Azeem Azhar: Research is often blamed for being a bit slow moving, (jokingly) “&ou know, I’ve been wondering about this topic. for 17 years.” Well, not in this case. And it was just over a month later that Moderna produced the first vials of its vaccine, 31 days later, after the sequence was initially released. And that is really, really remarkable., hundreds and 1000s of people.
Noshir Contractor: Welcome to this episode of Untangling the Web, a podcast of the Web Science Trust. I am Noshir Contractor and I will be your host today. On this podcast we bring in thought leaders to explore how the web is shaping society, and how society in turn is shaping the web.
Today,my guest is Azeem Azhar, an entrepreneur, investor and author. You just heard him talk about how exponential technology enabled rapid vaccine development and distribution. He’s the founder of Exponential View, a podcast and newsletter that explores the political economy of the exponential age, reaching an audience of more than 200,000 around the world. He’s also an active startup investor with investments in AI, work from home and climate change. He’s on the board of the Ada Lovelace Institute, and sits on the World Economic Forum’s Global Futures Council on Digital Economy and Society. Previously, he founded PeerIndex, a big data analytics firm acquired in 2015. His first book, “The Exponential Age: How Accelerating Technology is Transforming Business, Politics and Society,” was just published this month and was a featured book at this year’s ACM Web Science 2021 conference. Welcome, Azeem.
Azeem Azhar: Noshir, it’s wonderful to be with you.
Noshir Contractor: I’m so delighted to have you on the show, because you have been at the start of a lot of the things that will develop on the web. And I would love if you could start by talking to us about the role that you played — take us back to where things were at that time.
Azeem Azhar: Oh, I mean, it was just an absolutely amazing time. I first accessed the internet, in 1991, through a green screen terminal at university. It just opened my mind — the idea that I could talk to anyone anywhere in the world, and there was a real innocence and intent about how people spoke about issues. They shared a lot of material. There was no sense of there being ownership in a strange way. It was a real sort of commons of contribution. I graduated University completely unable to get a job — 53 job rejections, until eventually, The Guardian who had rejected me for several jobs, asked me to come in and help on a little event they were holding in an art gallery. And the event was a web event, and I showed up and the help they needed was setting up the modems to connect to the internet. And and that’s where we were in those day. It was very, very primordial, But if I think about what the Guardian in particular was willing to do and how they were willing to experiment in 1994, it’s pretty remarkable. I mean, it’s pretty far sighted to say, well, we should try and play around with the web.
Noshir Contractor: You mentioned in your book that you were amongst the first people to join social networking websites like 6 Degrees, like Friendster and MySpace. Tell us why you got interested in it. And what attracted you to those websites at those very early stages?
Azeem Azhar: Well, you know, I had already fallen down the hole that Tim Berners Lee, and then previous to him people like Jon Postel and John Licklider and then Leonard Kleinrock had created. And I was never going to climb out of that hole, as it were, there was too much to discover. And so you learned quite quickly, early on, if you’re an early internet user that the internet was really about people. So, when these first websites that allowed you to connect with each other, emerged, it was a really natural space, And of course, the challenges they had with this social network, was that computers were slow. And you wondered what the purpose of it was, because not everyone was on the internet. It wasn’t really your friends — it was just a bunch of people who happen to have discovered the service. But the power of being able to connect people together was really visible at that time.
Noshir Contractor: And we’ve come such a long way from those early forays. And you describe this journey as being an exponential change. Tell us a little bit about what you mean by that word, exponential in the context of exponential change and the emergence of the exponential age.
Azeem Azhar: An exponential change, you know, mathematically, is essentially any change of a constant proportion. So it’s compound interest. And what I define as an exponential technology is a technology that improves at a 10%, or higher rate every year for the same cost over many, many decades. And the consequence of that with those key technologies is that prices declined very, very rapidly.
As prices decline rapidly, elementary economics tells us that we’ll use more of this stuff. So as computing prices, courtesy of the exponential, decline in computing power dropped, we used much more computing — as we use much more computing, and I mean, billions 10s of billions, hundreds of billions of times, more computing.And because of that, what economics tells us is that complimentary businesses emerge. There are things that you couldn’t do with this technology that you now can do and businesses and services emerge on top of them. So from Moore’s law and silicon chips, we got cheap computers. From cheap computers, we got a web that could connect everybody.
