of the 11th Reunion Espanola de Criptografıa y Seguridad de la Información (RECSI), includes harassmnt and cyber-bulling as threats in the overall model, although dos not go into any detail on them.
• OWASP’s Threat Modeling page is a decent introduction to the general topic of threat modeling, although doesn’t apply it to harassment

Algorithmic Bias

• ProPublica’s series on machine bias, by Julia Angwin, Jeff Larson, Surya Mattu, Lauren Kirchner and Terry Parris Jr., was a Pulitzer Prize finalist.
• Big Risks, Big Opportunities: the Intersection of Big Data and Civil Rights: a White House report
• What does it mean for an algorithm to be fair?, Jeremy Kun
• Big Data, Machine Learning, and the Social Sciences: Fairness, Accountability, and Transparency, Hanna Wallach
• Critical Algorithm Studies: a Reading List, from the Social Media Collective at Microsoft: the literature on algorithms as social processes.
• Fairness in Machine learning, a slide deck from Delip Rao, includes a short reading list

Originally published at A Change Is Coming.