Emilio Ferrara: I feel like in web science and jointly adopting theories and data science, tools, and computational tools allow you to come up with the right blend of theory and data. That allows us to understand this phenomenon beyond just simple characterization, or simple theoretical explanation without a support from empirical evidence. This is really the fascinating power of web science, putting together these two things and balancing them together.

Noshir Contractor: You just heard from our guest today, Emilio Ferrara, who is an Associate Professor at the University of Southern California in Los Angeles. He has appointments in… Communication at the Annenberg School for Communication & Journalism, in Computer Science at the Viterbi School of Engineering and in Preventive Medicine at the Keck School of Medicine.  

He’s also a Research Team Leader for AI at USC’s Information Sciences Institute and the Director of the Annenberg Networks Network, ANN for short. And earlier this year, Emilio became the Chair of the Web Science Trust Network of Laboratories (WSTNet for short), which makes him an especially valuable guest today to talk with us about what he envisions as the next generation of Web Science. Welcome, Emilio.

Emilio Ferrara: Thank you very much, Noshir. I appreciate being here today.

Noshir Contractor: I’m just delighted that we have an opportunity to talk with you. I want to start by a tagline that I see associated with you, and that is your interest in networks and societies plus humans and machines. Can you unpack what you mean?

Emilio Ferrara: These new technologies emerge within our society, and they have effects on our society, their effects on the ways we connect among each other. They have effects on what we see on the web, and so on. So I feel like one of the most interesting opportunities in the context of web science that has been emerging over the last couple of years is certainly the ability to study how all these components interact — social media platforms, social networks, online and offline. And these emerging tools of artificial intelligence allow humans and machine to collaborate with each other. So this intersection of these different disciplines and areas has been the focus on my research for the last decade.

Noshir Contractor: You’ve done a lot of work, obviously, in the area of social bots, how would you consider or characterize the research that you’ve been doing, as an instantiation of this tagline that you have in terms of networks in society and humans and machines?

Emilio Ferrara: So we’re seeing the effect of social bots, which are accounts controlled in part are entirely by software, rather than human users, on social media platforms in a number of application domains spanning from politics, to public health, and so on. And these accounts are operating in public spaces in online networks, and they also interact with human users. So here you have the intersection between humans and machines affecting our social networks and our society.

Noshir Contractor: That’s terrific. Now, one of the first areas where I began to read about your work was the DARPA Twitter bot challenge that got a lot of visibility. DARPA has had an interesting history about technologies having been in involved with helping get the internet itself started. And this Twitter bot challenge was something that caught a lot of attention. Tell us what the challenge was about and what what what you learned from your experience with it.

Emilio Ferrara: Our team at Indiana University led by Fil Menczer and Alex Flamini was selected for this program. And I was lucky enough to be part of these efforts. And one of the goals of this program, as it developed, was to understand the possible effects of social bots on social media communication, especially in the context of public health.   

So DARPA  organized in 2014, this challenge for the detection of bots engaged in vaccine debate online. So the goal was to distinguish these anti vaccine and pro vaccine bots in the discussion. The challenge itself took place over several weeks over which we would receive Twitter content, sort of a playback of the to their content that will deploy our technologies to detect such accounts. 

So three teams did extraordinarily well. They detected all the bots, the team from the Subramanian and University of Maryland and the team at Indiana University led by Phil Manzer and Alex Flamini that I was part of, and then the team by USC led by Ron Gaston and Christina Lerman. And ultimately, this was just before I moved from Indiana University to USC.

Noshir Contractor: So you moved from one winning team to the other winning team, our competition? 

Emilio Ferrara: Yes.

Noshir Contractor: Back then, you were already looking at understanding the role of social bots and social media to deal with issues associated with vaccination, way before we all were experiencing what we have over the past year with the COVID crisis.

Emilio Ferrara:   I feel like public health has been maybe the most salient area where the manipulation of social media can have an impact on the real world and the change of behavior in the real world. Of course, there has been a lot of emphasis on politics, but public health sometimes goes under the radar. And it’s really not well established yet, the extent to which the manipulation of public health related discussion can be detrimental and dangerous for our society and social media definitely play a big role. So we looked at vaccine debates, long before COVID-19, we started looking into that around 2013 or so. 

And that was just around the time when a big measles outbreak occurred in California. And interestingly enough, the vast majority of anti vaxxer and vaccination groups, were actually aligning with left leaning or more liberal ideologies. Whereas today, when we look at the hesitancy around the COVID vaccine efforts, these emerge mostly from more conservative users. So these should tell you how much of a bipartisan issue is vaccine hesitancy, and how important it is to understand vaccine vaccine hesitancy through the lens of social media, because social media allow us to get a very diverse representation of political ideologies and how these ideologies interact with public health behavior.