But the thing that I found fascinating as I unpacked this question is the impact of this declining price, was that it’s not just that things got cheaper, we use them more frequently, we might use them in more areas. But that exponential reality transmitted up to products and services that are quite far removed from the underlying technology. So Facebook was the first product to reach 3 billion users. Many of us don’t think about Moore’s law when we use Facebook, but that’s why it got there. And we’ve just heard in the last few weeks, that TikTok is now the most downloaded app in the world, which didn’t even exist when I really started to think about the book. So this idea of exponential reality is that it weaves through from the kind of core technologies all the way through to the products that get built on them. And then the services and entrepreneurs and the market respond. So the technologies and the products demand very, very fast growth rates. And that requires rapid deployment of capital. And so this venture capital industry springs up around to fund these companies very, very quickly. And the thing feeds in on itself. So that’s what I mean by exponential technologies. And the exponential age is this notion that this pattern of accelerating change is becoming widely commonplace across our political economies. And I date that that inflection point at some point between 2011 and 2014.
Noshir Contractor: And then what do you mean by the exponential gap in this context, but as you point out, that exponential age comes with an exponential gap?
Azeem Azhar: The technologies and the businesses that are built on them, and the people who can take advantage of them, improve exponentially, and they create new potentials, and new potentials that we perhaps don’t have words for. But we as humans, live within societies that are regulated by, by habits, by norms, by conventions, by formal institutions, and by informal institutions. And largely, those institutions change incrementally at a linear pace. And so there is a gap that emerges, of the acceleration going upwards and this linear trajectory. And I think the exponential gap explains why we have a common pattern of a sense of friction, of division emerging about how we think of some of the fundaments of society in the political economy.
Noshir Contractor: What would be examples of our ability to try to address the exponential gap?
Azeem Azhar: I’ll give you, one example. If we look at companies. Traditionally, the way that we economists have thought about companies and regulators have thought about companies is that companies benefit from increasing returns to scale, and at some point, get to some diminishing marginal return. And that diminishing marginal return is like a force of gravity to hold a company to a certain size. The other force of gravity was that industrial inputs progressively got more expensive. The 1,000,000th kilogram of iron ore that you extract cost a lot more than the first kilogram of iron ore. And those things would slow down companies’ abilities to grow very, very big. Now, courtesy of essentially web based technologies and databases, we start to see companies being able to break that force of gravity, and they do so in two ways. The first is that a lot of companies now benefit from network effects. While the millionth kilogram of iron ore is more expensive to extract than the first, with a network effect business, the millionth customer adds value to all the previous 999,000. That’s a phone network. That’s Facebook, that’s Twitter. There are other types of network effects that emerge in this AI world that relate to our data network effects. So we increasingly rely on machine learning and algorithms to derive value in businesses. The data network effect means that the more people who publish web pages, the more people who search on Google and click or don’t click on results, the more information Google has about what good looks like and no competitive entrant to the market, however hard they try, can get that insight. And with every cycle, every click, every search we do, Google gets better. And barriers to its economic moat gets deeper and wider. And so those two things fundamentally change how we need to think about companies. In the 20th century, if a company had 70% market share, you can bet your bottom dollar the CEO had done something dodgy. They had bought up all of the silver, they had fiddled something with the regulators, they had done, they’d colluded with a competitor. In the exponential age, companies just get to 70% market share, because that’s where network effects take them.
And so then the question is not so much whether these companies are nefarious, or their bosses are good or bad. They may be, they may not be, it’s that the physics of exponential age companies is very different to the physics of an industrial age company. And that is the exponential gap.
Noshir Contractor: And this, of course, raises issues of ethical dilemmas that might come along with these rapid growths. And you founded in Peer Index, a big data analytics firm that was then acquired in 2015. And you talk in the book about how that experience in some ways shaped your thinking about the exponential age.
Azeem Azhar: There was a standard that sort of evolved in the early 2000s, called FOF, friend of a friend. And the idea was that you could use that standard as a way of keeping records of who you know, and what the nature of that relationship was. So there was some semantic depth to it. And I really fell in love with that idea. I built a FOF browser, in a blogging platform that I was running in 2003, 2002. And I had fallen in love with network science, and the fact that you could learn a lot about a group of people through their relationships without necessarily knowing who they were.
And by 2007, 2008, it was clear — Twitter had more than a million users, Facebook had more than 10 million — people were going to get addresses on the internet, they were going to be connected to other people.
