Nowadays, computational methods are essential in almost all fields of science. Social sciences are no exception, despite the fact that they are often considered to be some of the most “traditional” forms of science. In order to learn more about this topic, The KAIST Herald interviewed Professor Lanu Kim, who specializes in computational sociology.

Could you briefly introduce yourself?

I am an Assistant Professor at the School of Humanities and Social Sciences. I was recently hired in June, so I am very new to the school. I have a Sociology PhD from [the] University of Washington in Seattle, and then I moved to Stanford for postdoc. I worked with Professor Daniel McFarland, who is also a sociologist scholar, and then I was affiliated with the Convergence Lab. My advisor professor in postdoc scholarship was working with a computational linguist at the School of Computing and Linguistics.

So, that's my background. I am studying how knowledge is shaped based on big data and computational social science approaches. My BA is also [on] sociology and economics, so my career is full of sociology.

Could you tell us why you decided to come to KAIST? And to the School of Humanities and Social Sciences too, instead of, say, the School of Computing, as your work also has a lot to do with Computer Science?

First of all, I think KAIST was my first priority when I was looking for a job, [because] it especially [focuses on] science and technology. Another thing that was really attractive [about it] was that [it] really encouraged [us] to work with different disciplines. And that's really difficult, particularly in Korean universities. So, I was thinking it might be the best. Of course, there are very few jobs for professors, so I didn't have a lot of choices when I was looking for a job. But if I [could], I really wanted to come here because I know the background atmosphere — it's more free and it encourages convergence research. [Those were some] of the top reasons why I came here.

Also, I actually didn’t really think about going outside of sociology a lot because my background is really deeply rooted in social sciences. My major was really in sociology, so I feel that [what] I can give students the most is the way to apply those new methods based on digital data and computational skills to the traditional social science questions. So I thought, instead of [the] School of Computing, the School of Humanities and Social Sciences might be the best choice for me. And actually, I am also an affiliated professor in the School of Computing right now. So, I feel like I caught two birds with one stone.

Related to your previous answer, what made you explore the topic of computational social sciences, as opposed to more traditional social sciences?

That’s a really good question. I don’t know why. In the beginning of my career, when I decided to go to graduate school, I had always been interested in quantitative skills, quantitative statistics and analysis. And I was good at it, so I kept studying it for a long time. When I was in my PhD program, around mid-2010, there was a huge transition in the social science area. There were the new terms, like big data, data science, computational science, and information science [that] began to emerge along with the burst of informational data, collected mostly from smartphones, as well as many digitized documents. I was in the middle of that transition and I realized there would be no way back without learning about those computational skills. I was really lucky because I learned that when I was [at] the [turning point of] the transition process. I came up with a research idea, and then I realized it could only be solved by big data and computational skills. I requested a meeting with a professor at the Information School who had a background in computational social science, and then, I co-worked with him in my PhD program. That led me to my postdoc at Stanford, and that was the start of my interest. So, my research question hasn't changed, but the world has been changing. I was just trying to catch up by learning it.

What were you expecting when you decided to go into the field of computational sociology? After spending many years in the field, was there anything that surprised you, or did it still mostly go according to your initial expectations?

I think the most surprising thing for me is the amount of time I need to invest to write a paper based on computational skills, because I am still pretty much with the young [generation]. I can see that there are some really traditional social sciences scholars, who have a strong belief that actually we do not need these “fancy skills” to answer social science questions. Those kinds of data skills, which attract people’s eyes — people were just [being bombarded] with a huge amount of data and they think it is too fancy, it does not really efficiently answer social science questions. And then there are also young scholars arguing that we need to be in the flow of learning those computational skills. [These two generations] sort of have a conflict. When I receive reviews of journal articles I write, some of them say I need more theory: I need to justify my method. [Others] argue that I need to be more robust with my method: I need to collect more data. So, it’s really hard to balance [and combine these new things] in the social science field. I still need to personally convince the traditional scholars that it’s really necessary.

Also, social science scholars are not really accustomed to lab work. Being in the lab, sharing data, sharing ideas, and working together is still [a] relatively new [concept]. It’s another thing that I need to always explain. Even if there are four or five authors on my paper, usually the first author or the corresponding author [is the one] who has to put the [most] time and energy on the paper. That’s another thing that I’m looking at in the transition era.

