Control, Cell Biology and Success – interview with Albert-László Barabási

Albert-László Barabási is one of the best known and most cited figures of Hungarian scientific life in our days. His name is closely entwined with the most dynamically developing area of our age, network research. The researcher held a public lecture on February 18, 2016 about network theory in the framework of the PAGEO Club. Below you will find an interview with Albert-László Barabási.

Albert-László Barabási was born in Karcfalva, Transylvania into a family of humanities scholars. He attended high school in Csíkszereda, where he studied sculpting, then won the student Physics Olympiad. He received a Master’s degree in Fractals at Eötvös Loránd University, then he received his Ph.D. at Boston University in 1994 under the guidance of H. Eugene Stanley. Thereafter he was employed by IBM, where he first came into close contact with network theory, which was to become his specialty later on. His work led to the recognition of scale-free networks in 1999, when he created the Albert-Barabási-model, which describes the structure of the world wide web as well as complex metabolic networks and genetic systems. Until 2007 he was professor at University of Notre Dame, Indiana. Today he teaches at Northeastern University in Boston, where he is director of the institution’s Complex Network Research Center. In addition, he teaches seminars at Harvard University’s Medical Faculty. He is a member of the American Physical Society, the Hungarian Academy of Sciences and an external member of Academia Europea. In 2003 he was elected Scientist of the Year by Wired magazine. The world-famous network researcher gave a talk on February 18, 2016 in the framework of the PAGEO Club on how networks permeate our lives. Then he gave an interview to HuG magazine about his current research, and the links between network theory and geopolitics.

 

What are your most important research directions at the moment? What kinds of questions do you seek to respond to?

At the moment, we have three teams that do lab research. One question we are exploring is the control of networks: in the past fifteen years we have studied what networks look like, but now we are ever more confronted with what happens on these networks and what kinds of tools we have in order to exert some influence on network processes. We are calling this control, in part because the toolkit we are using for this purpose originates from control theory in engineering science. We are attempting to extend its use to networks. The other direction is understanding the networks within cell biology, explicitly focusing on understanding diseases. Our goal is to find a cure for existing diseases, or to improve the condition of the ill. The third direction unites issues related to social questions: more precisely, we study success, in particular scientific success, how one network node comes to stand out from others, how we can describe a scientific career quantitatively, how we can predict the potential effect of a Ph.D. thesis in numeric terms. These three research projects are running at the moment: control, cell biology, and success.

The central interest of your book, Linked, was computer networks, a topic that your latest volume, Bursts, also discusses. At Harvard, you researched cell networks. Which network interests you the most today?

For us topics open up and close continuously. We have not done any research on the Internet for ten years. The book Bursts, for instance, was about human locomotion, but that research is now over, too. As topics become too old, we close them down, we have no interest in carrying skeletons along.

Do others not continue them, either?

Yes, others research them, but there is a certain toolkit in our hands and applying it we can get a certain level of result. Thereafter we could only get derivatives of the same question, and we could not get anything substantially novel out of it. At those times, if a new, exciting topic comes up, we redeploy our researchers to that field. From this point of view, our system is good, because the students graduate, the postdocs get teaching jobs, and then we no longer employ new people for the old topic, but we strengthen our capacities in the new one.

In the future, what applications will your models have? Which topic occupies your thoughts the most these days?

All three topics I mentioned above occupy me simultaneously. Through network research we can focus on many, many things. There are researchers who study networks in space. This might in fact be closest to geopolitics, but one could study the world wide web, social networks, linguistic networks and so on. The area is inexhaustibly rich. We have taken on these three at the moment, In the future the next big topic potentially coming up is the brain, more precisely, comprehending the networks of the brain, as we can expect better and better data. Therefore, we will probably move in this direction. Which direction we can move into also depends on which fields yield data of such quality that are already useful for our tools, but the topic is not an old hoary chestnut yet.

What would you advise those young researchers who would like to study network theory in depth?

