← New Books Network
New Books Network
Thomas Hegghammer and Diego Gambetta eds., "Fight, Flight, Mimic: Identity Mimicry in Conflict" (Oxford UP, 2024)
01:05:05
Loading summary
Transcript54 lines
- [00:02]
New Books Network Host
Hey, NBN listeners. We're running our 2026 New Books Network Audience Survey, and we'd love just a few minutes of your time. NBN has been bringing you in depth conversations with authors and scholars for over 15 years. We haven't done a comprehensive audience survey since 2022, and a lot has changed since then. It's time to hear from you again. Here's why we're asking. We want to understand who's listening, what subjects and podcasts you love most, and where you'd like to see us grow. Your responses help us tell NBN's story to the publishers, libraries, and institutions we partner with when we can show that our listeners are serious readers, lifelong learners, and heavy library users. It opens doors to new partnerships, better resources, and ultimately a stronger NBN for everyone. And one more thing. If you leave your email address at the end of the survey, you'll be entered to win a $100 gift card to bookshop.org, a chance to stock up on books while supporting independent bookstores at the same time. The survey takes just five minutes. Your answers are confidential and your email will never be shared. Head to newbooksnetwork.com to take the survey today. We really appreciate your support. Now go take the survey.
- [01:14]
Thomas Hegemer
I was groomed to become one of his wives. This week on Disorder, the podcast that orders the disorder, an Epstein survivor tells me her story and what justice looks like for her. I want to see action, and I am demanding action. Do not just talk the talk. You need to start walking the walk now. It's one of the most powerful interviews I've ever done in over 20 years as a journalist. Search Disorder in your podcast app to listen right now. Welcome to the New Books Network.
- [01:50]
Interviewer (Raman)
What if what tells us about radical extremism isn't radical talk, but the amount of time spent in a certain form? That's an idea that Thomas Hegemer explores across his work, including in Fight, Flight, Mimic Identity, Mimicry and Conflict, a volume he recently co authored and co edited. Good to be with you, Thomas.
- [02:14]
Thomas Hegemer
Thanks, Raman, for having me.
- [02:17]
Interviewer (Raman)
So I just want to ask you more about this very provocative idea that's about time versus talk as a gauge of intent and signaling. Can you tell us more about that?
- [02:31]
Thomas Hegemer
Sure. So to get us there, I think I had to get a little bit sort of theoretical or conceptual and start where the book starts. So this is an edited volume that I co edited with Professor Diego Gambetta and this and Diego has done a lot of work through his career on what's Called signaling, and especially deceptive signaling, a phenomenon whereby people pretend to be someone they are not or pretend to be a member of the group that they are not a member of. And this is a very widespread phenomenon that we all know because it affects our daily life. The threat of being deceived by someone who is not what or who they say they are can happen everywhere. This is the reason we have, you know, we're skeptical of unsolicited emails. This is the, this is the reason we all have identity cards and passports and passwords and things. Because if we didn't have those things, people could go along and pretend they were you and misuse the, you know, you know, steal money or enter into premises where they shouldn't be, all these sorts of things. So in fact, our entire society is built up around mechanisms and institutions to prevent deceptive. So the society's ability, a society in which people can trust that what people present themselves as is really true, is a well functioning society. It's one in which it's much easier to do economic transactions or harder to commit crimes and so on. So it's just to say that the starting point is this broader domain of deceptive mimicry. And this is a field that's been theorized in sociology in particular by people like Diego and the. It can get a little bit technical, but the core idea is that when we interact with other people, we have to make judgments about what they represent and what qualities and intentions they have. The problem is that we cannot read their minds. I cannot know what you are thinking right now, I cannot know what you plan to do and one minute. But I can make assumptions based on the fact that other things I know about the eu, for example. And this is how we all operate in the world. We draw inferences based on kind of external signs about what people say and what people do, what people wear, things like that. So there's a filter in a way between what people's kind of true intentions and properties and what other people can know about them. And this filter is that this goes through science. So deceptive mimicry operates by imitating the signs of the properties that you want you to convey. And because these signs are often. Maniculable, they're vulnerable to being used for deception. So for example, a police uniform is a sign that police officers use to tell people that they are really police officers. But in