Albert-László Barabási is a professor of network science and physics at Northeastern University. He has pioneered the study of complex networks, discovering that most systems are not randomly structured, but contain highly connected hubs. He has authored popular books such as Linked: The New Science of Networks; Bursts: The Hidden Pattern Behind Everything We Do; and most recently, The Science of Science.
Transcripts of our episodes are made available as soon as possible. They are not fully edited for grammar or spelling.
Tom Burnett: Laszlo, welcome to the show.
Albert-Laszlo Barabasi: Tom, it’s a pleasure. I’m a big fan of your show. I, I listen to you regularly, so it’s wonderful to put the face on the voice.
Tom Burnett: Wonderful. I want to start by asking you a question that’s much easier to ask than to answer, but given your study of network theory, we live in a complicated world. It’s messy. Things are entangled, whether you look at biology, culture, technology, politics, you name it. If we want to start to understand this messy, complex world we live in, where do you think is a good place for us to start?
Albert-Laszlo Barabasi: Always think it’s the architecture is the first question you need to ask. That is, like, how the pieces of the system are connected. Because as yourself said, we live in a very complicated world, but what makes it complicated? When I started in science, I started with chaos theory, and we always thought it’s nonlinearity.
And I think my perspective has fundamentally changed in the last 30 years. Nonlinear is important, but the key is the complexity coming from the many diverse components connected by very diverse interactions. You name it, whatever complex system you care about, whether a society, whether a cell, whether a brain, it’s a system that is made of very heterogeneous component that are connected in a non-trivial way to each other.
In order to understand that, you first have to figure out the architecture, who’s talking to whom, how often, how strongly.
Tom Burnett: I know you’ve studied a lot of different complex systems in your career. Can you tell us where you started, like, your first big project on tackling a big, complex system?
Albert-Laszlo Barabasi: The first system that was consequential for me was the World Wide Web.
This was kind of, like, end of 1990s, and the web was just becoming what it is this moment. At that time, had only 800 million nodes, which is tiny, tiny, tiny. A, a, a university has more nodes than that today, right? And it was the piece of engineering, but it’s engineering itself, and that is that the network is growing and evolving what’s behind the World Wide Web, and there’s no central control over it, and that became a metaphor for much of my career and much of thinking about complex systems and networks in the many decades to come.
Tom Burnett: Yeah. What were the big takeaways in terms of studying the architecture as something as, even at that time, as voluminous as a World Wide Web?
Albert-Laszlo Barabasi: The biggest surprise was that it was not random. And of course, the question is, what do you mean by random, right? We know exactly what we mean by random because before us in the 1960s, both mathematicians and sociologists have worked a lot on what we call random networks because they said, “Listen, the world is very complicated, but we don’t know how things are connected, so why don’t we just assume that it’s all random?”
That is a coin toss decides whether you and I are friends or not. And so pretty much from the ’60s till the late ’90s, no matter what area you looked at of complex systems, there was an inherent assumption, and often not actually spoken out, but it was in there in the math, that the complex system we study are either purely regular, like a square lattice, or purely random.
And then we came around in the 1990s, thanks to the World Wide Web, and we realized it’s none of that.
Tom Burnett: So what features did the World Wide Web demonstrate or reveal to you that made you think this is not random, we really need to rethink how we understand networks?
Albert-Laszlo Barabasi: The hubs, the highly connected nodes, which in the hindsight is obvious. But what you need to know that in a random network, we are all alike. If the society would be truly random, that is pure coin toss decide whether we are friends or not, then you and I would have approximately the same number of friends. Very, very close, right? It would be hardly any variability. And in the World Wide Web was the first case where we clearly saw in the network context that this is not even closely resembling the reality.
Rather, you have many, many web pages in the case of the World Wide Web with very, very few links, and then you have a few major hubs that all hold together. Everybody connects to the hubs, to Google, to Facebook, you name it, right? And the top structure was a total surprise. It was an animal that did not exist in the scientific imagination up to that moment.