Noshir Contractor: I want to go back to something you said, and that is that initially, the anti vaccine will be largely from the left, and now they’re from the right. I understand the second one. What is your explanation about the first?

Emilio Ferrara: That is an interesting phenomenon that we have seen, and you’re definitely right early on. And, you know, a decade ago or so these anti vaccination movement, especially opposing mandatory vaccine regulations for children emerge mostly from liberal progressive users. Were individuals with high education, typically from good upbringing, upbringing, in urban areas in rich states, like California, and you know, states in the West Coast.  And this was not necessarily for religious beliefs, but really for personal beliefs, and concerns about vaccine safety, vaccine side effects in children and so on. Interestingly enough, we have seen this shift towards a more diverse population of users that are opposing vaccination campaigns and a shift towards more conservative users opposing COVID vaccinations over these recent last year or so. So it’s really a bipartisan issue. And it’s a very complex issue to explain that cannot be explained exclusively with political beliefs.

Noshir Contractor: You’ve talked about the sort of relationship between political polarization on the one hand and online conversations not just about politics, but also about the pandemic. And you published an article late last year titled “What types of COVID-19 conspiracies are populated by Twitter bots.” So Emilio, my question to you what types of COVID-19 conspiracies are populated by Twitter bots?

Emilio Ferrara: Unfortunately, a lot. And unfortunately, some of the worst conspiracies that you can imagine. So this is a paper in which I took an early look at the landscape of COVID related discussion on social media. So we were lucky enough to have this foresight in our lab to start tracking COVID discussion early on in, maybe before everyone else did, in January 2020. And we also published these data sets. We made it openly available to the research community, and publish the associated paper in the Journal on Medical Internet Research. 

We made it available because we thought these would eliminate one of the barriers to allow researchers to get large data sets and understand this phenomenon beyond what we could do in house in our lab. 

In this study, I highlighted the role of 10s of 1000s of bots over the first couple of months of COVID. And it turned out the bots were active in the spread of political conspiracies, conspiracies of various types, conspiracies pertaining the origin of the virus. Some of the bots suggested that the virus was a biowarfare that was deliberately created, for example, by China, and it was deliberately spread to the United States and the rest of the world. And this created a lot of anti Asian sentiment. So that was very problematic kind of conspiracy. The article also highlights how other conspiracies focus on misinformation about treatments. 

But I feel like the most concerning kind of conspiracy is that the study highlights are related to bots that spread extreme political beliefs, beliefs that are mostly aligned with the out-right movements and far right ideology and so on. So what this study highlights is the attempts to  hijack COVID the discussion and turn it into political extremism. So some of the most active bots that we uncovered that I documented in this study, are bots that effectively spread Q’Anon. And some of the most prominent hashtags that we see are spread by these bots are hashtags that are very popular hashtag for white supremacists.

There are very many troublesome ideas and ideologies that have been spread and injected into COVID. And many of them have been pushed by bots. One thing that fortunately we have observed is the fact that Twitter ultimately suspended a large fraction of these accounts. So there was a mitigation strategy in place. But this took place many, many months after these accounts started to spread these ideas. So it was already laid, in some sense, they had already a large effect on the network in terms of spreading these problematic ideas.

Noshir Contractor: You’re using web science to study this phenomena that was created on the web. What are some of the ways in which web science is providing you unique tools and techniques to study for better? And for worse? 

Emilio Ferrara:  That is an absolutely interesting question. And actually, I feel like web science was the catalyst that started this entire research direction. In fact, early on in 2014, we published one of the first studies that looked at how to use the tools of web science to study social movements on online. Over the last several years, we have been using the same tools of web science to study other social movements. For example, we have some work coming out immediately where we studied Black Lives Matter, through the same lens of social media and social media discourse. So web science has allowed us to focus on the behavior of individuals and the communities and groups on these platforms and understand how these collectives emerge and characterize their activities, not only from a computational standpoint, from a data-driven standpoint, but also from a theoretical standpoint, looking at these groups as organizations,  looking at these groups as collectives. 

And I feel like in web science and jointly adopting theories and data science, tools, and computational tools allow you to come up with the right blend of theory and data. that allows us to understand this phenomenon beyond just simple characterization, or simple theoretical explanation without a support from empirical evidence. This is really the fascinating power of web science, putting together these two things and balancing them together. And we have been learning a lot about how to increase diversity, how to understand the biases, and so on, through the lens of the web.

Noshir Contractor: I want to touch on something you just mentioned, the extent to which web science should be in your opinion, be primarily concerned with identifying issues, identifying bias, recognizing things that might not be obvious. And then on the other hand, for a lack of a better phrase, doing something about it. How do you see where web science is currently positioning itself? And how well it’s doing on either of these? And how much should it be doing on each of these areas?

Emilio Ferrara: It’s dear to my own heart and research agenda to understand how we can use the web and the technologies that are enabled by artificial intelligence and so on to improve the web and to improve our society in you know, very directly right. 

I was delighted to see that these years Web Science Conference topic is actually revolving around making the web a more diverse, more equitable place using web science as a as a framework. And I feel like web science, in its transdisciplinarity nature, provides the best tools to do that. We have the artifacts, we have the machines, we have the technologies and tools. And on the other hand, we have the collectives, the networks, the aggregation for where people come together, and they are at the same time using these tools, but they’re also affected by these tools, right. And in my opinion, if you look at these dimensions in a disjointed manner, you’re only going to be able to grasp a partial view of the problems. You need to look at both sides, if you really want to understand how you can improve society using these tools, and how you can mitigate the negative effects of these tools on human users. And, I feel like web science offers the best lens to do that.

Noshir Contractor: And you have done beyond your own research. When I introduced you, for this episode, I mentioned that you have recently become the director of the Web Science Trust network of laboratories. So first of all, congratulations, and thank you for taking on that important role. I would like you to tell our listeners a little bit about what the Web Science Trust Network of laboratories is, and what you see as a vision going forward for this network.

Emilio Ferrara: So the Web Science Trust operates a network of laboratories that are spread around the whole world, there are more than 20 section labs affiliated in this network. And they are all very well known groups of researchers whose work is often associated with web science and other sister disciplines, all revolving around the study of social networks, the web, human and machine behavior and so on. These centers have been pushing the disciplines collectively over the last two decades or so. And the network itself plays a role into shepherding in some sense the community and the official direction of this discipline.

I feel like as a director, my dream would be to enable all these labs belonging to this network to collectively operate and create new initiatives that can push forward with data, and can push the impact of web science into our world into even more evident, obvious avenues. So we are embarking on to our collective initiatives, initiative to pursue larger projects, collaborative projects that try to cross national boundaries, and push for more diversity in this field and more diversity in even in the labs and in the discipline as well. Coming up with our moonshot, which would be at least one major research project, bringing onboard as many of these labs as many of these countries as many of these sub communities as possible, and trying to pursue such moonshot project. So there is a bright future in my mind ahead for the web science community, for the Web Science Trust and the network of laboratories. 

Noshir Contractor: You did mention the word moonshot. If I had to ask you at this point, what would be an example of a moonshot study that would involve all the WSNET labs or at least a large number of them? What would that look like?

Emilio Ferrara: As for the moon shot, I feel like one of the major roles that web science and web science community can have in the future is really operating as a glue to bring together people from different fields and encourage them to pay attention to maybe the web as as a societal glue, as a as a system of systems, and allow the study of the web or systems of systems, with respect to some of the emerging problems that we collectively face, our society. 

The pandemic of courses may be the problem that is on everyone’s mind on these days. But there are many other problems that emerge. And the web can provide a lens to study them: sustainability, climate change, and so on are definitely very important problems. I feel like that’s something where the Web Science Trust can contribute because we can use the web as a monitor, as an observatory, to understand how people think about these problems, how people pay or not pay attention to climate change, to sustainability issues and so on. 

Obviously, there is another big problem that revolves around artificial intelligence and automation. So you know, as these AI revolution keeps emerging, there are going to be issues with job displacement. Web science, again, provides a lens to study human behavior in these new contexts and a way, maybe to anticipate issues and problems with that will exist in the future of work and society. 

And then, of course, there is always the aspect of democracy that is very dear to my heart, right. So as we observe the world change, as a reflection of all these phenomena, public health, pandemic, automation, and so on. Our countries, our democracy are constantly in peril in danger, right, because we have seen the rise of these nationalist movements, and extremism of every kind and sort that they’ve been leading and growing in the web. And we should study them through the web, right? Because that’s their natural environment. And these are the tools that we have at our disposal, and we should maximize them. So I think these are going to be part of the big moonshot, that the Web Science Trust, and web science as a community should, hopefully contribute to in the near future.

Noshir Contractor: That is extremely exciting. I think the idea of taking web science and training its focus on some of these grand societal challenges, would be incredibly powerful and compelling, if for no other reason, because so much of what is happening in all of these contexts is being coordinated via the web. And so as you said, using the web as an observatory and as a monitoring platform becomes important. And as you said, beyond just using it to monitor, you also have the ability to change some of these phenomena as a result of the tools that we have and the technologies and that we have related to the web, etc. 

I want to thank you again, Emilio, so much for all the excellent work that you’ve been doing in this area, for your leadership on the Web Science Trust Network of Laboratories, and for coming and sharing some of these ideas and exciting plans that you have with us today. So thank you again, very much, and we look forward to getting much more insights and research and leadership from you in the decade ahead.

Emilio Ferrara: Thank you very much. It was my privilege.