principle, anyone can don a police uniform and make an attempt at being a police officer. And so things like that, disguise, for example, is something, is a sign that it can be Manipulated. And this extends to all kinds of the whole ecosystem of signs that fill our social world. And then the next core idea is that when we make assessments about whether a sign is credible or not, we think about their cost. Not necessarily in economic terms, but in some other term maybe can be things like time or pain or hardship or whatever. So for example, if I told you that, you know, you may know me as an academic, someone as, hey, I'm a. In fact, during the 90s, I was a frontline special forces operator in the Balkans, for example, doing really hardcore stuff on the front lines. And you'd be skeptical, especially given my physique. But then let's say if I pulled up my trousers and I showed you like a bullet wound or something, or even if I had like an amputated leg, that could corroborate a story of battle, I really know action, you would be more inclined to believe me. Why? Because getting an injury like a bullet injury is very painful and very cost, very costly. So in that case, in that scenario, the scar carries more value than my worse. So ultimately brings us back to the, you know, the very, very well known saying that talk is cheap. So words are in some ways the least extensive signs that you can emit in this signaling domain. So that's sort of the background. So when we come to this idea about time, it has to be. So that idea becomes relevant when you move into the online space. So this book explores lots of different domains of deceptive mimicry. It's all related to deceptive mimicry and war and conflict in some way. But some of them look at, you know, real life scenario scenarios, you know, in, in the re. In the real world, what happened during the Rwandan genocide, waste ended and how did the, the, the sort of, the, the hunter tried to, you know, escape the hunters and, and, and so on and, and, or real world stories from the world of view. Left wing terrorist groups in Italy or the IRA in Morgan island. Various ways in which Dominic repelled in there. But in my chapter, I look at what jihadists do online and how they decide who to trust in in discussion forums. Because online, the online world is very different from the real world in that it's predominantly verbal. So when you and I are talking right now, you can see me. So for listeners, we're actually in because it's office and we're sitting here face to face so you can see me moving my hands and you can sort of, you can also hear, you know, my voice going up and down and there's lots of non verbal communication going on, whereas in online you have, you know, what's known as low bandwidth, you only have the words very often. Sometimes of course, you can have pictures and videos and such, but most of the time it's just words. And this means that there's a much narrower range of signs to play with, both for the mimic, as we call them, and for the duph, the opposing side. The question then becomes how do signal games work in such an environment?
- [11:37]
Interviewer (Raman)
What are the key kinds of signs that are important in face to face interactions that get cut out online? I mean, you've given some examples, but are there particularly important kinds of signs that get cut out that would have kind of made it a very different kind of communication on these jihadi forums that you've studied?
- [12:02]
Thomas Hegemer
Yeah, so perhaps the most important one is a face. So at least before the AI age, and we can come back to AI later on that, before the AI age, at least the phases, it started to be extremely hard to fake as a scientist to make a real 3D face in flesh and blood. That's exactly like the model as we call it. So the terminology we're using sort of fills the particular mimic, exploits the features of a model to trick and dupe. So, so the face, for example, in the 3D space is extremely hard to fake. So just having access to the person's face in real life gives you as a potential dupe, a lot of information. Whereas in the online space you don't have that, you just have like some kind of avatar and then the person's words, you have no way of where, even if you. So at least let's say so in the real world, if you, you know, if I were trying, trying to trick you, you know, you would may be able to uncover my deception, but you'd still know who I was, you'd recognize me on the street a month from now. Whereas online it's even more ephemeral. You would be, you could be tricked by what the person is saying or claiming, but you'd also not have any clue about who it was that was trying to trick you. So it's just a much more difficult space. But it doesn't mean that it's inherently easier for one or the other side of this kind of competition. It's not inherently easier for the person doing than the making is or the GOP system are different spaces, different dynamics.
- [14:10]
Interviewer (Raman)
So let's bring time back in. So this is very helpful to think about the narrow bandwidth online. I'm still not fully understanding that connection to time so could you say how that narrow bandwidth foregrounds time as kind of the most important signal?