About my initial expectations, of course it would be a lot of data — big data. So, [I thought that] if I have a lot of data, it [would] be easier to answer the research questions and write the paper, but actually, [the work] was really dirty. It took tons and tons and tons of time to understand how the data was produced. There might be some random typos that would make my results confusing. So, it’s not actually showing [any] pattern, but it might be just algorithmically induced results sometimes, or the results of [those] random typos. Figuring out these kinds of things, one by one after trial and error, tons and tons of [times], were somewhat out of my expectations. People think it’s fancy and cool, that I’m excellent in coding, but the reality is that it takes forever to clean the data. I bought a standing desk because I’ve been sitting in front of the computer coding for too long.

What about the things that did go according to your expectations?

Well, it’s in trend, so I thought it’s got to be the major way of doing social sciences. And I think we are [indeed] on the way to it. I believe, in the future, computational social science would be similar to our work. Quantitative analysis, everything that is going to use numbers and statistical skills, is going to require some sort of coding in the beginning, because most of the data that we are creating these days are based on these tools. So, I think it’s inevitable to move on, to learn about how to deal with digital data. It’s something [that is still] going on, as I expected before in the past. It’s going to be in trend and we will need more convergence to work with other areas who have expertise with taking care of the data.

Could you explain more about your research interests?

My doctoral dissertation was about how academic search engines like Google Scholars change the citation behavior of scholars. Google Scholars is a search engine [for academic publications] created by Google; it’s following the algorithm based on page rank. I cannot say the most popular, but [rather] the most relevant and popular articles — based on my search term or the engine[’s suggestion] — [would be shown] on top. As human beings, we are very likely to click on the results that come up at the first page, instead of moving on to the second, third, fourth pages and so on. So, I was just thinking that we, as people, might just be more engaged with this kind of academic search engine. We might just click on what the search engines suggest instead of really thinking [about] what kind of papers would be best fit to answer our research questions. And then we [would] just focus on the very popular papers that other people have been citing. That might stall scientific innovation, because we are only looking at the popular opinions and ignoring other minor opinions, [even though they] might be really genius. So, I was trying to look at how the citation distribution has been changed before and after the emergence of academic search engines. My timeline was pretty limited at the time, so it just says that it’s actually pretty stable and is neither concentrated nor diversified at that moment, but I think it might be changing by now. I think I need to update it sometime.

I saw your website and I read that you're also doing research on gender inequality, technology and society, and many other topics. Could you tell me more about your other projects?

The gender project is something that I’m really interested in, too. That was my postdoc project. Of course, there are some people who really love it and some people who hate it, because all the gender topics always have some fans and anti-fans. Mine is the same. The main argument that I’m trying to make in this paper is that occupations women are more likely to have are paid less than what men are more likely to have. For example, childcare workers are doing an important job, but they are paid much less than software engineers, where men are dominant in the occupation.

So, there is [an] occupational segregation that makes the salary inequality. And I believe that idea can be applied to academic knowledge as well. For example, in sociology, men are more likely to study labor, network, theory, and politics, those kinds of topics. Women are more likely to study childcare, women, and gender inequality; they are more likely to use an ethnographic approach. They are [also] more interested in poverty. [Men and women] have different research interests, and it’s genuine. It’s very natural that women and men have different perspectives and they might study different things. I argue that because of those different interests, people who study those topics [wherein] men are dominant are more likely to get a job in a tenure track position in comparison to those of female-concentrated topics, like child or family or aging. My paper says, yes, there is a particular preference that favors men’s topics instead of women’s topics. People still hate it because [they] believe science should be neutral and that we are just evaluating people’s scientific contribution. And that, they do not really believe in. We discriminate against people based on their topics, whether their topics are gendered or not. So it’s arguable. But what I say is that yes, there are still equality and discrimination against women. [Compared to STEM], it’s more pronounced in social sciences and humanities, which are, in general, still very struggling in these engineering-centered societies.

What do you think about the president's digital humanities vision?

I think the main vision that our president is pushing is that there needs to be more conversation between social sciences and engineering fields. The vision of digital humanities is that of the School of Humanities and Social Sciences. It [currently] only has an undergraduate program, [which provides] classes for undergraduate students [for a balanced education], with common sense about the society as well as [their] particular [field of interest]. But now, I think what the president is pushing is that he wants to make our School of Humanities and Social Sciences to be a Digital Humanities Department. He wants us to have a niche as social sciences scholars in the world, by leading the convergence [of]  engineering and science fields.

