I think we are on the point of seeing nonhuman intelligence, or, better put, nonbiological intelligence, in the twenty-first century, an intelligence that will probably be superior to human intelligence as we know it at present.
We have to learn to relate to more and more complex systems. I think we have to increase our ability to confront greater complexity yet further, to at least a few orders of magnitude.
—Michail Bletsas
Michail Bletsas is Director of Computing at the MIT Media Lab and Director of the Network Computing Systems Group, where he is responsible for designing, installing, and maintaining the infrastructure that the laboratory uses to produce, create, move, convert, store, and consume its “bits.”
He studied at the University of Thessaloniki, subsequently did postgraduate work in computer engineering at Boston University, worked for various wireless technology companies, and eventually joined the MIT Media Lab, where he helped create mesh networks technology (networks that do not need to deploy physical infrastructure in a zone to function).
Adolfo Plasencia:
Thank you for receiving me Michail, for this conversation.
Michail Bletsas:
My pleasure. Welcome to MIT Media Lab again!
A.P.:
Michail, you were born in Crete, a pivotal center of the Mediterranean. You studied at the University of Thessaloniki. Later you undertook postgraduate studies in computer engineering at Boston University. And later still, you arrived at the MIT Media Lab, founded by Nicholas Negroponte, also of Greek origin—whom you have worked with—and where you are now director of computing.
Is it by chance or has Greek vision got something to do with the Media Lab?
M.B.:
The fact that I ended up working at the Media Lab was something completely unrelated to my origin and Greek roots. I think I can safely say that my friendships, although they have a lot to do with who I am, did not play any significant part in my professional career. But I can say that Greeks fit easily into an atmosphere like that of the Media Lab. We don’t require much structure. We like a bit of chaos. We prefer it. And I believe that is what the atmosphere of Media Lab reflects.
A.P.:
I’m interested in your opinions as director of computing at the Media Lab. Yours is a symbolic location for people who think about advanced technology. Research undertaken here associates computation with intelligence in many ways. In the names of the courses and projects, terms such as “Intelligence,” “Thinking,” “Intelligent Computation,” “Things That Think,” or “Ambient Intelligence” appear. However, not everyone agrees with associating computers with intelligence or thinking. For example, Roger Penrose said at a conference in Barcelona, “Computers cannot be intelligent in any way because they are not conscious. Perhaps someday, in some laboratory, someone will construct a conscious artifact, but it will certainly not be a computer.”1
What do you think of Penrose’s opinion?
M.B.:
I don’t agree with him. I believe that human intelligence is something we have only recently begun to systematically analyze, and it’s a very complex system. I would agree much more with Marvin Minsky, who influenced me a lot at the Media Lab. He claimed that to actually construct intelligence is not so difficult. It can be constructed using a very simple set of basic primitives. It’s the massive connectivity among those primitives that gives rise to the complexity from which behavior stems. But it’s a system we have not gone deeply enough into. The study of the brain is something that has only begun to gather pace in the last twenty years. We have recently made great strides in our attempts to understand the codification of human intelligence. That acceleration is also happening in technology, although we often tend to underestimate the goals that we may achieve in the medium-term future. We get very excited about our short-term achievements and sometimes are so disappointed when things don’t go well that we completely lose track of what may happen in the long term.
I think we are on the point of seeing nonhuman intelligence, or, better put, nonbiological intelligence, in the twenty-first century, an intelligence that will probably be superior to human intelligence as we know it at present. We are part of evolution, no matter how you look at it. If past history is of any importance, the answer to whether it is going to emerge must be yes. People say that we are the most intelligent species on the planet and that we cannot improve any more. I tell them, Remember where we came from: remember the Neanderthals, remember Cro-Magnon people and the speed with which we have been able to advance through biological evolution. Now things are moving faster, and yes, I would call it non-carbon-based biological evolution. Biology will also evolve, not only intelligence. I believe that in the twenty-first century, we shall actually see advanced intelligence forms that are not Homo sapiens. This is the natural order of things. I don’t see evolution stopping with us. There is nothing less natural than thinking that everything ends here with us.
A.P.:
There are people who say that the general paradigm of informatics is, again, “turning a corner,” changing the direction of its evolution. I’m going to give some relevant examples:
In infrastructure, in the explosion of big data, in the server farms—by which Google with its complex software system Omega, and its new generation, Borg, which makes each data center with its hundreds of thousands of servers function together as if they were a single machine, just as Twitter does with its Mesos system. In computing: the “systematic” computer that repairs itself, which already exists at University College of London. In the online stores: iTunes and Google Play have sold billions of apps. In services: a single engineer at Facebook now provides support to a million users because the Internet has become gigantic, and so on.