- [14:31]
Thomas Hegemer
Right. So I came to this as someone who had spent a lot of time kind of lurking on radical Islamist discussion forums from the, from the early 2000s onwards. I used these websites as a source of insight into what people were thinking and talking about to sort of get a sense of how the ideology was evolving and who the most influential thinkers were and that sort of thing. And so I had a good sense of kind of the sociology of it in some sense and of the content that was being shared. And it was really. Diego Gamutau drew my attention to this more kind of deeper social science questions relating to it and the fact, to the question of how well this seems to be a reasonably well functioning ecosystems where people are collaborating and so on. How does it even work? How do they even know who to trust given the sort of low bandwidth problem? So in, so in this space, and I have to explain a little bit about how these forums work or used to work because there are fewer of them these days because of kind of state repression on them, more censorship and so on. But you typically have an open arena, a sort of a message board where any kind could join. It's like one kind of thread or feed of, of comments. People would sign up with a profile if they wanted to post and most of the time these would be fake names and you could have like an avatar or a picture or you could not. And also you had this sort of more kind of organizational type actors on these forums where there were like sort of like the Reuters or the AFP of the Jihadi world sort of agencies conveying supposedly verified information from the front line or from real groups and so on. So we had this sort of space with different types of actors with different types of identities and so on. And at the same time, and this community faced a constant threat of infiltration. Because if I, as an academic, you're doing open source work on this and then suddenly the police and intelligence services could do the same. Could be. And then reasonable to assume there might be. I was there just as an observer fly in the wall, as it were. But intelligence services could be using this as to infiltrate, to pretend to be a real Jihadi for a while in order to lure people into, you know, and doing some, you know, doing something incriminating and then arresting them, some false flag operations of which there have been many over the years. So when you are jihadi and I was trying to put myself in, you know, in the, in the shoes of someone who's actively bona fide fahjari on these forums. Like, how do, how can you be sure that the person you're interacting with isn't a police infiltrator? And how do you make sure that people are who they say they are? Because ultimately what some of these conversations could be about very serious. It could be about sort of, you know, everything from kind of sending money to a group somewhere or meeting up with someone in the real world to discuss maybe traveling to the front line or even, or even more serious things. So given the sort of really high stakes, how do people judge who to trust? And as I mentioned earlier, verbal claims carry a very little value. So as we backtrack by saying into this with very few priors, really just trying to sort of inferentially establish what seemed to work. And I noticed that people who are loud, who are kind of making the strongest claims in support of Al Qaeda or ISIS, were not necessarily taken seriously. Sometimes there could be buzz about sort of them being precisely infiltrators, pretending to be jihadi. So words alone was not persuasive. It could even be like a turn off. Also sometimes it would be a good thing from the point of view of the potential, the person fearing to be referring to, bearing deception if the other person could share something rare, rare materials, so a video recording from an actual front line or a type of cultural things like jihadi poetry. Sharing that. Because that would be something that's password. Yeah, you'd only know it if you had spent time in an actual group and so on. But I eventually noticed that what people paid the most attention to was the age of the profile and the activity of the. The basically the volume of contributions over the age.
- [20:45]
Interviewer (Raman)
Meaning the online history.
- [20:48]
Thomas Hegemer
Exactly. That's a better term. Yeah, the online history and the kind of volume of contributions. Because that was one of few very verifiable things in this ecosystem. Because unless you had, you were an illustrator, had access to manipulate those metrics, you couldn't easily pretend to or have your profile say that you've been there, been online for three years and have 37,000 comments.
- [21:21]
Interviewer (Raman)
So it's both longevity and frequency. Does that matter? Yeah, yeah.
- [21:25]
Thomas Hegemer
I think if you had a very old profile and very few comments, that would be suspicious. But if you have the right combination of, of longevity and contribution, because contribution, bear in mind it's a valuable sign because it's costly, because you're incriminating yourself. If you're posting a lot of basically pro ISIS statements online and sharing Videos on you are breaking the law in many countries. So we're
- [22:03]
Interviewer (Raman)
over and over again.
- [22:04]
Thomas Hegemer
Yeah, if you do that then a lot it's a sign of someone who is loyal to the cause and doesn't care. So this is by the way another very well known screening tactic that we all know from films. For example, many may have seen or some older areas may have seen Donnie Blasco which is based on the story of the infiltration of a mob in the US by Joe P. Stoner where they basically when a dupe which can be an organization where is that it's been infiltrated or they're being faced with a hostile intruder, they might put that person to a test by enforcing the person to commit a crime or to commit illegal on the premise that a real infiltrator would not do that either. Because in some countries it's not legal for a police officer to actually correct the law or do not be assault in serious ways. So incrimination can be self, incrimination can be a cascade. So that's back to the time and volume part. This is to say, yeah, basically the profiles with reputation seem to instill trust.
