Now that the showcase is done and our papers are handed in, my reflections on the Freedom of Speech* project are taking a different form. The intensely action-oriented thoughts and feelings of the last six weeks were marked by a combination of stress and optimism: there was so much work to be done, but I always found my self talk leaning towards “it’s going to happen” (until the last day, when it switched to a tired but satisfied-enough “it’s going to be what it is”).

Now, without the deadline of a presentation and audience looming, I’m much less stressed about the project. But without the motivation of that deadline, I’m less optimistic about its future. As a recovering perfectionist, I am now okay with the idea that a project can be useful and valid without functioning 100% seamlessly. But I want the tool to work as well as possible so that people can use it without encountering faulty information or frustrating glitches, and that’s definitely not the case yet. But it’s hard to imagine going back to it without my whole group there to share in the process.

It’s lead me to wonder if the site is a beta version or a rough draft. To me, calling it a beta version implies that the same site will be reworked and updated, and the final product would look and function similarly, but at a higher level. Calling it a rough draft (which is increasingly how I feel about the Explore page) opens the door to many more revisions, not just functionality upgrades. For one thing, I’m not nearly as excited about the topic model results now as I was when I first ran the model, and I wonder if they really add much value to the Explore page. The topic model proved a fun exploratory analysis tool, but I think its merits are more limited than I first imagined, especially in the face of the manual data grouping that Martin and Joanne did to make the eras on the timeline.

Thinking about that now brings me back to my first semester in this program — I took a GIS course at Hunter College (Intro to Cartography and Geovisualization), and my major takeaway was that good maps are 10% GIS and 90% context and design. The actual “truth” of the coordinates and GIS layers is obviously important to get right, but the vast majority of communicating the argument or point of a map comes from everything else: colors, labels, statistical breaks, symbology, and maybe most importantly, context. That’s how I’m feeling about the topic model: however cool or interesting the results are, they’re probably only about 10% of the way towards making a good end product. I didn’t leave enough time to build out that context or consider how I would lead a user through the topic model, and in the end, I find Martin’s context-rich descriptions of the eras of free speech to be a much more compelling part of the site.

I think the topic model has potential, but implementing it in a way that really helps users to gain new knowledge about freedom of speech cases would require, at the least:

• including the full 10-word topic somewhere, not just a representative title, so that users can see the words that make up a topic
• including a quick definition of what topic modeling is
• omitting the topic model data point for cases where it adds little to no value, as in cases where the top topic is a mix of unspecific words (this is true of a lot of cases) or cases that don’t have a strong one or two topics

Those are my thoughts for the future of the Explore page topic model, which in truth probably isn’t even top of my list to fix up (looking at you, wrong case showing up in the case modal). As far as deadlines go… I’m definitely taking the coming week off. Five fully online courses later, I maybe feel more beat than I ever have at the end of a school year and am 100% ready for the semester to be over: for some summer, some days outside without opening my computer, and some sleep.

After that, maybe I’ll hit up our group’s Discord server to run my tech edits by them, set a one week deadline, and see if I can squeeze a few more days of “it’s going to happen” energy into the project.

I think I have 5 personal blogs left to write, which pretty much coincides with having built an almost-fully functional website in the last… 3 weeks. I honestly haven’t had much to write since my topic modeling posts — the update each week is that I am neck-deep in coding, which is a state that I alternately love and hate.

I hate it because it involves hours and hours of sitting, often so engaged in a problem that I forget to move or drink water for an unreasonable amount of time. I hate it because it leaves my body restless and my brain knackered. I hate it because the emotional landscape sometimes involves great swathes of frustration with just pinpricks of triumph before I turn to the next tangled problem.

I love it, though, because it really is like learning a new language. It’s a language of functionality and precision, and of breaking the problem I want to solve down into a set of tasks that a computer can accomplish. I’ve enjoyed learning over the course of my degree that there’s rarely just one way to do that. There may be one way that’s the most performant, one that’s the most mobile-friendly, one the most visually pleasing, or one the easiest. Beyond that, there may be one way that accomplishes what you really intend and one that seems logically sound but ultimately fails (for example, select all the women in this dataset and select all the not men in this dataset are equally easy tasks for a computer but certainly not equal questions for a researcher — intention is key, as is a data structure that allows you to ask exactly the question you want).

Sometimes these different  “best ways to break down a task” overlap, sometimes they don’t. Sometimes I have the knowledge and bandwidth to make an informed decision or improve on an old way of doing things, other times I’m so relieved to hack out a way that just WORKS, I don’t even think about the others. That’s definitely a plus one for collaboration, since looking at other peoples’ code often teaches me about the ways I haven’t chosen.

I love coding because I get to not only think about all of that, but actually do it. I hope it never gets old for me to write out text commands and see them bring dynamic shapes and colors and movement and information to life online.

Here’s Part 2 of my walkthrough of the topic modeling process that makes up one part of the Freedom of Speech* project. (Part 1 here.) I’ll cover how I improved the topic model after showing the initial results to my group members.