Weren’t these things and computer technologies almost unimaginable a short time ago?
M.B.:
The Internet, and its giant computation system is already the most complex system that human beings have ever constructed. We have to learn to relate to more and more complex systems. I think we have to increase our ability to confront greater complexity yet further, to at least a few orders of magnitude. We are now beginning to do that with our attempts to decipher how the human brain functions, something we are having increasing success with. The most important thing is that many questions that we were unable to answer before now have concrete answers.
Today we have enormous volumes of data, and undertaking their analysis is very important and useful for us. I don’t think we should be put off tackling new challenges because of their complexity. The Internet has to be applied in many more contexts. The Internet of Things is now beginning, although we cannot yet say that it has been implemented.2 My documents are still not connected! We are learning to understand that huge complexity, and, more important, we are learning to respond to very complex questions by simply gathering and analyzing huge amounts of data (big data). Doing this in the past would have required very costly equipment and a great deal of effort. Today—no.
A.P.:
Steve Jobs, before he died, said the era of the personal computer was over, but not everyone agrees.3 Michael Dell and several well-off friends, among them several people from Microsoft, said that they disagreed with what Jobs said. Moreover, they wanted to refute that statement, wrong as far as they were concerned, by proving him wrong. They repurchased Dell Company, the former leading manufacturer of PCs, for $24,000 million. Intel also wished to prevent the end of the PC and considered the announcement, at least, premature.
What do you think? Is a slow death of information technology based on the PC taking place?
M.B.:
We have to separate questions of business from scientific questions. The important thing in business is not only what will happen but when it will happen. Michael Dell repurchased his great PC company. PCs continue to sell well, although their sales have reached a plateau. They have stagnated. The important thing to be aware of is that the profit margins of PCs are tiny now because it is a very competitive and mature market, whereas tablets are just arriving and so allow for a much greater profit.
Every time Apple sells a $400-plus iPad it earns a lot more money than Dell does when it sells a $1,000 portable computer. Michael Dell should be very careful when making predictions. He contradicted Jobs because there was a moment in the past when his company overtook Apple, but today Apple is the biggest and most profitable company in the market.
From a scientific perspective, we know that PCs are disappearing. Really, it’s not natural for us to interact with them using the keyboard and mouse. Computers will be everywhere, we will have screens and reproduction surfaces everywhere, and we will interact with them through gestures and voice commands. They will look at us and respond to our movements. So yes, I can tell you that PCs will disappear in the form that we know them today, but this is going to take quite a long time. And there is still a multi-billion-dollar business there. Apple perhaps can afford to say, I can abandon that market because I sell other products that are much more profitable, but that’s not the same as saying that Dell will no longer be able to sell PCs. Again, we have to separate issues of business from scientific arguments and predictions.
A.P.:
Nicholas Negroponte, with whom you have worked for many years, first launched the book Being Digital;4 years later, in 2002, he published an article in Wired titled “Being Wireless,” in which he compared the working of Wi-Fi to the action of frogs leaping from lily pad to lily pad on the surface of a pond (“lily pads and frogs” will transform the future of telecom).5 And after came something you had a lot to do with: the mesh networks or wireless networks that do not need infrastructure to be deployed. Then came the mobile explosion, the ubiquitous access to the Internet using mobile devices: smart phones and tablets. I talked about this article in another conversation in this book with Howard Rheingold, and he calls this type of communication “horizontal and lateral.”6
What do you call it? What do you think will be next in access technologies?
M.B.:
We have seen how wireless access technologies are converging. If we look at wireless in general, Wi-Fi and technologies using the Long-Term Evolution (LTE) standard are all very similar in their use of underlying technologies. One difference, however, is that LTE (4G and 5G) technologies use licensed spectrum and Wi-Fi uses unlicensed spectrum. There will be space for both, and we will see Wi-Fi continuing to grow. Over the years, telecommunications companies will shift toward Wi-Fi much more, even though they now see it as something they cannot use because of its unlicensed spectrum and their concern that they cannot guarantee quality of service. If you go back a few years, similar arguments were made against the use of packet switching and Internet Protocol–based communications when these started to replace traditional physical and virtual circuit switching. Nevertheless, unlicensed access is the best means available for deeper penetration. Wi-Fi is going to play a very important role as an access technology.