- [23:40]
Interviewer (Raman)
If you can't read people's minds, how do you know how longevity and frequency are being read by others on the forum?
- [23:54]
Thomas Hegemer
Depends on which side of the game you're on. I mean if you are, I mean if you are the mimic, if you are a real infiltrator, if you are, if you are working for MI5 and you are trying to build, you know, trying to infiltrate ISIS networks in Britain online, then I guess you're using reputation building like this alongside a lot of other strategies in the hope of gaining access. I don't think any. This is a part of this whole. All mimic games are very dynamic. Rapidly shifting work counts as a, as a convincing sign in one time and space might not do it in another. But people are constantly kind of from both sides of the game. They're constantly navigating in adaptogen.
- [24:50]
Interviewer (Raman)
How about you as a researcher? So what's the sign to you that time has had this effect for this profile's reputation on the forum, what are you trying to observe?
- [25:05]
Thomas Hegemer
So it can either be the explicit signs of distrust or explicit signs of trust and also actions that are proxies for trust. So actual exchanges of actual exchanges of information when someone is in a discussion with someone and they're being persuaded by other person to, to send money to isis, Syria or something, something like that. But when you think about it, this is not all that surprising. Because we have it in regular economies already. This is exactly what ebay profiles do. These are, this is what. And in business there's a word for this. It's a reputation system. Because there too they face the same. In this sort of secondhand trade domain, you have the exact same problem. You were about to engage in an economic exchange. Sometimes there were very valuable items. It's quite a lot at sake. How can you be sure that the person you are sending the money to is actually going to send you that use iPhone back. And if you did not have reputation systems, everyone was like a avatar with no image and Mickey Mouse 1, 2, 3 account created yesterday. Every, all, all vendors on ebay were like that, you know, the economy were grind to a halt or that that type of economy would grind to a halt. So we have, as societies developed things like reputation systems to sort of lubricate the, you know, the economy is to make, to make people trust one another more and therefore trade more with one another. And over time I think some of these reputations can become very. So I think on ebay, I'm not a huge ebay trader, but I get the sense that, you know, it's very important to have a fairly clean slate. I mean as soon as you get like a few reports or you know of someone who's unhappy with something you do that can become really hard. Your, your business, they draw. And this is, you know, this is what we all know these from like the restaurant business and people, clients, unhappy clients are threatening the restaurant owner are gonna write a bad review of you. They're basically targeting the reputation system, weaponizing it. Yeah. So the jihadis organically developed similar, similar systems and used existing sort of web architectures and platforms to build them. But it was always, it was with a twist that will have over time. I mentioned these sort of agencies, this kind of organizations like Jihadi, Reuters, or, or IFP or the media arm of a given group, they would all develop different identities for the different visual profiles and where that sort of branding was also an important part of the trust building.
- [28:30]
Interviewer (Raman)
So I'm not sure if you're lurking or buying on ebay, but you make very clear that you're lurking only on the Jihadi forums. Yes, you're, you say that you give yourself a nickname with a username account, but that you didn't post or interact. You could have, you have the language ability and you have a lot of contextual knowledge. Why did you hold back? Was it something about your methodology or something about safety or something else?
- [29:01]
Thomas Hegemer
It's a combination. But I think the theoretical scientific argument is the strongest, which is that if you are inserting yourself into your object of study, you are changing it. It's no longer an organic thing unless you're going to pollute your thinking and your inferences. So that was to me the strongest argument always. And you could insert yourself and pretend to be this, that, the other. But then you'd be making friends and those friends might think differently than they would have done if they hadn't met you, and so on. And then of course. In some countries it's illegal, depending on what exactly is it you do. So there are other considerations too, but I think. The scientific one.
- [29:59]
Interviewer (Raman)
Well, let's stick with the conceptual side for a moment. If you imagine someone asking quite similar questions as you are, but taking a more participant observation approach to kind of doing online ethnography and engaging and interacting with those same questions in mind, I mean, how would you, you've defended your lurking, but how would you think about what's gained and loss but lost with more of a participant observation kind of approach to the research.