And at the time, these networks were all open. And so I thought, wouldn’t it be really interesting if we could mine and interrogate and analyze and construct analytics in order to help people discover the richness of other people more easily.So the initial idea behind PeerIndex was to help answer questions like, tell me who knows something about sushi in Chicago. or help me find someone who knows something about shin splints in London. And by being able to look at the pattern of what people are posting on Facebook and Twitter and so on.But we could also then say to you, “ook, this is how you will be seen by systems.” And you can now look at the impact of what you say and do. And we could do that because Facebook and Twitter and these other networks were all open at the time.
Noshir Contractor: That sounds absolutely fantastic. What could possibly go wrong with it? Why are there troubling aspects? Because that sounds like an ability for us to globally know who knows who, who knows what, who knows who knows who knows what.
Azeem Azhar: It’s amazing and actually in this funny way, it’s the heart of the problem. The big issue I think ends up being around partly around consent. We used a model of implied consent, which is you can always make your Twitter feed private. And you can always ask us not to be indexed, but leave your things public.
And then, and then there’s the issue of the kinds of things that you can infer about people on the basis of their their behavior. We didn’t do this, but we could predict many, many types of personal classes and behaviors. And I think that that’s also also problematic. We battled with some of those questions. And in the end, the initial idea that we could provide this as a consumer product for consumers to use, didn’t really work out. And what worked out was it was a marketing analytics product that brands wanted to use to understand audiences. What was quite interesting about moving to the brands was, they didn’t care about individuals, they cared about averages and aggregates. So actually, all those problems went away. But it led to the next issue. Once you understand that you can affect people’s behavior. by tweaking aspects of an algorithm or showing them giving them a score, you actually have some kind of power over them. And that is not power to which they have consented to, or they have any way of challenging.
Noshir Contractor: One of the things you mentioned, as you described, the development of theory index was that at the time these platforms were open. You were able to get the data from there, even if you were implying consent on part of the users, it was still available. Since then, as you know, platforms like Facebook, don’t make that data available any longer. Why do you think that is? Do you think that they are trying to internally monetize the kind of peer index vision that you had?
Azeem Azhar: I think they do it for exactly that reason, which is that the data is in the core heartland of their network effect. So not only does it drive their monetization, because it’s the data that drives te the ad targeting. But the second issue is that, once you as a network, make your user data entirely visible, I don’t have to be part of the network in order to access your network. And so people forget that there was a product called friend feed, and friend feed aggregated Twitter and Facebook and a bunch of other things. So in a single panel, that was not run by these companies, you could look at all of your social networks in one place, you wouldn’t see the adverts because those were not in the content feeds. And you could message back into those networks. And that weakens the network effect, which is ultimately the source of these companies’ scale. The data policies of the networks were changing very, very rapidly, andthey were being tightened. I think the thing that was that would have been frustrating for me was that The honest reason for why they were being tightened, which was this is for our strategic long-term benefit from Facebook or from Twitter, was never the one that was presented, right, the one that was presented was, we want to provide users like you or I with a consistent user experience. And if you can access Facebook or Twitter from some third party application, they might not get a consistent user experience. So I think that the real argument, simply, it’s a business reason, “we wanted all — this is our pie.”
Noshir Contractor: One of the chapters in your book talks about the world being “spiky.” As you mentioned, this was obviously a play on Thomas Friedman’s 2005 “World is Flat.” And even before that William Gibson talked about the future being here, but it was not evenly distributed. How does your use of the word spiking build on or differentiate from those approaches?
Azeem Azhar: The key ideaof the world being being flat was this notion that there’s an equalizing force of around technology tied to a particular type of economic paradigm, that if people adhere to those rules, and things would be better for everyone. And, what I think has started to happen, and what we will see because of these, these technologies, is that, in fact, the local rather than the global, will end up being economically and socially more desirable in many, many contexts.
But there’s another part of it, which is, that if you randomly form a network, you get these nodes that have got more connections to them, you get agglomeration in a random network. But in a network where people are going to move for economic or emotional or cultural reasons, you are going to see even more agglomeration, because you’re going to see intent as to where people will go. And I think as the world moves to a more complex, advanced economic position, that kind of agglomeration will continue. So my view about the world is that while we will maintain global relationships, and we need to maintain a sense of global governance to certain types of problems — many relate to the web, many relate to things like climate change — we will also start to see increasing spikes emerge and some of the assumptions that were really of the the that neoliberal era, unpick.
Noshir Contractor: So Azeem, I wanted to take us to the present, while you were writing the book, the world was confronted with the pandemic. Clearly, there are two aspects of exponential change. On the one hand, the spread of the disease, but to me more interestingly, in terms of development of the vaccine, and then in the getting people vaccinated, also represent exponential change. I was particularly intrigued by your description of a particular website, Virological.