We have an ever greater number of devices that are wireless. Now hardly anyone uses just Ethernet without good Wi-Fi access. As access technologies converge, we see that, at the purely technological level, LTE and Wi-Fi utilize very similar radio technologies. The main difference is a decision of business policy as regards the markets and the use of licensed or unlicensed spectrum. I think we will increasingly use devices that connect online using Wi-Fi. In any case, most of our devices today are wireless, and we will see them prevail over wired devices. Today we have certain things that are wired, but now the tablets, mobiles, the electrical appliances in the home are increasingly wireless, and we can’t be creating cellular provider accounts for each one of them.
On the other hand, the value of having a network managed by a telecommunications company is something we cannot ignore. However, we will see part of that migrate toward the Wi-Fi side and Wi-Fi networks will obtain elements of the larger (wider area) networks. We will see a convergence between the two. The difference is that you will have control over the Wi-Fi part of the network and that, in turn, will be linked to the macro network, which today are two separate and mostly distinct parts. We are going to see all these parts converge, and most access will be wireless. Also, the first networks to be established in developing countries will directly be wireless, so bypassing the copper wire stage.
A.P:
You are a computation scientist but have ended up in computation applied to the Internet.
Do you think there is a before and after in computation sciences following the Internet explosion?
M.B.:
Here we can talk about two stages. As an information technology scientist, I could say that there is a “continuum,” as the explosion of the Internet is accelerating several technological trends. Obviously, if you are outside, you are not going to recognize the concepts such as those that existed before the explosion of the Internet. What we are seeing now are the effects and the growing number of applications that those concepts have had in our daily lives. I would like to point out here that, after the Internet boom, we are beginning to formulate a more participative Internet, and we are realizing the potential of the Internet, which was always supposed to be a bidirectional medium from the beginning. That is the essence of what Tim O’Reilly formulated with his definition of Web 2.0, which tells people that they can also produce bits, so that there is no need for them to only be passive consumers. That is the most important difference in itself.
A.P.:
The networks and the Internet are another of your fields of research. The sequence we mentioned before for hardware has a parallel in the evolution of the network: the Web (Tim Berners-Lee), Web 2.0 (Tim O’Reilly), the Social Internet (social networks such as Facebook and Twitter), the Internet of Things (more than half of Internet traffic is no longer between humans), the Internet of Everything. …
What would the definitive Web be like, if it were possible?
That interplanetary Internet that Vinton Cerf announced some time ago?
M.B.:
At the moment, we don’t have much interplanetary activity, although the interplanetary Internet exists and will increasingly be seen. However, the most important thing is how the finely granular Internet is going to reach Earth. By that I mean that there are still thousands of millions of humans who are not connected, so there’s a long way to go before we all are. This is one of my favorite problems and it is where I have undertaken most of my research.
A.P.:
The term is “finely granular Internet”?
M.B.:
It’s not just a term, connectivity everywhere. There are still many people without connectivity to the Internet because they cannot afford to pay for it. We have to work honestly on the commitment to Internet being a fundamental human right. And until we make that come true, I’m going to continue to work on achieving it. I’m going to leave the interplanetary question to Vint. Right now we have several specific ways of connecting to the Internet—increasingly those are wireless ways—but we are going to witness ever more widely reaching methods over time.
I don’t know exactly what they’ll be like, but we will be more and more directly connected as humans and, of course, everything else will also be connected to the network.
A.P.:
Let’s talk now about the evolution of the MIT Media Lab. After a glorious stage led by Nicholas Negroponte, in which you all brought to fruition the sentiment expressed by your friend, Alan Kay, “The best way to predict the future is to invent it,”7 your laboratory grew and was extended with an impressive, transparent, and light-filled second building, which is now fully operational.
What computation and connection technologies have you deployed in the Media Lab this time?
M.B.:
The most interesting thing for me is that when one builds one’s own laboratory, you notice more what goes on inside. The network of sensors that we have deployed and interconnected throughout the interior of the building allows us to know in real time where there is activity at each moment and so immerse in the laboratory. We can do all kinds of things: switch off the lights in places where there is no activity at that moment, so saving energy; control the air conditioning or refrigeration, for the same reason; tell people in which other place there is activity so that they go with the others and maintain their social activities. We can recognize and know who people are when they arrive, we can help and guide them to where they wish to go. If they want, their visit can be recorded, and we can send them it as a diary so they know whom they have spoken to, how much time they spent and with whom.
We don’t have many fixed signs: most are touch-screen monitors with which one can interact, with maps and interactive catalogues of Media Lab activities. The building is much more interactive than the old one. At a more mundane level we have a wireless network that provides connectivity throughout the building. We have point-to-point real-time video capabilities, which are used a lot because of the continuous flow of events that occur here.