Quite a few students ask me this question. Then we always discuss what it is in particular that they are interested in, and then we try to find something for them in that smaller field. If you tell me that you wish to study biological systems at all costs, obviously, I will not give you a problem from the field of sociology. The other thing we need to consider is where the big questions are. There are two big question packets: on the one hand, in basic research we have a very good understanding of network structure, but not yet of network dynamics. We have no comprehensive theory about what happens in networks, and how the network process influences the structure of the network itself. There are many partial results, but there is no comprehensive theory. If someone feels that they wish to study fundamentals, and has a very strong background in mathematics, then this is the direction I would send him in. Here, a solid background in mathematics is indeed very important, as this is not an empirical question. Here, the student needs to come up with a theory. Simultaneously, there is great potential in the application of these tools in new fields.

In the coming 10-20 years, the largest potential field of applied research, as I have already mentioned, is the brain. For we have just come to the point that the brain must be mapped. The conceptual toolkit is being developed in the world, primarily in the United States and in Europe, using which we can map simpler organisms’ brains for the time being, but later on the human brain as well. Once these maps appear, an enormous set of questions will emerge, namely, how we can analyze these maps, how the brain works as a neural network. Here, applying the network theoretical toolkit will become unavoidable.

I am not saying that the network will be the solution to the understanding of the brain or the explanation to the questions of consciousness or memory.  Many kinds of breakthroughs will be needed, but without network theory it will not be possible to answer these questions. I would advise those thinking long-term to start out in this direction. In this field, there are those students who want to tackle such fundamental questions as what a network looks like, how it influences the brain processes that participate, and those students too, who want to tackle biological or medical questions, such as in what manner can we “fix” the brain, i.e. how we can cure the brain’s diseases.

The brain problem is doubly a question of networks. Partially, the molecule networks within brain cells collapse, when we say that there is a lack of dopamine, but as a result of this, the neuron network itself may collapse. These two questions overlap. For instance, Alzheimer’s and Parkinson’s are both collapses of a particular neural network. A certain part of the brain quite simply dies. In another case, certain molecules form, and therefore the network cannot run. This twofold network question will take a very long time to untangle and for us to understand in some way.  This will keep several generations engaged.

To what degree are brain research and research on artificial intelligence connected?

These are two very different worlds. The issue of artificial intelligence is at present in the hands of programmers. It might be inspiring for us with regard to the functioning of the brain, but it is not obvious that artificial intelligence is going to be solved on the basis of the brain’s operation.  The hardware is namely very different. One possibility is that we build a software on it that simulates on the hardware what our brain does, the other possibility is to create a software that suits the hardware better.

In my opinion, the winning software is likely to be one that maximally exploits the hardware’s capacity, and not one that emulates human thinking. This is only a hope, that the difference between human thinking and machine thinking will always remain and will always remain qualitative. Because the strength of machine intelligence is not to reproduce human thinking in its entirety, but to complement it.

The toolkit of network research so far has not really been deployed for the examination of geopolitical or international relations issues. What could the reason for this be?

I am not sure whether it truly has not been deployed or whether we merely do not call it network research. In 2007 we published a long article in Science, in which we studied import-export networks. This study not only yielded a book, but also an entire movement. We examined which countries export what, and out of this we constructed a network. Through this we succeeded in posing such questions as what territories may become reachable for a given country in the future. Let me take an example: if a country is very good in exporting bananas, this means that it has the appropriate plantations, climate, that there are agricultural experts and know-how on packaging and shipping. All of this may be really easily used towards, say, fig cultivation. Thus, the country may easily switch to fig exports, if that happens to be a more profitable branch. However, all of this infrastructure is of no use if the country wants to switch to iPad production. However, if someone has the technology to produce laptops, that may not be great help in banana cultivation, however, he may easily produce iPads. Thus, from this point of view what the spectrum of possibilities is lies in the hands of the country. It is possible to say very precisely which branches of industry are achievable for them at present.