At this point, I’ve run three topic models with the goal of increasing the clarity and specificity of results each time. The number of topics seems about right, and my main focus was on removing words that didn’t contribute to the topics as knowledge-producing documents. For example, here are the topics from the first model:

It was an exciting start, particularly since topics like 1, 10, and 11 immediately speak to (un)protected speech themes we’ve been talking about for weeks: broadcasting/advertising, obscenity, and communism/McCarthyism. Another part of the fun is also seeing words that expand or clarify our understanding of a topic. It’s not surprising that the words “foreign” and “control” are in the communism topic, or that the sexual/obscenity topic includes the word “children.” It does, however, help to solidify our understanding of the motivations of these battlegrounds: communism in speech matters because the U.S. government cares about the impacts of foreign influence; obscenity in speech matters because the U.S. government cares about protecting the rights (and souls) of children.

On the other hand, topics like 7 and 14 are essentially useless for telling us about themes in the cases, since for the most part they just include high incidence court-related words (“plaintiff”, “justice”, “court”, “district”, etc)  that aren’t on a general purpose stop word list. Topic 15 also includes several Justice’s names: Blackmun, Brennan, and Rehnquist. I took these words out and ran the model again:

These results were better, particularly since every “unhelpful” word we remove from the model makes room for a more interesting one. For a perfect example, I took out the word “statute” and it’s replaced in the obscenity topic (now called V18) with the word “minors,” a much more descriptive word for that topic. Topic 10 distills a clearer picture of the topic about broadcasting regulations, compared to its corresponding Topic 1 in the first model.

New topics also appear: Topic 16 shows a new topic about fraud/soliciting/telemarketing. Topic 17 brings together the words “flag”, “symbol”, “peace” and “group.”

BUT, another limitation of the first and second models arises: Joanne pointed out that Topic 14 contains court-specific language that makes an interesting group of “court verbs” but doesn’t help us with thematic topics. So, one more time! Here are the topics in the third model:

Topic 11 shows us a “libel” topic, an important battleground of free speech that was missing from earlier models. Topic 13 also brings out a new thread with “university”, “students”, “message”, and “viewpoint”.

There are some more words that could be taken out (“John” and “jr”, and what is “FALSE”??), but for now, this is the dataset! Each topic is represented in some percentage (often 0) in every case, so the dataset we’ll use for describing whether a case is about obscenity includes that information. We’ll have to see what the threshold is, i.e., if a case is 30% obscenity topic, does that make it an obscenity case? What about 50%? That’s a task for this week, and one I’m excited to share with our group as well. (Subtext: there is also a boatload of web development work to do and I’m grateful we can share the load of this data work yayyyy)

When I started the DAV program in the Fall of 2019, I was faced with the ostensibly unfortunate reality during registration that there was not a single open course in my program. I ended up signing up for two DH courses and one GIS class at Hunter College. In the end, it was a semester marked by transformative thinking: about data (what is it? where does it come from? who makes it? should it actually be called capta?); about categorizations (is it possible to reconcile the inherent messiness of our world with the binaries required to communicate through digital means? who is left in and left out when we decide which structures compartmentalize the world? or, even more important, do we recognize categorization as a subjective, historically situated decision, not a reflection of “inherent truth”?); about visualizations (is there an inherent lie in representing 3D space on a 2D map? what about an inherent lie in representing data sets visually? or an inherent truth in legibility and access?). So many questions!

Since that semester, I’ve taken 6 classes in the DAV program. I’ve continued to think about, and be pushed on, these and other crucial questions about humanist data inquiry. But they’ve more often been from the DAV perspective — that is, the goal was generally to produce data analysis and data visualizations as the deliverables. The questions are important and absolutely considered in the process, but they are on some level incidental to classes that expect a final product of shareable insight via data, rather than, for example, a paper or a round table discussion as the fruits of knowledge production.

The return to the DH side this semester has made me realize how much more action-oriented I’ve become in the last year. I’m constantly thinking about how my work relates to the news, to my life, to the U.S. at large, to the jobs I’d like to have in one year, or five, or ten. Right now I’m both enjoying the opportunity to think deeply about data management plans and group dynamics, and rearing to get started on making stuff. In the last week or two, the rearing to go side has been shouting louder and louder.

I’m happy to have both sides. My background is largely in well-funded academic spaces, where talking about data and equity can happen without the urgency of needing to actually get a project done on a deadline. At worst, this has at times led me to feel like the work I do in private, academic spaces feels irrelevant to the work that’s needed in public spaces. The DAV program is, for me, a great antidote to that. I guess this is all to say, I’m not “move fast and break things,” I’m “think deeply and make stuff.” And I’m ready to make stuff!

Eva Sibinga is in her final semester of the Grad Center’s Data Analysis & Visualization program. Her research interests include data ethics, the intersection of race and technology, and the application of feminist theory to contemporary data questions. With a background in English Literature and Visual Art, her approach to data analysis and visualization is motivated by a desire to expand the way we tell stories and understand the world through our own eyes and others’.

Eva is one half of Freedom of Speech*’s core data and developer team. She will also support the project’s outreach effort, and hopes to improve her understanding and skills in UX/UI design by applying some time and effort there as well.