A.P.:
You are one of the inventors of mesh technologies for wireless networks. I imagine there will be a lot of that in Media Lab. …
M.B.:
Yes, all the computation is ubiquitous and connected: coverage for smart phones, Wi-Fi, and so forth is very, very dense. That was a problem in some of the MIT buildings, but I believe we have solved it here quite well and to the point that we are providing the model for the others in terms of ubiquitous mobile connectivity and for places with a communication system and interactive screens with a lot of feedback.
A.P.:
Do you have your own “Internet of Things” in the building?
M.B.:
Yes, “Internet of Things” at the Media Lab means that all the thermostats are controlled, that the whole Media Lab is online with a whole range of flow sensors: a microphone network that does not record what you say but just records noise, so that we know where there is human activity and where not, always respecting privacy, obviously, at the same time. We also have little robots that go around acting a little like your “agent,” all connected to the network. Another important thing we have is telepresence, as there is a lot of collaboration with companies throughout the world. For them and for us, it is important to what we used to call videoconferences and what these companies call telepresence. You can place your avatar in any place in the building and let it interact with anyone.
A.P.:
At the MIT Media Lab, almost every day, new artifacts and technologies are being invented that have never been seen before. Your capacity to amaze seems limitless. How do you do it?
Does it still surprise you as much as in the past that young people come here every year from all over the world to invent things in your building?
M.B.:
It doesn’t surprise me at all. The lab’s recipe is pretty simple at its base: find a corner at the best technical university in the world, bring in brilliant people that don’t fit well in other places, give them freedom to pursue what they feel passionate about. Make them build things to demonstrate their ideas, and get feedback often from the lab’s sponsors. Make the place fun, and repeat, year after year.
A.P.:
Is innovation something natural here?
M.B.:
Yes, it’s what we do. When you bring so many intelligent people here and you give them creative freedom, that’s what happens; innovation takes place day-to-day. It’s the best thing about working here. Something comes up week after week.
A.P.:
Talk to me about the relationship between the MIT Media Lab and the companies that finance a huge amount of research and laboratories here. Hal Abelson says in another dialogue in this book that the companies that finance MIT and its laboratories cannot “redirect” research in line with their own company interests and, at times, the convincing arguments of the companies have to be resisted.8
What is the technological relationship like at the MIT Media Lab?
M.B.:
The Media Lab was a pioneer and was the first to go into this. Many companies came and became members of the laboratory because they were impressed with what they saw here and wanted to support it. They see the laboratory as a window to the future. We shouldn’t discredit models of directed research that seek to solve very specific problems that require directed research. It’s another way. Those models, of course, have their place at an institution like MIT. That research will always have a boost effect on undirected research, because specific problems have to be solved and so resources have to be made available and efforts focused on finding the solution. The problem is overcome by evolving. You need not only innovation but also evolution, incrementally improving concrete things. That is important.
The world does not advance only in quantum leaps, but at times you have to go to a specific place and pursue a specific goal intensively. Most often companies encourage this. Most often when a company finances you, it comes to you because it thinks you are more intelligent and capable than it is in certain things and it pays you to solve something specific. The companies, by definition, cannot tell you how the research must be done. That is good engineering, that is evolution. To be a good engineer, you have to know beforehand what you want to achieve. Here at the Media Lab we advance the other way around. We say, We are not going to go very deeply into this. We are going to try to create for you a window to the future and show you that there are ideas on what the future will be like. How do we get there, exactly? That’s another question, but we tell them that we are going to get there.
A.P.:
Here you have strong relationships with communications companies. The boom in social networks, the boom in smart phones. This is all around you here.
How do you relate with the large companies of ubiquitous communication?
M.B.:
Telephone companies are very hierarchical in a top-down fashion. Control flows “from above.” Innovation that comes from below is foreign to them. They don’t understand the peer-to-peer effect. In Usenet culture, there used to be an ethical criterion: do not disperse, do not deviate, do not waste time. Those companies do not understand that. They don’t understand “opening up,” and in the long term they are going to regret that. Without this bottom-up movement, there is no innovation, because that’s the direction that the best innovation takes.
A.P.:
What is the importance of the boom in social networks and of the competition between Google and Facebook for the Internet?
M.B.:
At this point in time, our lives would be poorer without Google than without Facebook. Google is useful in a much deeper sense than Facebook, because it satisfies our need to find relevant information for all aspects of our activity. Facebook, in contrast, although it also satisfies a deep need (communication), it does so in a relatively superficial manner.
A.P.:
Michail, many thanks for this dialogue.
M.B.:
Thanks to you too.