I think his vision is wonderful, and I think it has a reason and a background. Why should we go this way? Because we need to be innovative to achieve a niche that is really helpful to make us stand out in the world competition of social sciences and humanities scholars. I also think there are a lot of data science scholars, and a lot of professors in the School of Computing too, [who] begin to actually study social phenomena based on their data these days, [such as] with Twitter and Facebook. You can just name it and [that media data will have been used by data scientists]. [Data scientists] can [benefit from] some of [social science] theories and explanations, and [social science scholars can also benefit from] their knowledge on [computational] skills and data collection methods. So, I think we can have some synergy effect by communicating more.

On a related note, just a few hours ago, I received an email about the GSI conference on metaverse. What are your opinions about metaverses?

I just saw the poster in the morning too, and I was really shocked that our president will give a keynote speech. And then there were a bunch of people from the industry and there were only one or two professors from academics. So I think it’s really a good mix of industry and academia for discussing metaverse, but I don’t know — it's a very difficult question and I’m learning, [too], as a newbie in the field, but I definitely think that [the metaverse is] a doable technology and if it becomes possible, there will be a lot of [changes] in social interaction between people. So far what I found and what people argue is that face-to-face interaction cannot be replaced by online interaction because of its limitations. We cannot feel other people in the world and we think [online communication is] just complementary instead of being the major method we’ll be interacting with [others]. But I think if the metaverse world becomes really stageable, our hypothesis of face-to-face interaction might be changed first. So, I’m looking forward to the conference as well.

Traditionally people see social sciences and computer science as two unrelated fields, but computational sociology is gaining more attention these days. Do you think this convergence of studies is also the trend for other humanities subjects and other fields in general?

Yeah, it’s in trend basically everywhere. Unfortunately — sometimes I say it's unfortunate because these trending fields are mostly for men and there are only very few spots for women, so I’m sorry about it. But anyway, I think it’s not only sociology; there’s definitely economics, and political science is [also] a huge field that’s adopting computational social science. Psychology, of course, is trying to converge with neurology as well. Anthropology might be a slight exception because they still [need] to do ethnography, but [in general, the convergence with engineering fields] is pretty dominant in the social science field. In humanities, like philosophy or classic literature, there is [a field] called the digital humanities, which is trying to digitize all those analog data into a digitized format and analyze it. I think you have run into works like analyzing classical paintings by digitizing them and [running a machine learning model trying to deduce the perfect frame for those paintings. So those kinds of approaches trying to convert to those computational skills with ancient literature and paintings, become very popular. Yes, it’s really in trend.

[On another note], there was a new song that came out with ABBA’s voice, [based on artificial intelligence]. They are very old singers from the 1970s and [scholars] recreated a song based on [the singer’s] voice by running a machine learning model based on previous albums.

We feel like they are alive and in the present. [Those are what] we are trying in the social sciences, humanities, as well as art. [Those are all] very interesting [trends of bringing back artists by AI], but we need to always ask why we are doing it — why we should recreate ABBA and why we shouldn’t recreate ghost paintings, for example, when there are tons of artists there.

Do you think that the rise of this digital humanities has a lot to do with the COVID-19 pandemic?

I think the increase in computing power is a huge reason why we can do [digital humanities], because when the computing power is weak, [the study] becomes inefficient. [People] don’t really [have] time and energy to try it.

[I] do this kind of work because the computer is really fast now. We don’t need supercomputer powers to run [huge sets of data]. I think this advancement of technology and our behavior is moving into digital tools, which makes it easier to record it. [This] definitely pushes the need for digital humanities as well as the sweeter size.

Do you have any tips for young scientists who would like to pursue fields related to this convergence topic?

As a young scholar who just finished my PhD, I don’t recommend going to the graduate school, but, for those of who want to do computational social science, I want to emphasize that it is a really interesting field and there can be so many more things [that] we can do than we think. It’s really exciting, people are excited about it, and if people want to listen to you, sometimes it’s easy to connect with the industry as well. But the disadvantage is [also] there. My friends and I call it the “two PhD problem” because now mastering one discipline is not enough. It is always my substantive knowledge of the discipline plus alpha, which is the coding and computing skills. And that [“alpha factor”] is becoming more and more important. It’s really a lot of work. Because there are not so many people, particularly in the social science area, who can guide me on [what] to learn and [what] not to, there are few mentors, [making] it difficult to navigate. That’s a kind of disadvantage; [so], your PhD program might be longer than you expected, but as soon as you figure it out, I think it’s a really interesting field and also a growing one.