- [30:33]
Thomas Hegemer
So I should preface this by saying that I'm aware that there are some academics that do this to varying degrees and who would view these things probably be different from me. And I recognize that there's a trade off. You lose someone and you gain, you gain some. One very tangible thing that I lose by not participating is access to privileged spaces. So this has become more important over time. As heat has been put on these online spaces, there's more monitoring, more infiltration and shutdowns, et cetera. So what, you know, in the mid-2000s, the jihadis were discussing quite openly very sensitive things on the open forums. But as pressure was put on them, the more sensitive conversations moved to sort of private or closed, but by invitation only groups. And you, you could only get in there if you, if your profile was seen as worth inviting. And of course I had, you know, I don't know, history or not being one of the cool kids on the forum would not get invited to those spaces. Whereas if you are doing more active participation, you might actually end up there. So yeah, This probably maybe worth having a sort of a longer debate about the trade offs here. But I think there's, if you combine the features of sort of the various dimensions of this problem, the scientific one and the ethical one and the legal one, I always fell down on the kind of the, through the absention side of things.
- [32:45]
Interviewer (Raman)
There's a specific point you raise in your chapter, which is about kind of the limits of online spaces and how ultimately many important interactions happen face to face. And I guess on this specific question I wanted to push you on that. I mean, is there a way to dig deeper into kind of, I guess following the trail in understanding interactions in the online space and then how they persist or not into the face to face space? Or would that require a totally different kind of methodology that would require a whole new apparatus and approach?
- [33:29]
Thomas Hegemer
So for this particular domain, I think it would be really hard unless you are an intelligence service and can follow people in the real world and follow them over time. My window into this was very narrow and quite superficial in some ways. I was only looking up at basically the public square of jihadism. Whereas as you say, as you rightly point out, to really understand what calculations people made and what real life consequences it had, you'd have to follow them from the public square literally to their bedroom and ideally also query them, talk to them about their calculations. Like why did you trust him or her? Why did you not respond to that email? You see what I mean? So you need to go a lot deeper, of course, and I think this is one of the, as one underexplored part of this whole domain of mimicry, accepted mimicry, the calculations that dupes in particular make we often at this point, at least in the research on this, at the system level, sometimes it's at the game level, but there's very little it is that I've seen research that systematically collects the calculations and try to find ways to sort of get
- [35:17]
Interviewer (Raman)
more
- [35:18]
Thomas Hegemer
than anecdote about how people judged and then a pre gain.
- [35:24]
Interviewer (Raman)
I'm thinking about one book you've written in which you step at least one foot into an interactive frame and that's the biography the Caravan, where you don't interview the subject of the biography, but you interview people who were around him. And I wanted to bring that in for this conversation to ask whether, I mean, there were specific reasons why you wrote that book in biographical form to talk about kind of the local dimensions to a global Jihadi movement. But I wonder whether there's also some bigger picture reflection you have on, I guess, whether it's possible in your ongoing work to look at both the public square, as you put it, and things that kind of matter a lot to people's calculations and motivations outside of the public limelight that you might get. Only if you step into an interactive frame or kind of an alternative set of, of of ways of gathering information about someone or something.
- [36:29]
Thomas Hegemer
Can you, can you give an ex have anything Specific in. In mind.
- [36:36]
Interviewer (Raman)
I. I guess. Well, so here is a. More I guess one worry is whether determining the public square as the parameter of inquiry leads to some kind of bias in the questions we ask and the answers we're looking for. And I think writing a biography, I imagine it as an exercise where you're looking. You have to interweave the public dimensions of this one person's life with what happened behind the scenes. And I wanted to ask you to use that example or just use it as a launchpad to. Yeah, I guess I'm pushing you to extrapolate from that one project into a broader question about whether it is possible to link up public and non public dimensions of a phenomenon like global jihadism. But also there are many other examples in this book where it might be helpful to look at both public and. And kind of non public spaces.