Tom Burnett: So having a network, there’s many things and only a very small number of them have a huge number of connections, and almost everything else has a very small number of connections. That is not random. And that is not random. And
Albert-Laszlo Barabasi: there was a precise math, right? In the case of the random networks, the number of links follow what we call a Poisson distribution or a bell curve, right?
In the case of the worldwide web, we discovered that this is a power law, which suddenly took out the worldwide web from this random system paradigm and put it into different space. Something that physicists like myself were very familiar with because since 1960s, physicists were very busy with phase transitions and critical phenomena, and phase transition and critical phenomena had lots of power laws within that.
It was about self-organization. It was about emergence. It’s about emergence of properties that are not just simply the sum of the pieces, but together we do something different. And then of course, by itself, the worldwide web would had not been surprising. But then we and others started to look at many other networks and we realized, wait a second, there is no network out there that would be random.
Tom Burnett: Really? None at all. Basically, it went from you thought nearly everything to nothing?
Albert-Laszlo Barabasi: I cannot think of a system, and I, I’ve been studying these systems for a while, that would be truly described as a random network. They don’t exist. They’re beautiful mathematical construction, and this is where I start my class in network science.
And at the end of the class, I tell them, “Okay, let me tell you the truth. There’s no system out there that will be random.” So then you may ask, why did we waste an hour and a half studying random networks? And the answer is because in order to understand order, you first need to understand randomness, because order is a deviation from randomness.
So without understanding random networks, you will never understand emergence. So it’s a necessary step in our journey that we cannot skip to really understand how the world is organized.
Tom Burnett: I think in 2026, it’s much more intuitive for us to imagine that the World Wide Web has these massive hubs, and then nearly everybody’s personal page and all these other sites are slightly linked. What are examples of some of the other networks that you’ve explored out in the wild that also exhibit these same features perhaps we can, I think, intuitively relate to as non-experts?
Albert-Laszlo Barabasi: Absolutely. I mean, the first was obviously the World Wide Web, and immediately after the structure of the internet, which in people’s mind is the same, but it’s not, because the Internet is the infrastructure of how computers are connected by cables, by, uh, optical lines and so on. But then soon after, we started to look at the actor network, how Hollywood actors are connected through joint movies, and that also followed the same structure.
There were a few actors that were seen to be in all movies, right? And then most actors are very specific character actors in one or two movies, right? And then we actually started to look at the social network when the data emerged. Obviously, once Facebook and Twitter emerged, then the data showed they follow exactly the same pattern as the World Wide Web.
And these are all human-made networks, right? With something that we created and kind of made you think, “Hey, maybe this is something that we like to do,” like we like to create hubs. And that was until we actually started to look at the networks within the cell. First, the metabolic network of how the chemical reactions are connecting the chemicals together, and then the, what we call the protein interaction network.
And both had the same property, that there were a few major hubs that hold the whole network together. And let’s just say the technical term today, we call these networks scale-free networks because there’s no intrinsic scale in the system. Many small nodes coexist with the big nodes, so the scales coexist.
And we found, and others actually have found in others, biological systems are fundamentally also following this paradigm. And what’s interesting about it is the World Wide Web at that time has a history of twenty, thirty years, right? The internet maybe forty, fifty years. But the cell is a four billion year history.
So then the big question really is how is it possible that we created the same architecture today as we built our internet and social networks and so on, that life has converged to over billions of years of evolution? And that is really the mystery of the networks, and that’s the beauty of the networks, and that’s what we physicists call a universality.
Tom Burnett: Can you explore that why, or should we be very satisfied with just knowing the what and that there is that deep congruence?
Albert-Laszlo Barabasi: As an intellectual construct, you can be very satisfied, but as a scientist, you cannot, right? And as a scientist, you need to answer the why, and we did answer it in ’99, kind of my claim to fame, right?
My biggest impact paper that I ever wrote together with Réka Albert, who was at that time my student, was the explanation. And the explanation is very simple. Networks are not static objects. Every network that you see out there is the result of some kind of growth process. You start with nodes, then you add more nodes and more nodes.
And so when networks grow, they behave differently from the static randomly connected nodes. So that was feature one, growth. And the second feature, what we call preferential attachment, which is when the new node comes into the system, it’s much more likely to connect to more connected nodes because those are easy to find.