Banana cultivating countries need a giant effort spanning several decades to come into a position where they are able to manufacture iPads, as they need to acquire new knowledge, new technologies, and train engineers and experts. It is not realistic when a banana cultivating country, which currently possesses no experience in electronics at all, declares that it would like to become an electronic powerhouse in the next 20-30 years. Thus, networks have succeeded in measuring the economic potential of countries. This tool is now widely used, for instance by the World Bank. If a country applies for development money for a certain branch of industry, they then examine with the aid of networks whether this switch is realistic. Then they either reject the proposal or they help them in developing a realistic portfolio based on their current resources.  A specific example occurred for instance when Pakistan turned to the World Bank with a great new development, and was advised to go in a different direction on the basis of our map, for what they originally planned was simply not realistic. An ex-student of mine, who is now a professor at MIT, César Hidalgo, developed this toolkit, when he was still researching in my lab with Ricardo Hausman.

In the meantime, they have further developed this topic. Starting from the World Bank, they advise other countries in this regard, and they help map the given country’s opportunities. This comes to the forefront in geopolitical areas as well. After all, who may manufacture or produce which good is very much determined by its environment. There are very clear clusters in this regard. For instance, the Southeast Asian cluster is very strong in computers. This is by no means a product of chance. They learned from one another, similar toolkits and knowledge is available to them, etc. As a result, capacities form. Thus, this is very much a question of geopolitics.

Do you have any connections to security policy?

We do not, but there are many who study this topic. Security policy has very many different aspects. One that is interesting to network theory is this whole war on terror. These days the war on terror has been transferred to a network theory basis. After all, in today’s conflicts it is not two great armies facing one another, but large armies face little groups who organize in networks.

In the United States, a new doctrine appeared, accordingly, which is called net-war. In the next 30-50 years, all of America’s wars will mean fighting against smaller groups, not a real, regular army.

In these cases, fighter planes and tanks will not be as useful. Understanding communities, mapping emotions, learning who is friend and enemy within a village will be much more important, because the civilians and militarily active population has become fully intertwined and indistinguishable from one another. We must figure out who is our friend, and who is our enemy. This is a very important paradigm shift all over the world.

Is your model used in order to obtain information from Facebook and Twitter profiles and activities for security policy purposes?

These kinds of data are public; therefore, they are used ad nauseam for all sorts of purposes. This really belongs to the category of data processing. I am a founding member of a firm called Maven7 here in Hungary. It’s a very successful firm, we have clients all over the world.

We do Twitter data processing, social network site data processing. We look at what is relevant for us from the mass of data that is available. Today this has become a separate profession. It is not routine in the sense that network-type thinking is not necessarily the heart and soul of every tool, but in more and more cases, it is. There is a constant tension between data mining and network thinking. However, network data mining tools are beginning to appear, which have absorbed network thinking. For instance, in our Boston institute we recently hired a lady whose specialty was network-based data mining in particular.

Many people see the future in the disappearance of borders. What is your opinion, will the role played by borders shift?

Borders have been eroded systematically for a long time. There are certain types of people, like myself, for whom borders really no longer play a role. This process is unstoppable in my view. Obviously, this is in stark opposition to national identity and community awareness.  The question is how these local interests can be connected to these “borderless countries”. If we consider Hungary, we have interesting double interests: we would like our border to disappear so that Hungarian communities abroad could become part of the home country, meaning that there would be no difference between an ethnic Hungarian from Transylvania and a Hungarian from Hungary. At the same time, we would prefer that our borders not disappear, so that we could hold unto a certain national identity. This is a Gordian knot and there is no magic bullet, such as network theory, which could provide a solution. Network theory is an appropriate tool to map these various points of view, to make us understand the issue, and the effects and availabilities of potential solutions. In the hands of decision makers, it could perhaps operate as a tool, so that they can figure out what the next move is.

Is network model appropriate for international relations, or in economic interactions and their analysis?

At present, we do no such research, but our students constantly study these areas, here in Budapest as well as in Boston. I have an annual network theory course, and in it students must prepare a project working in pairs. Each year there are a few groups who make a network about relations amongst countries, then they analyze it. Last year a group examined how countries connect to one another through their minorities. For instance, there is a Hungarian minority in Romania, thus these two countries can be connected, you can build a network and then analyze it. Of course, we cannot yet call this research. It’s more like a finger exercise, but the fact that the students can do it with such ease in a four-week course shows that this is a feasible option.

I do not follow this literature much, but today ever more details are becoming available in relation to international relations, therefore the area would be relatively easy to examine.

Would you say a few words about your new book which is about to be launched (Network Sciences)?

This is a textbook that I wrote over the past five years and I taught it in Boston as well as in Budapest, at the Central European University. As the book was being written, the chapters of the English version were published on the net. Still, interestingly, the first published book is going to be in Hungarian. Although this is of course a translation from English, but the English version will only come out in June-July. This course, I find, is actually made up of two very different communities. On the one hand, there is the Budapest community, which is made up primarily of economists, sociologists, political science students (though at times medical student also attend), whose mathematical skills are very different, therefore I show them the mathematical equations and we have a conversation about them when I teach, but I don’t expect them to approach the problems from a mathematical point of view. There is a software packet which aids them in applying their knowledge to real networks. On the other hand, I teach the same course in Boston to Ph.D. students with a solid background in Physics, Computer Science, and Mathematics, for whom the mathematical derivation is also very important. So, the question was this: how to write a book that satisfies both types of audiences? I structured the ten chapters of the book in such a way that the chapters contain formulas, but there are no proofs or mathematical derivations. I say what the formulas are good for, how they can be applied, I illustrate them with concrete examples. Each chapter has a section at the end for advanced students, which contains the derivations as well. Obviously, I wrote the textbook primarily for those who found Linked exciting, but found it contained insufficient mathematical content. This book has another interesting aspect, namely the question came up, why should it only be accessible in English? Obviously, it is possible to have it translated the classic way: the publishing house in each country hires a translator, then publishes the book. However, it is a rare instance in the scientific world to translate such textbooks, as those students who navigate in this direction typically try to read the book in English, anyway. So, by having a Hungarian translation, one and a half years ago, we started an interesting experiment. I kept the digital rights to the book entirely to myself. Thus, I provide opportunity for communities to assemble, translate the book together, and make the book available on the Internet. To help the process, we provide the entire software packet, so that the final book looks exactly the same as the original English, so that the images integrate nicely, the entire layout is beautiful and so forth.

We started experimenting with this in Hungary, and we learned a great deal from it, so we take this experience along everywhere in the world. Along with the Hungarian translation, a Chinese translation is underway (true, this happens through an official publishing house). However, we have also heard from Japanese, South American, German, and Italian groups, who assembled such communities and would like to make the book available. This is very exciting for me, as it is a new model. What is important to me is that the knowledge, the toolkit becomes available. We are trying to spread this model worldwide, eliminating the obstacles that copyright used to create earlier.

You studied in Bucharest, Budapest, and Boston, and at present you teach at Harvard and Northeastern University.  What is the most important difference between the education and scientific life in Hungary and in the United States?

The differences appear on many levels. The educational philosophies are fundamentally different in the American and Hungarian systems. My children attend school in Budapest, so I have insight into the Hungarian school system. One difference is that the European education system is knowledge-based. The American system however is thought-based. In the United State, we do not transmit knowledge, rather, we instruct them in how to think. It is for this reason that European students in secondary education and indeed up to the first and second years of college seem to know a great deal more than Americans. But in the long run this rate changes, as the role of universities and education is not necessarily to put a certain amount of knowledge into our hands. After all, what really remains of school: we can write, read and count. Perhaps we remember that there was a battle someplace. Everything else we forget; all fact-based knowledge is gone. We can look it up if we really need it. What is, however, crucial for us is the way we see a question that pops up, where we look for answers, and how we approach problems. In the long run, or at least in the world of research, and in other areas of life this active system that incentivizes thinking is somewhat more successful.