- [37:40]
Thomas Hegemer
Oh, sure, sure. Yeah. So first I should say that the research on the jihadi forums and trust building, I see it more as an inquiry into signaling and how signaling work in online spaces more than an inquiry to find out how jihadism as such is working. So in some sense I probably could have done the same type of study in any other high risk, sort of clandestine space in a far left, far right, hardcore nationalist in some part of the world. As long as the conceptually, what was important there is that these are very high stakes decisions that people are making. They make the wrong one, they can end up tortured or killed or arrested. So, and what happened? How does trust work there? But of course, I chose that particular domain because I came from research on jihadism. Yeah, I totally agree that if the question is to do specifically with the domain of militant Islamism, then you have to, you get deeper by combining different types of data and different types of methods. But very often here, as with many other domains, it's usually a question of capacity. So people who do qualitative work often criticize the cons because this by saying, well, it's very superficial. You abstract little things you're studying, for example, recruitment to jihadism by looking only at 10 variables, socioeconomic variables, or other variables that are effectively just like numbers. And then you run into statistic analysis on them so that it's broad but not deep. And conversely, the consul criticize picklejung or kind of quality or ethnographic based work, or saying, well, it's ultimately anecdotes, how to generalize from them, et cetera, et cetera. And it's hard for any one researcher to do everything some ways. But, but I think you know, as. But people, I think on average are doing more and more like you. You see more scholars doing a broader range of things. Like, you know, where I see at least the political science people combining several approaches, often at quite an advanced level in a way that people did not do 20, 20 years ago. The bar has been raised and it probably is going to be raised even further now. With artificial intelligence, it becomes easier to do certain statistical tests or certain large
- [40:52]
Interviewer (Raman)
scale textual analysis in terms of putting hypotheses under test. It's obviously partly about work in social sciences. But you also say in this chapter that there are direct implications for counterterrorism policy. And in fact, counterterrorism policy is one domain in which there are assumptions about how the online and the offline link up, because any policy of that kind would need to make such assumptions. What are the key assumptions in that space that you think academic research can and should put under pressure or to test?
- [41:37]
Thomas Hegemer
You mean assumptions that governments seem to fold?
- [41:41]
Interviewer (Raman)
Yeah, about how online spaces link up with offline spaces. I mean, I'm thinking about assumptions like what radical talk means. I'm thinking about assumptions like how someone's considerations offline can be deduced from various metrics of their online presence. Are there, I mean, this could go on, but are there specific assumptions you see in that counterterrorism policy space that you would like your research to speak to most directly?
- [42:17]
Thomas Hegemer
Not really. I mean, not for this particular project. In a way. I honestly don't think I went into it in order to kind of help intelligence services infiltrate better. And that is a byproduct of the research, I think. But you could also flip it and say, you know, there are elements of the research that can help the jihadis operate better. You know, and in some, this is a feature of, by the way of signaling generally as a field, it's sort of an amoral, filial way. Everybody can use it. And everything has a flip side. So if things that will help will empower users of Jihadi forums will also empower pro democracy activists in a high repression context, because the structural, the structure of the arena which they're operating is the same.
- [43:31]
Interviewer (Raman)
But
- [43:33]
Thomas Hegemer
of course, there are lots of conceptions that you come across in the public debate around the significance of online extremism that I think can be debated, or some of them that I'm just disagreeing with. But often I think they come not from. They come neither from academics who specialize in it, nor from people in government who work seriously with this. They come more from sort of pundits, but sort of your observers or generalists who have an interest in this. So I don't know, I think one blind alley in the space is to think that the Internet has one sort of effect on something, that the Internet inherently helps extremists or that the Internet inherently does one thing or another. Because I think that's one thing we've learned over the decades, is that this could change. So I think it's much more useful to think of this as a game, as an evolutionary system as, as it were. And we can see this very, you know, we don't need to read the, the titles of works in this space over from the 90s to today. In the beginning, we talked about the Internet as a liberation technology. This was going to be, this was going to bring, you know, peoples together and, and, and, and marginalized groups, good causes. We're going to be able to use this to get their voices heard and et cetera, et cetera. And then, you know, things flipped and you understood that states could actually leverage the Internet to control to, to, to enhance repression or to manipulate people's opinions and that sort of thing. So, and the same thing with, if you bring it more narrowly to, to radical, to sort of extremist sites like, you know, ISIS sites or neo Nazi sites and so on too, how useful they are to the militants depends entirely on the level of repression that they are facing. So a jihadi forum or a neo Nazi forum that is completely shielded from infiltration or bad publicity or anything where they can just operate completely on their own is operationally empowering. It means that they can freely, you use the online digital communications for what it's best at, which is reducing transaction costs and speed of communications and sending of money and all kinds of things. They can coordinate meetups and training camps and attacks. Even imagine if there was completely safe spaces online that people could coordinate plan attacks. So the amount of the utility of an online tool depends entirely on the, on what sort of their opponents are doing. And that's. We've seen that over the past decade where jihadi forums have gone from being very lively, consequential places because they were under less scrutiny, to now being places there much where it's much harder to kind of do anything serious. There are still spaces in there, they're still sharing propaganda and kind of winding each other up. It's very, very rare to see that they are sharing phone numbers or how are numbers or using it for something consequential because that will get spotted and exported immediately by Intelligence. So point being that one I think one often sort of misconception about the online world and extremism is this notion that there is an inherent kind of effect. That being said, there may be kind of general directions that we can speak of. It can fluctuate but the, the curve can fluctuate and hydropensibility can still have a long curve in one or the other way. Numerous isn't. I think there's, there's, there's more and more that makes me think that it been the really the very long term effect of this is empowering States written
- [48:15]
Interviewer (Raman)
about this too because of the repression effect that once they know how to wield it they have kind of first mover advantage of determining what the spaces in which activists are operating in or something else.
