For every one good idea that actually becomes a paper, there are a zillion that became stalled collaborations, fruitless goose chases, and little balls of frustration that roll nowhere. When I become truly overdramatic, the ideas that go nowhere become indictments highlighting my lack of productivity, but other times they provide my ever-growing idea compost pile some new life. In the best moments, I hope that these puzzle pieces become something a little promising in ways that I could have never foretold much farther down the line. But other times it's totally fine to just leave them as they were, where the ideas find another person who might be able to make better sense of it. Maybe that person is you, my dear reader. The ever-wise Gillian Hayes writes about how she freely gives away ideas since the hard work is really in executing the project. In the spirit of that kind of sharing (and to document some of my own thoughts from the last couple months), here are some projects that just didn't make it into a paper or otherwise have me sort of baffled.
LinkedIn and platform accountability
As someone who loves social media research and has been saddened by the dismantling of the APIs (or access to other research data sources) of Twitter, Reddit, and Facebook, I was really excited about doing some platform accountability research on LinkedIn. In my head, there are more than a few foundational papers on the major platforms -- I can point to the papers on radicalization and YouTube, for example (and in my humble opinion it is everything that Becca Lewis writes). But I struggled with thinking of something similar for LinkedIn, and the user norms on LinkedIn are unlike most of the other platforms I'd ever read about or studied (I love some good LinkedIn cringe as much as the next person). The other reason why I was really excited about a project on LinkedIn is the pervasiveness of hiring algorithms as one of the formative examples about algorithmic discrimination in computer science and tech ethics classes, despite the fact that I never read papers about this kind of discrimination being audited on platforms like Workday, Monster, Indeed, or LinkedIn. I suspect that part of the reason is that many -- if not all -- of these companies are tightfisted about their data and any researcher audit creates legal liabilities for them rather than any value-adds (which partially explains why LinkedIn created an academic access program in 2019 and then completely stopped it). There are many different approaches that one could take with a platform project on LinkedIn: an anthropological study identifying cultures of cringe and community, a quantitative project on which accounts get surfaced in recruiter searches, a mixed methods approach that audits the different approaches that recruiters take when sifting through resumes. (For the last part, I went down a huge rabbit hole on Recruiter YouTube, which is itself a fascinating project.) At one point, I did get access to a LinkedIn recruiter seat (which is $10k/month per seat!) and played around with the different options, but didn't really get anywhere meaningful. Despite doing all the IRB work, I also realized that there wouldn't really be a way for me to do this project without compromising people's anonymity or otherwise fucking about with peoples' livelihoods. Combined with a lack of a clear computational path (via API) to do the kind of analysis that I might be interested in, the mixed methods approach slowly fell apart. There were a lot of thorny issues about research design that I just couldn't quite make progress on, and so this project slowly slipped away from me.
Email tracking and political campaigns
The second goose I chased came from my brilliant friend Eeva Moore, who was fascinated by the way that email data is bought and sold in political campaigns. As someone who loves thinking about the circulation of knowledge, this was captivating at the outset. I wanted to track how pervasive data selling was across political campaigns, the kinds of messages they used to convince voters who ostensibly already wanted to vote for them, and the extent to which candidates traded email lists in exchange for political capital (e.g., did Peter Buttigieg give up his email list as part of the deal to become Secretary of Transportation?). Email listservs are tremendously valuable assets in political campaigns, so it makes sense that they would be commodities that one could trade to pay down debt or to bolster a larger political campaign. To address this question, I ended up making a little database of major elections (e.g., all House, Senate, gubernatorial, and secretary of state races) and signed up for email updates from all the candidates using different email addresses in hopes of seeing how specific email addresses travel across political candidates or to a more centralized party database (especially in primary races where the party does not pool resources for all the candidates). In the end, it seemed like this was a question that could be answered by asking campaign managers rather than running a big analysis. The answer to the original question -- are emails from campaigns traded around for political capital? -- is yes. While it is difficult to truly quantify how pervasive this phenomenon really is (which may be its own project?), it is also a phenomenon that can be somewhat tracked by the FEC (see Elizabeth Warren, for example, and her campaign's payment for the Iowa Democrats email lists). Investigating FEC records is not perfect, and often these payments and transactions are not that traceable as they go under many names. But suffice to say there is still some stuff beyond the surface that I couldn't quite scratch.
Scaling responsible computing
I think rejected papers are the bread and butter of a good academic compost heap. Rejections are humbling and necessary, and in this case, I was incredibly grateful to the reviewers who did not let me or my coauthors off the hook (and did so rather kindly). It was one of those reviews that affirmed all of the things that you knew were wrong with the paper, but then also pointed out some more we missed (totally constructively!), which made you rethink the structure of the paper entirely. This paper was an offshoot of some of the pedagogical work I do at Mozilla with folks at the Responsible Computing Challenge, and I'm still thinking through how best to manage the programmatic aspects of thinking about pedagogy while maintaining it as an active part of my research agenda. There is a broader question for me, too, about what good pedagogical research looks like, as someone who does not have any formal training in the learning sciences or education.
Things I learned about doing research
These are the three main projects that made up the compost heap, though there are many other smaller ones that still haven't quite made it even to the compost stage (e.g., some gestating work on tech work and surveillance) -- they're little specks of yeast waiting for the right time and place to make a good, crusty baguette. I don't regret the time I spent on these projects, truth be told, but their relative failure -- at least for now -- makes me nervous as a first semester faculty member, that's for sure. Composted projects like these are diversions from the monumental task of writing two books almost by definition, and I would be lying if I said that I didn't feel guilty about not making more research progress. One day I'll be able to get to a place where I'm less afraid of these necessary components -- which is to say, failure -- to a fulfilling research career, but that day is not today.
One thing I really struggled with throughout the past semester is really finding good research questions: I was tortured by the idea of working on something (waves hands around) "important" on a vague theme that I found interesting, but sometimes I really came up short when it came to figuring out a good question to meaningfully explore (and for which I could collect data). I worked myself into a knot about my quest for good research questions and how I couldn't collect data in the absence of them, but Graham Jones very helpfully pulled me out of this rut by pointing out that research questions evolve as you collect more data, and that there is an iterative, dialectical relationship between research questions and data collection (particularly in more interpretivist traditions). My other advisor, Arvind Satyanarayan, also helpfully reminded me that this job is the most fulfilling when you work on things that genuinely interest you, rather than work that you think you ought to be doing. I definitely felt the grandiose, normative impulse to be Thinking Big Thoughts™, which is both misguided and unproductive. I think this semester was helpful in identifying where my priorities were (or should be), even when the projects don't always pan out.
Here's to another couple years of generative intellectual composting!