PyCon 2016 in Portland, Or
hills next to breadcrumb illustration

Wednesday 1:10 p.m.–1:40 p.m.

More Than Binary: Inclusive Gender Collection and You

Anne DeCusatis

Audience level:
Best Practices & Patterns


Many people identify their gender in many ways. So why do we build systems to capture accurate gender information with a dropdown that only lists “male” and “female”? This talk covers why you might want to consider alternative ways of selecting gender for your users, a brief overview of the current best practices, issues addressed by my project Gender Amender, and why more work needs to be done.


At Meetup, one of the main ways we make our platform personal is by targeting recommendations based on gender. Unfortunately, the gender selection on our platform at the time I joined was a single dropdown, with options for “Male,” “Female,” and “Other.” If you don't identify in accordance with one of those, not only do we not target recommendations effectively for you, but we probably misgender you – not exactly a friendly or personal experience. ## Current practices Most forms which require you to select a gender require you to select only one option from a list of two or, if you're lucky, three or four. There are a few notable exceptions to this. Facebook, if you select “Custom,” allows you to enter text, and attempts to suggest matches to what you've typed so far beneath the text field. OKCupid allows you to select up to five gendering words from a list of their curation. ## #genderUX and why it's not enough; what current best practices leave out What's seen as the [current best practice][1] is to provide some sort of freeform input. It's [been argued][2] that this is the only way to ensure that everyone has a fair shot at representation. A tagging system has [also been suggested][3]. I think that freeform input is great, in theory, but what happens to the freeform input once it's collected? If you are Facebook, maybe you have enough resources to build a natural language processing system to determine which genders people might have from what they say about themselves. Otherwise, developers can only really ignore freeform data. A tagging system like OKCupid's is also a step in the right direction, but it's limited by the tags that developers come up with – a potential member still might not see themselves. The UX portion of genderUX is important, but I feel that collecting data that can be handled in an inclusive way once it reaches backend developers is as important to the user experience as what the field looks like. After several discussions, and a survey of existing inclusive gender options, I created my own library to try and bridge this gap. [Gender Amender][4] uses JavaScript and Python in order to maintain a list of relevant gender options for a member of your site to select from, and knows enough about each of the genders submitted that no natural language processing is necessary. ## Why collect gender at all, if it's so hard to do? The short answer: You probably don't have to. But if you value knowing about the genders of your users, hopefully getting accurate information about them will also be valuable to you. ## What Gender Amender is not for; why we still need other solutions in this space Gender Amender is best implemented where it's not prohibitively difficult to provide typing input, and is only suitable for systems which provide feedback in real time. Gender Amender is great for web forms! Gender Amender is not great for text messaging. [1]: [2]: [3]: [4]: