Richard Rogers: What has interested me and what I’ve developed as, as a so called web epistemologists, is thinking about, not just what’s specific about the culture, so what one would call web or platform vernaculars nowadays, but also what’s specific about the methods.

Noshir Contractor: Welcome to this episode of Untangling The Web, a podcast of the Web Science Trust. I am Noshir Contractor and I will be your host today. On this podcast we bring thought leaders to explore how the web is shaping society and how society in turn is shaping the web.

My guest today is Richard Rogers. You just heard him speak about what he terms “digital methods.” Richard is a professor and chair of New Media and Digital Culture at the University of Amsterdam. He also is Director of the Digital Methods initiative, known for the development of software tools for the study of online data. And he is the author of two award winning books Information Politics on the Web and Digital Methods among others. His most recent book is titled Doing Digital Methods. He is currently working on a book titled Mainstreaming the Fringe: How Misinformation Propagates in Social Media. And Richard was Program co-chair for one of the very first Web Science conferences back in 2013. Welcome, Richard. 

Richard Rogers: Thanks very much. Great to be here. 

Noshir Contractor: I’m delighted that you’re able to join us today. Take us back to those early days when you were first getting involved in the web. What prompted you to think about focusing on the web as the object of study?

Richard Rogers: Oh, that takes me way back. So I think it was in the mid-90s, when I was asked to write an article about climate change, I started sort of surfing around and noticed that certain websites linked to other websites, but then the websites didn’t link back. So that’s when I started thinking about creating software that actually maps how websites linked to one another, ultimately resulting in a piece of software called the issue crawler, which To this day, is still crawling the web and mapping links between websites.

Noshir Contractor: Tell us more about the issue crawler, that was definitely one of the first tools to study the web. And tell us what you intended it to do, why you call it the issue crawler, and where it is headed these days.

Richard Rogers: So when we started looking at links between websites, what we noticed was that a lot of websites would be linking to one another around social issues. So we coined the term issue networks — and well …coined the term, sort of repurposed it, looking at how not only NGOs and academics but also governments and corporations would be interlinking or not linking and so that we came up with a kind of link language. So there were critical links, this is like Greenpeace linking to Shell. There were aspirational links, there were these NGOs linking to governmental organizations or international organizations, and then international organizations wouldn’t link back. So there were these missing links. We call this sort of the politics of association. And that’s what we were putting on display with our link maps.

Noshir Contractor: How would you interpret when one website link to another and the other did not return the link or reciprocate the link?

Richard Rogers: It’s about reputation, largely. We found that, for example, in one small study of Armenian NGOs, so they would link copiously to one another, and then they would also sort of aspirationally link to UN organizations, and the UN organizations would link to one another, but then they wouldn’t link to the Armenian org — So it’s a kind of a lack of recognition. It’s about reputation, it’s about relevance, in some sense. 

Noshir Contractor: How relevant is web linking today as compared to what it was when you were first developing issue crawler?

Richard Rogers: So it’s interesting, when I first started writing about hyperlinks, I talked about them in terms of a sort of link economy and link economy actually supplanting an earlier economy, which I refer to as the hit economy. And so now, you could argue that the like, economy has taken over from the link economy. And of course, we’ve seen the sort of widespread industrialization of the hyperlink. You also see that links have changed, right. So it’s quite actually quite complicated, more complicated than it used to be, to map links. 

Noshir Contractor: You talked about the evolution from the link economy to the like, economy. Tell us more about what you mean by the like economy.

Richard Rogers: There’s a term that I sort of repurposed from sort of critical business studies called vanity metrics. And so I’ve been studying, quote, unquote, vanity metrics. And this is follower counts, like counts, view counts, all of these numbers that show how well you’re doing online, especially in social media. This is what you could summarize as the like economy. 

Noshir Contractor: One of your major contributions to web science over the years has been your work in the area of web epistemology. Can you tell us a little bit more about how you got interested in that, what it means and what have we learnt about that?

Richard Rogers: So generally speaking, web epistemology is the study of the web as a particular knowledge and or information culture with its own specificities. Wh at has interested me and what I’ve developed as a so called web epistemologists, is thinking about, not just what’s specific about the culture, so what one would call web or platform vernaculars nowadays, but also what’s specific about the methods and so what I’ve tried to develop over the years, or what I’ve called digital methods, 

Noshir Contractor: What are some of the things that we have unearthed that we would not have been able to do if we didn’t think about the web from an epistemological standpoint? 

Richard Rogers: If you think about web science, in particular, I think it came from a particular insight about the web — that the web is not just like a cyberspace as we once thought, this sort of realm apart, it’s not necessarily only to be studied as, as the virtual or as a virtual society, but rather, that the web has interesting societal data, right? How do you then capture this data, and think about making findings that you then ground in some ways. Amongst those ways, would be to ground them, quote, unquote, online. So this is one of the notions I’ve tried to develop, online groundedness. So the idea of using web data to make findings about what’s happening in society and culture, and then go out and grind, grounding them in the online, of course, we can triangulate, we can we can bring in other data from you know, the ground. But, but this is, we can also bring in data from different realms online. 