They’re more visible. They’re more accessible than the less connected nodes. So these two things, growth and preferential attachment, are mathematically now proven to be responsible for the emergence of the hubs. And the mechanism behind is as simple as if you have no friends, you don’t have, don’t have anybody to introduce you to new people.
If you have lots of friends, you’re gonna gain many new friends.
Tom Burnett: I wanna drop in a literary reference here. In Tolstoy’s novel Anna Karenina, it starts with, “All happy families are alike. Each unhappy family is unhappy in its own way.” How does this apply to networks that you’ve examined in the natural world?
Albert-Laszlo Barabasi: It does. It does. And that’s kind of the tension between universality and specificity, right? Yes, there are architectural features of networks are, that are shared across many, many different systems, but at the end, the World Wide Web is very different from the social network or the network within this, right?
And that’s really where the meat of the network science field is coming from, is to understand not only what is common, but also what is different and what makes it different, and how does the differences enable functions that you have in one system and you don’t have in the other one.
Tom Burnett: I wanna turn next to a topic that you brought up, the networks that we find in the biological world that has been present on this planet for billions of years, and it’s complex at multiple levels.
We have individual cells, building blocks of life. We’ve got brains, which are networks of millions and billions and trillions of cells. Then we’ve got billions of brains that are networked together. So I wanna kinda explore those three levels together with you at cell, Brain, and then large populations of organisms.
So the cell. In one sense, I think we see them as simple or primitive. As you look at individual cells with your network theory and your scientific tools, do you see simplicity or complexity in just one individual cell?
Albert-Laszlo Barabasi: Certainly complexity, and let me tell you why the cell is certainly very amazing. The cell is the first network in the history of the universe, at least the slice that we know about the history of the universe, that has learned to reproduce itself.
And this is not trivial, right? Because we have networks before the cell, like the networks within the sun, right? For example, the nuclear networks in the sun. They don’t reproduce. So the cell was the first network that successfully reproduces over and over, and I’m really fascinated of how it does it, right?
How it does it with the fidelity that is required for the survival not only of the cell, but of the organism.
Tom Burnett: I wonder, too, in terms of, like, the sort of chicken and egg problem. As far as I understand it, in order to reproduce DNA, you need the products of DNA, the various proteins and molecules in order to do it.
So have you wrestled with that kind of problem of what came first, and how, how do you get a kind of reproductive device that requires the very outputs of its reproduction ability?
Albert-Laszlo Barabasi: I am fascinated about reading about it, but I also think that one could look at the higher level and say it doesn’t matter, right?
Of course it does matter. Yes, of course. We humans wanna know how it happened, right? But also, if you look at from a bigger perspective, what matters is somehow this new function was kind of given to networks. Networks got endowed with the possibility of reproduction, and whether that happens the tape first and the pieces after or the pieces first and the DNA to record that after, it’s a very, very important details that you can spend a lifetime on it, and it’s will be a satisfying lifetime.
But you can also look at it from a distance and say what a miracle that happened there, and what an interesting endowment that the networks were given to, right? And then just run with that, which is by itself is amazing.
Tom Burnett: Yeah. That reminds me a quote from E.O. Wilson that he said the universe is biophilic.
So there’s something just fundamental to the structure of the way the universe is, is that life emerges from it, and you used the word endowed. There’s just, there’s something there, and that life’s gonna come about in this kind of universe that we’re in. That’s what we have, and maybe that’s as far back as we can go, though it’s not gonna stop us from trying to go further back.
Albert-Laszlo Barabasi: Yep. It’s one of the most fascinating questions of biology in a way, or like whether is there a driving force that towards this more complexity, and some people think it is, or is it just a sum of many lucky accidents? And I don’t think we have the scientific language to fully answer that, nor do we have the receptiveness of the scientific community to do so.
Tom Burnett: I want to turn now from cells to a higher level. Billions of years, we’ve had unicellular creatures. Then I was talking to Simon Conway Morris and this Cambrian explosion of many new and different multicellular organisms. More evolution, more time passes, creatures with nervous systems develop, and at some point you develop brains.