- [48:28]
Thomas Hegemer
I think it's more about accumulation that nates have a much larger ability to accumulate knowledge than a militant organization. Militant organization's have a short lifespan, so the time they usually involve mainly young men. So even the individuals in the organizations are only in the gang for five, 10, maybe 15 years. Estate on the other hand, lives forever in a way. And the bureaucracies. And advanced intelligence services in service in the west is an incredibly complex system that just accumulates, accumulates, accumulates not just data but institutional memory in our whole kind of army of analysts who worked there their entire career. So it's just. So what I've seen at least for the years is in this technology race is that rebels are the first movers. They're the ones who pick up a new opportunity provided by tech innovation. For a while it gets used to good effect. The rebels can separate for a while, but then the state watches, sees this and takes action against it and finds ways to make it less useful and, and. The balance of power switches it in until some new innovation comes along.
- [50:13]
Interviewer (Raman)
There's a bit of a tension here. I mean there's this insistence about the Internet being indeterminate in its effects. It depends on various conditions. And then there is this at least hinted at a long run view that kind of the state wins. And I guess I, I might be tending a bit towards the latter. I'm thinking about a case like the Chinese Internet which every Internet ecosystem has its fluctuations, but I think probably a bit less dramatic than the jihadi forums you're looking at in sense of there are openings and closures but on the whole you see for the past 15, 20 years a state that has become very good at wielding the Internet for its purposes, whether it's informational or direct repression. Otherwise Fluctuations seem minor in relation to that arc. Obviously that's one country, not the whole world. But what's the strongest evidence that this more you might call it pessimistic long run view might not transpire in the end that it might be more dynamic than kind of this just gradual accumulation of state power?
- [51:39]
Thomas Hegemer
Well, first of all I should say that I probably, although we're in the guessworks state, my guess would also be that ultimately I don't know what the time frame is probably heading in that direction. But the question, I guess your question is or is your question what would it take to reverse that long term trend?
- [51:59]
Interviewer (Raman)
Yeah, what would it take for that alternative to that long term trend to be plausible?
- [52:07]
Thomas Hegemer
Well, I would need to see non state actors seeking to to make substantive changes society survive in the competition with a state for a certain period of time. For a rebel group, for lack of a better generic term to not be ground down by a state appointment over time. I would need to see. I don't know. Yeah. Any type of. Whether it's sort of a type of extremism that we don't like or godism or far right stuff or whether it's stuff that we do like or pro democracy activism or things like that. Any non state actor with a controversial objective that was where a lot is at stake trying to achieve something that some state really doesn't want to happen. And if they are able to leverage the digital communications to their advantage over time that would be the evidence that I haven't seen to be. But I've yet to see that all through so far in all of these sort of competition episodes, some of which have lasted maybe five years, some 20 years, the state had tempted to win.
- [53:39]
Interviewer (Raman)
You mentioned AI just now. My hunch is that that tips the scales in favor of states. Do you agree?
