Critical Data Literacy with Dr. Eldar Sarajlic

A conversation with Dr. Eldar Sarajlic, Associate Professor in Philosophy at the City University of New York.

The following is a conversation with

Eldar Sarajlic

, a philosopher and college professor with some, in my opinion, radically novel ideas on education. He teaches critical thinking with a data science mindset. In this interview, we discuss his radical approach to improving critical thinking education with an algorithmic twist.

---

Alejandro: Let's start with some introductions. You describe yourself as "a philosopher hooked on data science." Could you tell me a bit about yourself, your background, what you are doing right now, and what it has to do with data?

Eldar: Yes, I am a philosopher. I have a Ph.D. from the Central European University —previously in Budapest, now in Vienna—where I wrote my thesis in political philosophy and theory. I am an Associate Professor at the City University of New York (BMCC), teaching critical thinking, scientific and data literacy. My fascination with statistics and data is of recent date.

For most of my academic career, I've been focused on normative issues, first in politics, then in education, but as I've grown more intellectually and as I've taught generations of students, I realized that statistical reasoning is a very important, yet often neglected component of critical thinking in general, so I decided to shift my research and teaching focus a bit to help students develop these skills.

Even before, I was interested in reasoning under uncertainty, but the pandemic nudged me to make this shift. I realized many people have poor skills in generalizing from data. I created a new Critical Data Literacy course, which will start this coming Fall semester. To best serve students, I had to learn a lot (and I'm still actively learning) about things I have not learned before, from formal statistics to machine learning. These fields fascinate me; I see them as amazing epistemic frameworks that deserve to be more popularized within the standard critical thinking curricula.

Alejandro: Critical Data Literacy is such an intriguing concept for me. From the computer science perspective, I'd argue we tend to see philosophy in maybe two separate ways: one as a mostly abstract exploration of what it means to know something and how to achieve it —what, in my layman's understanding, would correspond to epistemology?— and another as the exploration of the social aspects of scientific endeavor, things like biases and moral issues of automation, etc.

However, Critical Data Literacy sounds like it would touch both aspects. Is this distinction meaningful at all? How do you see this interplay of the more abstract and more practical aspects of philosophy in your course?

Eldar: I don't see abstraction and practicality as separate things; that attitude is reflected in how I think and teach. For me, philosophy is an endeavor to help humans think more clearly about anything. To achieve its aims, philosophy naturally engages in abstract thinking. Similar to computer science, in philosophy, we abstract to understand the underlying structures behind processes, events, and phenomena.

While some philosophers see that abstraction as an end in itself, which is legit, too, I see it more pragmatically: it serves a broader aim of being a better thinker and decision-maker. So, yes, my idea of Critical Data Literacy combines abstract thinking about what data means and how it reveals the world to us with practical questions about what we want to do with data, how we process it, and for what purpose. I think this two-pronged approach is most beneficial for developing substantive skills in reasoning with data.

Alejandro: It makes a lot of sense to me, though I wonder why such an artificial distinction seems to permeate at least the formal sciences. Math and CS have a strong philosophical tradition. From Descartes to Newton and Leibniz to Russell and Godel and Turing, our greatest mathematicians were also great philosophers —or maybe it was the other way around?

The origin of CS can be traced back to the philosophical inquiries of Leibniz, and the very theoretical foundation of CS is deeply grounded in Logic —which is a gift from philosophy, I'd dare to say. So I wonder why it seems there's such a divorce in formal sciences from the, like you say, more practical aspects of philosophy. Is it a curriculum issue?

Eldar: Well, this 'divorce' is a distinctly modern phenomenon. In the times of Descartes and Leibniz, the gap between philosophy and other disciplines was much smaller. Even the terms used for different sciences admitted the deep connections between them: remember that physics, biology, and chemistry were called 'natural philosophies' all the way from Aristotle to the 19th Century.

My guess is that the divorce happened in one part because of the advances in all fields, where specialization and the amount of knowledge accumulated simply don't allow a single person to know that much; and in another part because of the professionalization of the academia. We must not forget that academia is an industry of sorts and its dynamic influences how knowledge and sciences are distributed.