These are networks of many, many, many individual cells. So pick a species and just tell me a little bit what is the architecture of a brain?
Albert-Laszlo Barabasi: So this is my other fascination and the parallels between the cells and the brain. So we know since Ramón Cajal that the brain is a network, right? Actually, the first medicine Nobel Prize was given for the discovery that the brain is a network.
Yet we didn’t know that network until a few years ago. That is, we had no means of mapping it out. And only about ten years ago, the first connectome map started to emerge. And now we have actually full connectome maps for not only the C elegans, which is little worm with three hundred neurons, but also for the fruit fly, the Drosophila, that has about an order of ten thousand neurons.
And we have segments of the human and of other higher level organism brains. So we live in a very special moment now because for the first time, we actually started to see the network. And what is the network? The neurons themselves, which are very elongated objects. We always tend to think of a network as nodes connected by links.
But really the neuron is a very weird animal because it’s a very elongated cell that simultaneously serves as nodes as well as links. ‘Cause it has these long axons that kind of go out, and wherever they meet with other neurons, then they form synapses between them. And that architecture has really become available only in the last kind of four or five years to us.
Tom Burnett: So I am imagining now this brain, I’m imagining it as a complex network of which you and I have talked about several so far. I guess maybe it’s twofold. One, does the brain have these like hubs Like we’re imagining with World Wide Web and internet and other things. And two, how is that brain controlled?
Albert-Laszlo Barabasi: So this is a big question, right? Because as soon as kind of Scalfi networks were discovered, people immediately tried to apply it to the brain. But at that time, the brain only had three hundred neurons, and it was too small to decide. And only recently we had the opportunity to look more carefully into that, and we did so.
Actually, I have a project with my son, who’s a brain scientist, so it seems to be a family affair. And he and I were actually looking at the structure of these connectomes that have emerged in the last years. And the answer is yes, the brain does have hubs, and that’s by itself not surprising, because, uh, brain scientists have documented a couple of cells that have tens of thousands of synapses to other ones.
But we actually were able to see the precise mathematical structure. And the hubs are never as big as on the World Wide Web. That is, they don’t grow as far, they don’t go in as big in size. And the reason they don’t is because the brain is a physical network. That is that at the end, whom you can connect depends on how many other nodes you can connect to, depends on how many nodes you can get nearby.
And that is limited by the physicality, right? On the World Wide Web, you have no physical limitations, right? As far as I’m concerned, a billion people can connect to your website. It would not require more resources on your end to accept them, right? But when it comes to the brain, the neurons need to find each other and need to get physically next to each other.
Tom Burnett: So how does that affect then how the network operates, given those kind of limitations to the physical architecture?
Albert-Laszlo Barabasi: It does, and we’re still figuring out how. So let me tell you one aspect of it, right? Many processes in the brain involve synchronization. And fundamentally, the big hubs let you synchronize a system in a much more effective way than the smaller hubs that we see in the case of the brain.
This could probably be good news because pure synchronization is really a disease, right? Schizophrenia is one of those, right ? You don’t want global synchronization in the brain. You want local synchronization, right? So to some degree you could say that the lack of the brain’s ability to synchronize because the hubs are not so big allows us to have multiple thoughts at the same time in our brain, right?
Allows us not to be dominated by one process, but to have multiple processes running side by side. And actually one of the things that we see is that the presence of a somewhat smaller hubs are affecting many, many different features of what we know that happens in the brain in a quantitative way, and it kind of limits them, and that’s probably a feature and not a bug.
Tom Burnett: I imagine just intuitively for a biological creature to live, it needs many things happening simultaneously and independently. All my organs need to work. All my senses need to work. We’re taking in all these different kinds of data, both inside our bodies and from without. And so somewhere in the brain there is something unified somewhere.
Yes, and or
Albert-Laszlo Barabasi: not. I, I tend to think it’s not unified, but rather allows the many independent processes to run in parallel so they’re compartmentalized. And I think that’s the key of not having to be hubs, uh, because it allows you to be a local boss and not a global boss in the system.