Azeem Azhar: Virological is a sort of GitHub for virus scientists. And very early in 2020, on the sixth of January, an Australian University virologist put a very simple statement on Virological — This is a website that typically gets a few dozen visitors a month. And he simply said, Look, the Shanghai Public Health Clinical Center is releasing a Coronavirus genome from a case of respiratory disease from the Wuhan outbreak. The sequence has been deposited on GenBank and will be released as soon as possible. Now GenBank is a code repository for sequences run by the National Institutes of Health, and people flocked to it.
Within a matter of days, hundreds of researchers are looking at this genome, because it’s new and it’s interesting. And we’ve not really got cases outside of China by this by this case. But what I found fascinating, is that research is often blamed for being a bit slow moving, you know, “I’ve been wondering about this chapter for 17 years,” well, not in this case. And it was just over a month later that Moderna produced the first vials of its vaccine. 31 days later, after the sequence was initially released. And that is really, really remarkable. What’s remarkable is not just that we could sequence the virus so easily. And that’s as a consequence of another exponential technology, which is genome sequencing. But then courtesy of the web, which is another exponential technology, we were able to get it out to, you know, hundreds and 1000s of people. And then the techniques that Moderna used, many of which relied on a machine learning based system to help manage data, discover data and look for patterns, were also applications of exponential technologies. And so you end up within 12 months of the virus being identified, we had seven different vaccines that had been approved, and 24 million people had received their first shot of the vaccine. A large part of just being able to do this and coordinate people to deliver and then receive the vaccine is entirely dependent on computers and databases and smartphones.
Noshir Contractor: One of the things you talk about in the exponential thesis is that there was a change, an exponential change both in the amount of things being invented, and the ways in which they get scaled. Tell us a little bit about the difference that you see between the exponential change in invention versus scaling up?
Azeem Azhar: There are larger markets to go after. And it’s cheaper to do this invention than it ever has been. One core idea that I talked about is the idea of combinations, the fact that technologies from different domains can combine and they’re reliant on there being open standards and modularity.
On the other hand, the question is, why can we then adopt them so much more quickly, and the reason we can adopt them so much more quickly. And I think this is where part of the thesis is a bit complex, right? It relates to the fact that there are global networks of information and global networks of, of distribution. And I think back to the first iPhone, which was launched in 2007. And it was available in one store in San Francisco, just off Union Square. And when the iPhone 12 was released, it was available in 300 cities around the world on the same day. And that is a testament to being able to coordinate and deliver these products over the place at the same time. And I think that that’s one of the interesting wrappings of the book and my argument, which is that the exponential age isn’t just about a process where silicon chips get faster and faster and faster. It’s that, that speed that acceleration has a way of echoing through other parts of industry, and then butting in quite quickly into our, the rest of our lives.
Noshir Contractor: That teases well for my closing question. We spent some time talking about the pace with which the exponential age is upon us. Will it ever stop?
Azeem Azhar: Well, I think in the timeframe that I’m thinking about in the of the book, which is, you know, decades, it will continue. I think we were scratching the surface, there are still incredible breakthroughs that are happening. And even things that happened while I was writing the book, I talk about in the book about a Romanian company called UiPath. And when I wrote the first draft, UI path was one of the fastest growing software companies in Europe and was had was worth a billion dollars. By the second draft, I’ve had to write that up to 7 billion, by the third, it was past 10. And just as we’re going to print, I had to quickly go in and change that number to $35 billion. So it’s a 35x increase in the valuation in the year or so that I went from first draft to to go into print. So I think it does continue. There’s a more metaphysical, I suppose question, maybe it’s a physics question, which is, can it continue forever? I mean, physicists will tell you that, ultimately, there are a limited number of atoms in the universe. And there is there’s sort of issues of their complexity and what can an atom really support. So I’m sure there could be some physical limit to all of this. But that is the subject of a book that will have to be written by somebody else.
Noshir Contractor: Very good then. But speaking of your book, I really enjoyed it. And I would recommend it very much — the title of the book, The Exponential Age, how accelerating technology is transforming business, politics and society. Azeem, thank you so much for taking time to talk with us today about the exponential change that we are witnessing and in particular, being able to tie in many cases to topics of interest to those who are following the web and in web science in particular. Thank you so much again.
Azeem Azhar: Thank you, Noshir. Really appreciate it.