Alejandro: Certainly. Job market dynamics surely played a role as well. In summary, I feel we have a critical need for, well, critical thinking across all disciplines, especially in formal and natural sciences. My students often look with disdain at all the societal implications of the technology they're helping create, as if by being on the engineering side of things, one could shield oneself from the moral dilemmas.

In your experience, what are the most lacking skills regarding critical thinking in modern college curricula?

Eldar: It is clear that to be a good critical thinker, we need a good balance between technical and conceptual skills. I think the ideal configuration between the two cannot be determined generally but has to vary depending on the discipline of study. For example, CS students will perhaps need 80% technical and 20% conceptual skills; humanities students need the inverse of that.

So, what may be lacking is the awareness that there has to be balance rather than some particular ratio. In my disciplinary corner, I see a lack of emphasis on technical skills. I define these as skills needed to implement and act optimally based on a sound understanding of the world. For example, many of our students may recite complex philosophical accounts and know the name of logical fallacies by heart but then go out and be scammed by a used car salesman in a blink of an eye. There is a disconnect between abstract thinking and practical action.

I'm guessing the situation is the opposite in the CS field, where students master complex software development but then lack an understanding of its ethical consequences?

Alejandro: Yes, but in a sense, it is similar to what you describe. Our students can make complex proofs in predicate logic and calculate characteristic statistics of whatever probability distribution they want. However, they fail to see when a politician makes a clear non sequitur or faulty probabilistic reasoning. However, I do feel there's more need for a larger discussion about the societal impact of technology in the "hard" sciences.

This brings me back to your Critical Data Literacy course. I'm intrigued to know what is on the menu and how it differs from classic critical thinking curricula.

Eldar: I'll be happy to share it. I'm (literally) taking a break right now from writing chapter 6 of the textbook for this course to answer your question :)

I usually like to describe the course by saying that it is akin to an intro to the philosophy of statistics or an intro to inferential reasoning. I'm trying to offer students a very basic and down-to-earth overview of the most common reasoning principles used in statistics (and machine learning). I've been learning a lot about these disciplines in the last few years and I became fascinated with the ingenuity of some of its concepts, which allow us to develop practical insights into reality with very clever methods. I realized that these methods should (and could) be popularized among the general student population as a way to develop their critical thinking skills, especially now when technology takes over more and more space in our existence and interaction.

For example, I think that every college-educated person, regardless of their field of study, should understand concepts such as probability distribution, linear and logistic regression, decision trees, Bayesian update, cross-validation, neural networks, etc. These can be simplified enough to help non-CS or non-stats students understand how data can be used to make inferences and predictions and minimize the possibility of being wrong. I also believe it can help 'innoculate' some of them from believing that technology is perfect and infallible, and make them see the importance of human presumptions that are entangled with any technology. Basically, I see statistics and machine learning as very practical epistemologies that deserve to be more 'mainstream' in critical thinking and philosophy curricula.

In any case, that's the main idea, and it is still in draft form since I haven't taught this course yet, so it may go through changes. I'm also a firm believer in the importance of 'coding literacy': I think all college students should be able to write some code, and I plan to integrate this in the class as well. I have (perhaps a foolish and optimistic) plan to teach philosophy and humanities students basic Python so they can import, clean, and conduct basic analysis on a dataset on their own. I tried to find a textbook that satisfies this hybrid approach but could not find any, so I decided to write my own.

I'll make it free by writing it as a Jupyter Book and uploading it on GitHub, so others can see it, use it, criticize it, and perhaps help create momentum to popularize this approach a bit more. I also have to say that since I am not a computer scientist or a statistician by training, I'm a bit insecure about the real value of what I'm trying to do, so I do hope that friendly folk who know more about these things than me will provide some insights too.

Alejandro: That is a very radical approach, to say the least! There's a lot to unpack here, let me try to digest it into a few questions to take it piece by piece.

First, I agree that everyone should develop algorithmic and probabilistic thinking skills, even more than actual coding skills. However, I do believe learning to code, even if just superficially, is the best way to develop those more abstract skills.

In that regard, maybe it's obvious, but I want your opinion either way. Why Python?

Eldar: Well, a) because I’m familiar and comfortable with Python, and b) because it is versatile and easy to learn. What are other (similarly easy and versatile) options? I’m curious: in what way do you think this is radical? Does this look like a feasible project from the perspective of a CS professor?