Tom Burnett: Artificial intelligence and computers. We’re talking about networks and structure and architecture. Computers now are able to do many amazing things that at least at one level look like human intelligence. Does the architecture of computers have anything to do with, say, the architecture of brains?
Albert-Laszlo Barabasi: We’re gonna get into deep waters with this conversation- Okay but let’s try. So first of all, let’s distinguish the notion of the computer from AI, right? A von Neumann computer is a miracle, but it is a tool for fast computation, and the AI is not a von Neumann computer, right? And in my mind, AI is brain dematerialized because if you look at the history of AI, it has a very checkered history with many kind of stops and breaks and then restarts.
And every time there’s a restart, it comes from an idea that is emulating closer and closer the brain. So fundamentally, what is the AI? Is a neural network where the weights are trained to carry information. What is a human brain? Is a neural networks whose weights are trained to carry information. What is different between them?
The AI is trained over a short period of time in historic context with a massive amount of data. You gotta show them ten thousand rabbits to get a rabbit, right? The brain is pre-trained over millions of years of evolution. We are not an open source computer. That’s the difference. The underlying mechanisms are really the same.
It’s a network that has acquired intelligence through the many, many weights it has.
Tom Burnett: Another feature of biology that fascinates me, the next level up besides individual brains, is brains working together What they’re able to do collectively is extraordinary. So maybe in the case of ants, how does that collective intelligence arise and their ability to act in the world in ways that no single member of that population could do?
My colleagues
Albert-Laszlo Barabasi: actually in ecology have mapped out the networks that they have, and they build very reproducible networks among themselves. But very interestingly, the role that the ant gets in the network is not born with it, but it’s really defined by the environment. So in that sense, I think it’s emulating a little bit the brain’s intelligence by giving functions that are environmental dependent.
But then you can step actually a step further and say, let’s talk about humans, right? And let’s talk about the much more complicated networks that humans organize that eventually we do the same, right? We are born all equal with an open mind. And yes, you become a podcast host, right? And probably many other role, and I a physicist and a network scientist, right?
So the same type of differentiation happens in the society. And what is amazing about these networks is that the redundancies existing in these networks at the same time gives the, the possibility to carry out the functions that we need, which is to teach the network science class that I do, but also to be robust, because if I step out, there’s somebody else who can do the same thing, right?
So there’s this redundancy in the system and adaptability. And by adaptability, I mean when I finished my degree, there was no network science, and now I teach, right? So I think it’s pretty amazing of how we build this kind of ability of the networks to adapt to the environment in real time and to come up with the new functions that that environment needs at that moment and abundant old functions that were not necessary.
And in a way, I think the human society is more interesting than the ants because the ants, for now, they’re reproducing the same network over and over, just like the robot lab brain does. The human brain and the human society is always reinventing and adjusting to the environment in a way that, yes, of course, ants can do too, but I think it’s fair to say that one of the reason that it is you and I having this conversation and not two ants is because we managed to learn to do that at a higher level.
Tom Burnett: I want to turn to our final section here, and it’s a combination of looking forward and looking back. Ten years from now, how does scientific research look in a decade?
Albert-Laszlo Barabasi: Well, I mean, it’s very interesting because of course now in everybody’s mind is AI, right? And whether it will replace us or it will just be a partner.
And I’m of the school that as amazing AI is, it’s still very dumb and will continue to be dumb. This is because it’s a very intelligent executor, but not yet the source of ideas. And the reason why it cannot be the source of ideas is because ideas are not objective. Ideas are driven by human interest This is not to say that there is not a fixed s- set of mysteries to describe how the universe works.
But when it comes to a complex system, there’s so many different ways we can evolve and we can go with them, and hence it is really driven by our own interest, our own affinities, our own moral compass, our own kind of sense of direction that determines where we want to go. And the question is, will that ever be replaced by AI?
I don’t only hope that it will not, but I also think it’s not possible. That’s one of the things that were left to us. But on the other hand, we got a fantastic assistant, one that Really empowers us and supercharges our ability to go in those directions. And I think as time goes on, that will be even more supercharge our ability.
