Ali Alkhatib

One of the busier intersections in the study of algorithmic bias has been in automatic judgments along race and gender, but decisions about how to frame and present information to people making decisions are immensely important, and from what I understand it seems that the evidence from studies isn’t consistent about whether something like a photo of the person involved in a given context helps or harms them. I’ve also been thinking about what the human component of AI-training systems might look like (for people evaluating appeals, or for people doing labeling tasks, for instance), and this paper seemed to approach that topic pretty directly. In this paper - “Do I Look Like a Criminal? Examining how Race Presentation Impacts Human Judgement of Recidivism” by Keri Mallari, Kori Inkpen, Paul Johns, Sarah Tan, Divya Ramesh, and Ece Kamar - Mallari et al. replicate an important study (and I think come up with problematizing findings, but offer some context to help it make sense), and go further by thinking about what this might mean when we design systems.

In statistical terms, “significant” approximately means “this probably didn’t happen by random chance”; in everyday language, “significant” means something more like “that’s a big deal”. Don’t speed past that sentence - it’s important. In academia we put untold numbers of hours into communicating our work to the people who have similar technical backgrounds to us - and use language in the same way - so when something we do pulls in huge numbers and garners public attention, we’re rarely well-prepared for it. I lost track of the number of times students and even professors expressed apprehension about journalists asking about their work when I was in grad school. There are unique, complicating factors in that case, but I think everyone’s anxious that they’ll say something and it’ll get misquoted, or quoted but out of context, or the audience of everyday intelligent people will read something like “significant” and take it in a direction we didn’t mean. One of the papers I read today - “Disseminating Research News in HCI: Perceived Hazards, How-To’s, and Opportunities for Innovation” by C. Estelle Smith, Eduardo Nevarez, and Haiyi Zhu - offers a sort of taxonomy of pitfalls, and how to try to avoid them.

It’s April 2020. There’s a pandemic sweeping across the world, and many of us are basically locked in our homes reading the news and consuming visualized data about infection rates, mortality rates, charts telling us what trajectories we’re on in comparison to other countries around the world, and trying to make sense of the magnitude of the trouble we’re all bracing ourselves for. One of the pernicious, kind of evergreen problems in HCI - or more accurately in data visualization - is how to communicate really out-of-scale data in a way that doesn’t totally lose the person trying to grok what’s going on. That’s sort of the whole point of data viz, after all, isn’t it? So this paper - “Du Bois Wrapped Bar Chart: Visualizing Categorical Data with Disproportionate Values” by Alireza Karduni, Ryan Wesslen, Isaac Cho, and Wenwen Dou - was really timely; this paper offers a historically inspired data visualization tool that (at least in some situations) helps people juxtapose bits of data with wildly different values in a way that doesn’t totally lose all meaning.

I alluded to this in the last post, but I’ll be trying to write a bit about each paper I tagged as interesting in the list of CHI papers that caught my attention. If you’re one of the authors in one of the papers I listed, and I haven’t linked to it (or if you’d just like to chat about it), please reach out! My goal isn’t to give a deep reading to papers. If you’re a first-author and you’d like to do a deep-dive conversation, I’d love to try that (maybe we could video chat, or just DM on twitter, or something), but my goal is to get across what sparked my interest in the paper. If you’re new-ish to the HCI space and you see a title and you’re thinking “why did Ali gravitate toward this paper?”, then hopefully these posts will help give some insight into that. My thinking is that you might have read a title, thought “that doesn’t sound like my thing”, and passed on it; if my blurb sparks your curiosity enough to go check out the paper, then I’ve succeeded. That being said, if you read my blurb and you’re like “okay, got it, but that’s still not my jam”, then that’s totally fine.

Academia is a difficult line of work to “succeed” in, where success is often defined as being noticed in a noisy field where people work for long hours in obscurity, sometimes feeling like you only surface to get smacked in the face with harsh rejections. Conferences are a rare opportunity to be up on a stage talking about something you did that other people collectively decided the field should be proud of. So it feels extra painful to imagine early career researchers who were looking forward to CHI, hoping for a rare opportunity to share their work and perhaps make themselves known to people they respect, only to have that opportunity taken away by a global pandemic.