There [will be] more people who want to listen to you and expand [the field], so [this part] is encouraging.

The way of doing research [is also very different] compared to before. Online communication becomes more dominant and normal. I cannot imagine myself defending my dissertation at the end of my PhD program online, but now all the dissertation defense is done through Zoom. Conferences are [also] moving online; it has become a new normal.

And I’m sorry to say, but it is becoming more competitive [as well]. You will be pushed to produce more in the very beginning of your graduate program. I think it's very necessary, but on the other hand, [some] people will not have enough time to think about what’s [their real interest area]. They might just [end up] following their advisor’s idea instead of [coming up with an original] dissertation topic idea. If I have a student in the future, I really want to encourage them to find a research topic by themselves, instead of asking me to give them one. I think that productivity pressure is one of the [most difficult] things.

During your work as a sociologist, did you find any results that you found especially interesting? Any particularly insightful findings about our society?

There are a lot, so I cannot pick one, but the actual axiom of my life is based on a sociological research that came out in 2004. It’s by Ronald Burt, a very prominent social science scholar. What he argues is that a good idea [emerges] when two people who do not really share a lot of their backgrounds begin to talk. Then they can find a way to integrate the knowledge and make a real innovation. Of course, in my experience, it takes a lot of time because sometimes those who are from different fields do not share the same terms. Even if they are talking about the same thing, because they are using jargon, they might not communicate well. So it takes time, takes effort. It’s difficult to meet, but good ideas really come from what we call “structural holes”, networks that are vacant and haven't been connected yet. That was a really brilliant research in my idea, [that is] guiding me [on] what to do in my future. [I know] I need to talk to someone who is new, who is very foreign to my field, and then I can get a lot of insights from those kinds of conversations.

Was there something that you thought should be “common sense” when you found out about it in your research? Perhaps it should have been something that a lot of people know about, but in reality, not many people are aware of?

Much of the good research that I haven’t seen is to talk about something that we [think is] common sense. We know that something is common sense, but confirming it scientifically that it is actually true is completely different. I realized those kinds of observations are really interesting in my area. [For example], there was a huge discussion on whether feminization of occupations causes the [gender-salary gap, or the other way around]. That discussion has been going on for about 15 to 20 years, based on a lot of different research. What they found at the end, [was that] there cannot be a perfect answer, but I am sure those two factors are related. What I found interesting was that there was somewhat [of a] causal argument that feminization of occupations — I gave you an example of childcare workers and [how] the wage is likely to go down because it's what women do — might [indeed] be the reason why the overall wage goes down. That was pretty interesting in my opinion. People will not buy this argument in their face, but when you look at actual scientific evidence over time, it's likely that it's strongly related with [this causal] relationship.

What I also found interesting is that there was the polarization of political views in the Twitter era. The hypothesis is that when people are given better and opposing information, they will understand the balance, and the polarization will be reduced. But one of the papers [that] was published in PNAS, one of the top journals, argues that it actually increases polarization. When I am exposed to the opposite political view, I feel upset, [so] then my view [gets] stronger and stronger. [As a result], polarization [becomes] more serious. That was pretty insightful.

Finally, do you have any messages for our readers?

I have been in the student publications when I was in undergrad. So, I have a huge interest and love for all the student media. Sometimes I read the Korean newspaper as well as The KAIST Herald. I am happy as a reader, that there is a media that's followed by students and I'm very supportive about it. I feel like I'm talking to reporters, not the reader.  I was so excited when you actually asked for an interview, because I wanted to meet someone from these kinds of newspapers. 

I guess it's different for humanities professors because in other departments, usually professors lead a lab and have graduate students under them?

Actually, we are trying to launch a graduate program. Pretty soon. We do not have a graduate program right now, but we are aiming to have graduate students from 2023. So I think once it begins, it's going to be a new source. Of course there's nothing going on right now, but we are planning it. And then we do want to have the students under us because that agenda definitely needs a lab and I need people to work with, but it's impossible to do it alone. So that’s what we are aiming for.