So yes, 10 years from now, science will be super different, but it will be different in our ability to execute on the ideas and to come up with ones and so on. And once you come up with the idea to kind of test it, its validity and its reach, and I’m really looking forward to those changes that we’re going through now and what new capabilities we’ll get in the future.
Tom Burnett: It makes me wonder about the network of the scientific community. And I’m wondering if you could speak to that specifically, what kind of features it has, like hubs, growth, preferential attachment. How do you see science in this sort of network perspective?
Albert-Laszlo Barabasi: Well, this is a fascinating question, and it’s been actually fascinating me for the last fifteen years.
And I even written two books about that, particularly one called Science of Science. And Science of Science now is a very robust field that is really using the tools of network science and data science to understand how science works, how the scientific community works, and how innovation happens, and when it happens, and how we could supercharge it and empower it, right?
Thanks to digitalization, we have a full record of all scientific accomplishments. We have every single paper published in the history of humanity. We know their authors, we know where they’ve worked, we know what was the impact of the paper, we know who it inspired and who did not notice it, and so on. So we have opportunity to look at an intellectual achievement, which is what science is, and totally track its history and understand what worked and what didn’t.
And what did we learn through that? Well, we learned that there are very reproducible laws and mechanisms that describe both the emergence of impact in science, as well as the emergence of impactful individuals and collaborators. Yes, we do have hubs, individuals who get exceptional number of citations.
We have collaborative hubs, right? Individuals who have thousands of collaborators on a regular basis, right? They can animate many other individuals to work with them, right? And there are funding hubs, right? Individuals who get a huge amount of money to achieve some bigger goal. And within that, within those structure elements, we see lots of conditions under will things work and under which things don’t work.
So science of science are asking questions. How do you put together a good collaborative team? When is the right time to attack a question? What is the role of a superstar in a department? How do you really measure of an impact of a paper who is not cited any longer, like Einstein’s relativity, because you don’t need to cite it.
Everybody knows about that, right? And yet it’s impactful. And of course, the big gorilla in the room these days, how is AI changing all of this? And we have colleagues who have measured how that now as much as forty percent of the papers in some areas are AI assisted in the sense that their writing is already heavily using AI.
It’s a very ongoing and very fascinating problem now.
Tom Burnett: Sure. Let me ask you one more question about science itself. What are the advantages and disadvantages of whether you have a Small, really smart team working together versus a really big team doing something like a project in, like, at CERN with a super expensive super collider and thousands of people involved.
You know, are there pros and cons either way?
Albert-Laszlo Barabasi: Actually a question that science of science has answered, and there’s a beautiful nature paper about that, and if I wanna paraphrase and summarize, new ideas are born in small groups. The r- role of big groups is to clean up the ideas and find definitive validation for them.
Tom Burnett: Looking back and reflecting on yourself, how has your study of networks in your career influenced the way that you see yourself and your place in the world?
Albert-Laszlo Barabasi: I was trained as a physicist, and like many of my colleagues, we’re very, very introvert people. But then when I started to study networks, I realized this has served me very well until now, but really, if I wanna be successful both in my career and I wanna have an impact on my colleagues, I need to talk to them.
That little realization actually has ended up defining much of my life, and this is real. I remember when I became an assistant professor at a relatively young age at University of Notre Dame, and I went to my first conference. I had convinced myself that I should go up to the senior members of my field and say, “Hey, I’m Laszlo, and now I have faculty, and would you like to have dinner or lunch?”
And this was a major effort for me, and I was very pleasantly surprised that none of them bit my head off. And this kind of mini realization has followed me throughout life. If we wanna be part of a discovery, if we wanna be part of the society, we have to engage with each other, and I transitioned from being a very deeply introvert to what people perceive me as being much more socially inclined and so on.
What they don’t understand that it’s an acquired skill. I acquired it because I was inspired by the need that I need to be a part of the networks that matter for me.
Tom Burnett: Who would have thought that studying physics would make you more social and extrovert, Laszlo? That’s- Yep … unexpected development. That’s right, but it did happen.
Laszlo, this has been a f- very fun conversation for– Thanks for joining me today
Albert-Laszlo Barabasi: . Well, this was fun. I’m so glad that I finally made it to the Templeton program.