Alejandro: I agree with your choice of Python. It is probably the most accessible programming language not only because of formal characteristics like its syntax but also pragmatically because it has a huge user base, good documentation, great tooling, and works everywhere.

I think it's feasible, but it is a huge challenge, not in the least because —same as my CS students look with disdain at social sciences— at least in my experience, I've seen philosophy students disregard math as something too technical and irrelevant to their interests. So I think it is as hard to make either side understand how critical is the other side's viewpoint.

I meant “radical” in the sense of "attacking the root of the problem". You're asking what's missing for critical thinking and deciding to go to the roots, directly teaching coding, probability, and stats instead of just some high-level intuitions.

Eldar: That’s a great way to describe it, actually. Exactly, I want to dig deeper below intuitions and provide some practical skills that address some of the main challenges of our time. That’s why this kind of conversation we’re having is very important. I’ve never done anything similar, but the more we converse, the more I realize how important it is.

Alejandro: Backtracking a bit, another radical move (now in the more colloquial sense) that I find in your approach is to make the documentation free and open source. Academia has a long tradition of jealously treasuring its knowledge. While disciplines like physics, math, and computer science tend to be more open in publishing, I feel this is somewhat novel for philosophy. Am I right? And in any case, what prompted you to go open source instead of the more traditional academic textbook publishing approach?

Eldar: Yes, you are right; this seems novel in my field. The reason I decided to take this approach is twofold. First, I have earned tenure this year, so I'm under no pressure to publish for academic publishers anymore. Second, I can finally act upon my profound disappointment with academic publishing in general. I guess this is the case with a few disciplines, but it is acute in philosophy: philosophers just talk to one another via journal articles and books, and it is not clear what's the objective value of our writing.

Sure, it benefits us to get promotions and tenures, but beyond that, there seems to be little of value. Our writings are too specialized and complex to be of interest to anybody outside our narrow area of expertise. I decided that I didn't want to do that anymore. It's not even fun to write, thinking only about how to satisfy some abstract reviewer out there. Once I wrote the first chapter of this textbook, thinking about students as my readers, it felt incredibly liberating, and I realized that I enjoyed writing more.

Why open source? Well, most textbooks published by fancy publishers get very expensive (teaching at a college where most kids come from the working class made me very sensitive to price). This limits the book's reach; some people will not read it because they cannot afford it, and I'd like to reach a wider readership. In addition, my impression is that the deals that most textbook publishers offer are not super beneficial for the authors either.

So, all of this combined made me decide to just put it out there and let anybody read it for free. It also allows for easier updates and editing. I've learned a lot by looking at how folks in CS do this, and I think it is amazing and should be replicated in other disciplines too.

Alejandro: The state of academic publishing is certainly depressing, and I think that speaks at large about the health of our scientific endeavor as a civilization in general. I've written before about ways academia can embrace openness, so I'm thrilled to see your efforts in spreading knowledge beyond the gates of traditional academic publishing. This is a topic I'd like to explore with you on another occasion.

For the purpose of closing this interview, I'd like to ask you how can people like me and other readers of this substack, who are mainly on the technical side of the discussion, get involved with your project. Where can we find your book, and how can we collaborate?

Eldar: Well, I will upload the first version on GitHub sometime by the end of August (at least that's the plan), and then I'll share it on Twitter. I'm not sure yet what would be the best way to have people involved, but for now, I plan to use Twitter as a vehicle for that (maybe Discord, too, I have to learn more about it to figure it out). Once it is set up, I'll focus on each chapter individually, perhaps every month, to tease out some details and hopefully have people give their own five cents on the chapter topic. Something like that, but I'm still searching for the best modality.

This was an awesome discussion, Alejandro! I think we should do one where I interview you and focus on how you are trying to bridge the gap from your perspective.

Alejandro: Sounds like a plan! Count me in for both to continue the talk and help you in any way I can with your Critical Data Literacy project. Thanks a lot for your time, and let's talk again soon ;)

---

This kind of in-depth conversation is something completely new I’m trying out. If you enjoyed it, please comment and tell me how I can improve it. Feel free to suggest people (and topics) you’d like me to talk to.