- [53:47]
Thomas Hegemer
100%. And I mean these people are still not. I know there are people who don't know, there are people who disagree with me. And much has been and is being written about potential users of AI by terrorism to terrorist groups and so on. And although I think in some, some ways maybe answering different questions, I don't disagree that some groups or some individuals might use AI once or twice or a few times creatively to their advantage. But I would be very surprised if at the strategic level states are able to leverage AI for repression much more efficiently than rebel groups will be able to leverage AI and for their rebellion. It's a question, as I said earlier, about capacity for accumulation. It's about money, is about access to the Hardware it's about access to the manpower and the best scientists and so on. And it's just if we look at the complexity and the sophistication involved and developing really good models now and we compare that to what an actual rebel group looks like on the ground and that in places where we're most likely to find them is just. I struggle to see how it can be leveraged strategically. Yeah, you will see some groups that try and use it a few times but then the states will find out how they did it, they will find out what the voter necks are, what to look for and then the next time around it'll be harder.
- [55:42]
Interviewer (Raman)
Is my house before we come to a close, I wanted to bring back in time but stick with AI. So I mean it does, you know, the cost becomes so much lower with AI and so I'm wondering what time signaling. Yeah. So how does that change time as a signaling device? Because I mean frequency for example, doesn't really seem to. It gets even kind of cheaper. So how do you anticipate time as a signaling device changing with technology?
- [56:24]
Thomas Hegemer
Well, for some sort of poorly regulated places like jihadi forums, it is going to be a big problem because the baseline knowledge among all users will be that it's perfectly possible now to manipulate the metrics of longevity. So you know, in 2002 very few people could very, could easily sort of create a copy of say a forum or manipulate, knew how to manipulate like the joined date or the number of contributions can for example that would take like infrastructure and access because now it's much easier to just spin up an entirely fake page site or just to manipulate digital information more generally. So it's not, maybe not that time as such is going to lose value but you know, the signs, the proxies for time are more easily manipulated. So in some sense the signs go back to the signs that we use to judge someone's time investment in the cause are going to be harder to read. In a way this is one out of many, many examples of things that are being changed by AI in this space. So yeah, as you said, the cost of previously very costly size is being driven down to zero. And that is a problem. Now I don't think many people know how it will play out, not least because the technology itself is evolving so fast. And one of the big questions is of course will we have. Yeah, okay, AI makes it easier to lie and deceive, but it's conceivable that you might also use AI to detect and stop deception for the moment not there. It's very, very hard to detect fakes, all those, you know, there are companies that, there are people who are working on this problem, but as far as I can tell, yeah, deepfakes are still a real problem and it's creeping into ever new domains. So as we're questioning, can someone, to put it very simply and bluntly, can someone develop an AI decision detect system that will kind of counteract the effects that AI is having on the opportunities for receptive mimicry? That's a big question. And I listen to a lot of AI podcasts and, and stuff like that. I have yet to hear. Like, I don't see signs on the horizon of like a solution right now.
- [59:49]
Interviewer (Raman)
So just to kind of bring things together, I mean what we get from a face to face interaction when we go into an online one is a narrowing of bandwidth and the possibility
- [60:02]
Thomas Hegemer
that you are being tricked into. You're being tricked by a host of. Yes.
- [60:10]
Interviewer (Raman)
And in an age of AI, I just want to check that whether I understand your perspective. Is it that we remain with the similarly narrow bandwidth of signs that we have in general on the Internet and then we have an intensification of the ways in which one might be duped or is there some other transformation that's taking place as well that I'm missing with that characterization?
- [60:38]
Thomas Hegemer
So I think one way of thinking about it is just to put yourself in the shoes of a militant of some kind. If you, let's say you had a cause that you really believe in and you wanted to carry out some Cl to draw attention to that course, like in Britain today or somewhere else, like how would you go about finding and you needed to draw in new people to your course, you know, to get it, you know, find people willing to, to commit a serious crime or to join, to join you. And knowing that you know all the capabilities of the police and the risks of the Internet. I don't know, I mean, in some sense my inclination would be to think that with AI you can't actually trust anything in the online space at this point. You have four like high stakes decisions. So if I was a, you know, a terrorist group leader, I would be probably go, go analog. So either you go completely analog, you go old school, or you embrace the, and lean in with technology and try to find some new technological solutions to these things. But I think at least many of the old tactics in the online space used by rebels and disciplines and whatnot are, are going to be quite risky with AI. But this is one of these are just hundreds of mip. And I think the effect of AI on deceptive mimicry is another major area of research that remains.
- [62:43]
Interviewer (Raman)
Well, and watch this space, because I know that Thomas's doing work on that right now. So I think that's a good place to end on. So thanks a lot for talking with me today.
- [62:52]
Thomas Hegemer
Thanks so much.