Noshir Contractor: One of the things that you touched on here is the ability to be able to study all of society, not just the online world, but by using tools that are gleaning information from multiple platforms online. Could you give me an example to make this more tangible, a concrete example of an issue that is more pervasive, but that you’re able to glean information from one or more online sources to get insights into it?

Richard Rogers: Well, I mean, you know, the flagship project was Google Flu Trends. And that was a very interesting project, and it ran for a number of years and and what it did was anticipated. The incidence of flu by search queries and what went wrong with Google Flu Trends? Is it sort of just a general warning about this sort or admonition about this sort of work? Right. So, when people are searching? are they searching? Because they have symptoms? Or are they searching because it’s flu season, and they’ve heard about it, flu season on the TV news. So is the phenomenon happening in the wild? Or is it happening in media? I mean, that’s for me was one of the more interesting examples also, because of the critique there have, but there are others as well. So for a number of years, for example, queries on AllRecipes.com were used in order to sort of map the geography of taste in the US.

Noshir Contractor: This area that you just talked about, the example that you gave, which is fascinating, is part of the infrastructure that you’ve been developing, more generally called the digital methods initiative. The goal of that is to do research that goes beyond the study of online culture only. Can you tell us more about the genesis of the digital methods initiative? And what are the kinds of things that you believe you could observe and study as part of the digital methods initiative.

Richard Rogers: So it goes back to the beginning of web science, in fact, so it goes back to 2007. And it’s been around. Since then, we’ve developed I think about 100 tools. And most of it is situated software. So we come up with software that we need for a particular research project, and then a lot of it sticks around, it becomes more sort of, like general purpose, but other tools go away, depending on use. But right now, we maintain quite quite a lot.

And we use this software both for societal and cultural research, as well as sort of media research, media critique. More specifically, a recent study that we did was we looked at what happens to about 20 so called extreme internet celebrities when they were deplatformed from mainstream social media platforms. And then they migrated to telegram. So we built a telegram data extraction tool in order to see what they were doing online there and to see whether or not they were acting in the same ways that they were acting before, for example, 

Noshir Contractor: And what did you find?

Richard Rogers: We found a few things, some intuitive, but a couple of things that were really counterintuitive. So the intuitive findings were that their audiences had thinned considerably. Counterintuitive was that they were still posting the same amount, or they were posting very, very frequently. And this went on for quite a few months, despite the fact that you could say that the media that the platform had less sort of oxygen giving capacity in the sense that there are ewer viewers. But the most counterintuitive thing that we found, was that their language became far less offensive over time, which then led to a number of different speculations. One speculation was that maybe they were offensive before for their audience. And not they’re not just generally that offensive, for example. Or that they entered such an offensive space that they couldn’t be more offensive than the space that they ran. So these are two different scenarios, let’s say. But nevertheless, those were some of the constitutive findings.

Noshir Contractor: I want to take us to an exhibit that you were involved in, which was featured at the Zed KM, entitled Making Things Public Atmospheres of Democracy that was curated by Bruno Latour and Peter Weibel. That sounds fascinating. Tell us more about this exhibit. 

Richard Rogers: We built a couple of exhibitions interactives. One is called the issue barometer. And the issue barometer would basically show the rise and fall of attention in particular social issues.So we took a set of NGOs, multi issue ones, also single issue ones, made an issue list on the basis of what it is that they were campaigning for on their websites. And then over the course of three years, we followed their campaigning behavior, showing how attention to particular issues rises and falls.

Noshir Contractor: To what extent do you think this helped illuminate this issue for the general audience of policymakers, do you see that these kinds of tools might increase literacy or awareness about some of these issues?

Richard Rogers: Yes, I think so. This is sort of issue trend research, if you will. You can imagine policymakers these days with issue trend dashboards, so, this is one of the earlier ones, but this was also in some ways a mirror for for non-governmental organizations. So are you demonstrating commitment, despite changes in funder agendas and sticking with particular issues? Or are you sort of following the money, so to speak. And so this was also part of the critical angle to this particular exhibition.

Noshir Contractor: To what extent are you able to use these kinds of methods to uncover disparities that may exist between the global south and the West, for example, or other forms of disparities that we see in society? Are there some examples from your work that show how these methods can bring exposure and bring those issues and those disparities to light?