How would it be possible to “copy” this a little in Hungary? We must learn out of this in some way or another. I am not saying that we must implement something like this already on the elementary school level, but at universities drastic changes are necessary. Having gotten to know the American system, I was astonished at how simple the Hungarian-Romanian system really was: the whole year through we goofed off, then we prepared really hard for the year-end exam. In other words: in two weeks, we absorbed the entire material, then we were examined, and afterwards we straightaway forgot it all. In the United States, things do not work this way. There, grades represent an entire year’s activities, and a student cannot afford to only prepare for the year-end exam. For the educator, this means a great deal more work. Having constant expectations where we keep testing the student, however, means much better learning opportunities for students. When I teach, the grade is not based on students’ examinations. Rather, the students typically map a project, they create a network, which had not been mapped before, they figure out which network this should be, then they demonstrate on this network what they learned in the course, and how well they can apply the tools. I certainly don’t sit down to examine and say hey, you, now go ahead and tell me about the basics of the Erdős-Rényi model. This they can look up in a book. But do they understand, how to analyze a network? Then they have learned everything necessary. If we managed to make this shift, this could be very successful. There are other aspects of this same issue. In Hungary schools, even schools in Budapest, the elite schools, struggle with tremendous budget deficits. I don’t even dare to think what happens in rural areas, in villages. The same is true of universities as well.

Interestingly, when we take a look at what the Hungarian nations is proud of, it’s our Nobel Laureates and Puskás. That student of mine who studied export-import questions, César Hidalgo, also created another tool, in which he took a look at the cultural exports of specific countries: in other words, what other countries are familiar with. What is the cultural export of Hungary? Music, science, and soccer. This is where we have excelled. The question is whether we can maintain this.

If you were to decide in what scientific areas the Hungarian Academy of Sciences should finance basic research, which ones would you choose? In your opinion, in which scientific area could Hungary become world-famous in 10-15 years?

There are areas where we have no solid traditions, and there are others which continue to be important, where we must continue to invest, and where we must pay attention that the current momentum is preserved. Such a field is mathematics, or brain research in Hungary in the last few decades.

Perhaps one of the questions ought to be where have we got solid traditions, and within that scientific which specializations could become potentially even more important in the long term. I have not studied science in such depth to be able to answer this question, but the role of a leader or a thinker is really not to say, hey, here’s where we need to invest, but rather he or she needs to approach the question properly. This proper approach is to ask where the necessary gray matter is, upon which we can build and develop. So, for instance it would be pretty hopeless for Hungary to enter the race in particle physics, as we shall never have a particle accelerator.

This does not mean that we should not participate in CERN’s work, where Hungarians indeed perform very well. Rather, I am saying that we have certain abilities upon which we can build, and upon which it is worth building. I just randomly highlighted two fields, mathematics and brain research, and of course I would add network theory, too, where there is a lot of overlap with mathematics in Hungary, and in the long term there will also be a great deal of overlap with brain research, too. Certainly, there are numerous other fields with great potential that I have no knowledge of.

Do you plan to write a book in Hungarian at some point?

This is a question of markets. I just sent off a book proposal for my newest popular science book, and it will also be written in English.

I do not write in English because I find it easier, or because in some way I have greater affinity towards the English language. I would love to write in Hungarian, but the market is elsewhere. With regards to everything that I do, my goal is that it become available in as many platforms as possible.

If I write the book in English, it can be very easily translated into very many languages, including Hungarian, so the Hungarian community is not excluded. If we think about it, Bursts is actually a very, very Hungarian book. But it was written in English. Its theme however is very strongly Hungarian, and it was a great deal more successful in Hungary than in any other part of the world, because in other parts of the world people could not identify with its story, whereas in Hungary the entire public identified with it.