Richard Rogers: what I just described, colleagues and I termed issue drift. And so particularly, nongovernmental organizations or governmental organizations sort of drifting away from things that are important when they could be sticking with them. One of the kind of critical projects that we undertook along these lines was called issue celebrities. We looked at a very important issue in the global south. And that is awareness of mines, land mines, and the clearing of mines and landmine related injuries. And we looked in particular at a charity or funding organization that was set up by Paul McCartney, and his wife at the time, Heather Mills. And, and it was quite serious that so, they raised year after year, something like $4 million, which was quite close to the total UN budget for the same activity, but then they broke up. So what happens to this global South issue when these celebrities break up and then leave it? It seems cynical on the one hand, but it’s quite serious on the other with when we’re talking about this kind of money. So this is one project that addresses that particular aspect.

Noshir Contractor: You’re working on on this book on mainstreaming the fringe — how misinformation propagates on social media.

Richard Rogers: In the run up to the 2020 US elections, we studied the extent of the so-called misinformation problem with a cross-platform analytical approach on seven social media platforms and we found that each of them in quite specific ways, but generally speaking, they all marginalize the mainstream. So for example, Twitter amplifies what is referred to oftentimes as hyperpartisan sources. On TikTok, they use particular sort of ronic sounds to instill mistrust when a mainstream media clip, for example, is played. But in all very specific ways, each of them sort of marginalized the mainstream. And of course, this has, you know, quite some implications for you know, taking seriously, news. 

Noshir Contractor: I’m still stuck back on what you mentioned earlier about TikTok sounds, tell me more about what you mean by that.

Richard Rogers: TikTok is this sort of, sort of music-driven platform and on the interface, when a particular sound is used, you can click on the sound and see other videos with the same sound. And so you can sort of map the use of particular sounds, okay. So there are certain sounds, which are used to instill mistrust in what it is that you’re looking at. And so this is this is quite interesting. And it turns out that a lot of the top let’s call them political videos on Tiktok, in the run up to the 2020, US presidential elections, were using those sounds. It develops in a kind of new type of misinformation. A lot of the videos are satirical, right? So that you think that, oh, that it’s no big deal, but at the same time, the satirical videos are introducing other sort of misinformation techniques. So you’re getting these hybrid types across social media platforms, you get new hybridity is that complicates the sort of typical topologies of misinformation, but the one on TikTok, I found was particularly interesting.

Noshir Contractor: I think one of the things that has recently emerged in web science is the endeavor to study multiple platforms. And you Richard have been at the forefront of being able to look at these multiple platforms. What I found interesting about the examples that you gave is that in many ways, while multiple platforms might allow us to triangulate some insights, you’re also finding that each of these platforms are used in distinct ways. 

Richard Rogers: I’ve been working on the kind of difficult problem of commensurability and crossplatform analysis. Especially in marketing research, a lot of the work that’s done on crossplatform analysis is about the study of engagement. So each platform has metrics. But each platform is also quite specific, right? So you can’t just blindly think that a hashtag usage is the same in Twitter as it is on Facebook, as it is somewhere else. My sort of short answer is that you need to understand the quote unquote, platform vernacular. So which types of digital objects are privileged and which are not privileged. And with that knowledge, you can then move towards something that is a more satisfactory striving for commensurability.

Noshir Contractor: That’s really been a challenge. I noticed that you’ve been spending some time focusing on a technical definition of “memes.” Tell us more about what got you interested in this particular topic at this particular point in time?

Richard Rogers: There was a Facebook engineer who was quoted a year or two ago saying, you know, 95% of the content that’s passing through memes. And I was like, oh! What I came across is, depending on the software, the memes are defined differently. For example, Know Your Meme, which is this sort of well known database that started in 2006 or (200)7, it has a particular way of thinking about a meme, and that is sort of the special internet phenomenon that requires a literacy in order to understand On the other hand, if you go to Crowdtangle, which is Facebook’s data collection software, both for research, as well as for marketing it has a meme search. And what it finds are images with text. Okay? So images with text is a very, very roomy definition of a meme. And the database definition is quite different. 

And then in the middle are a number of other ones. What I was looking at recently were: Okay, so what’s a meme?What’s a meme according to, for example? IRA disinformation operatives? So I went through about six or seven of these different ways of thinking about about memes.

Noshir Contractor: And what would these definitions allow us to do more specifically? you know, what is the advantage of creating this classification? What new insights that we gain by using this classification?

Richard Rogers: When thinking about how to study memes, you want to think about how to sort of demarcate this this phenomenon, right. And there are a variety of different ways, and I think that I think that’s the largest contribution. More specifically, what I’ve been doing is thinking about different kinds of sort of automation practices of, meme detection. And what we’re finding, generally speaking, is that the automated detection mechanisms are currently not that good at detecting what a sort of person or set of people, who are doing close reading, would call a meme. 

Noshir Contractor: Well this is interesting, thank you again, Richard, for giving us these little peeks with specifics, and all the rich kind of research that you’ve been doing and all your contributions over the years to a broader understanding of web science. And I wish you the best as you continue some of these efforts, and we’ll be tracking them in the years ahead.

Richard Rogers: Yeah, till then. My pleasure.