[Editor’s note: This is part of a series examining the internet’s first 50 years and predicting the next half century. Join Xconomy and World Frontiers Forum on July 16 for Net@50, an event exploring the internet’s past and future.]
No one has done more than Donald Norman to teach us that every piece of hardware and software—in fact, every human-made object, including the internet—embodies design decisions that aren’t always made with the best interests of users in mind.
On top of writing the best-selling 1988 book The Design of Everyday Things, which popularized the concept of human-centered design, Norman has spent decades working inside universities and companies (including Apple (NASDAQ: AAPL) and Hewlett-Packard (NYSE: HPQ)) to remind his colleagues that technology should accommodate human needs, capabilities, and behaviors, rather than the other way around.
He’s currently director of The Design Lab at the University of California, San Diego. I talked with Norman for my feature article “Special Report 2069: Predicting the Internet’s Next 50 Years.” An edited transcript of our conversation is reproduced below.
Xconomy: Back in 2011, in a column for the website Core77 you said you were worried that in the absence of some fundamental innovation, the internet might evolve into a patchwork of walled gardens. I think that was prescient—today, more and more of the internet is behind paywalls, inside social networks, or quarantined by authoritarian nations. I know you’ve continued to think and write about the design and workings of the internet. What’s on your mind right now?
Donald Norman: I have many, many concerns about where we’re moving to and where we are today. ARPANET started off as a sort of informal communications network among remote facilities, in Massachusetts and California and maybe Utah. Nobody at that time predicted what was going to happen, and I think it may have been impossible to predict it.
So, the systems were kludged together. They were used by highly technical people, mostly by people who are friendly with one another, and they started sharing information. And then they started developing methods of doing this over long distances, and over the newly developed packet-switching networks, which eventually became ARPANET, and later combining multiple networks into the internet.
I remember in the days, in the early days, not only was there no security and no attempt to have any, but people bragged about that fact. At the MIT AI Lab, they bragged that if you connected to MIT, you entered the debugging state of their PDP-10 computer. And the argument was that anybody who can work the debugger is somebody we welcome to the community. But that’s the most dangerous place to be in a computer because you have complete control of what everybody does!
I recall when a student at UC San Diego really violated the trust. I forget even what the student had done, but it wasn’t acceptable, and there was a big discussion about, “Oh, what should we do, how should we punish the student?” And I said, “Look, I’ll just go talk to him.” And that’s what I did. And we talked, and I said, “This isn’t the way we do things.”
And so, that was the early attitude. But the problem is that all the fundamental infrastructure was built with this kind of trust and openness in mind. And it’s really hard, once you have this infrastructure and it gets in place to a large extent all across the world, to change it easily. And that’s, I think, one of the major issues because now we’ve let everyone in. And people discovered that they can steal. They can steal data, they can steal privacy, they can modify documents, you can spoof identities, you either hide your true identity or take on someone else’s. We can produce fake news.
And let me keep going. The other thing is that Google, Facebook, and others—when they started, they didn’t have a clear business model in mind. And they discovered advertising, which by itself is not a bad thing. But they discovered that advertisers said, not unreasonably, the more we know about people, the more we can provide information that’s relevant to their interests and needs, et cetera. But that has now carried on to an obscene level. And we, as individuals, no longer have any control over information about ourselves.
I was just at a conference on transportation, and one of the companies is proud of the fact that it’s introducing a new fare system for the MTA in New York City. So, their equipment allows you to take the New York City subways and buses and so on. And they were proud of the fact in this conference that, “We say we don’t own the data, the city owns the data.” And I said to myself, and then I said publicly to the conference, “Why should the city own the data? Isn’t it the individual’s data about what routes they are taking, and when and where they’re going, and who they’re with?”
And now, the psychological sciences have been mined to the utmost to create little tidbits of information that are so exciting that they become addictive, until the social networks have learned how to present little tidbits continually. We want to know what’s going on, and we don’t want to be left out. And this is, I think it’s much to the detriment of society, in decreased performance and decreased quality of the work that you can do when you’re continually interrupting yourself.
And then the horrible license agreements that we’re forced to accept even though we can’t read them, and can’t understand them even if we could read them. The argument is, no, you’re not forced. You could always say no. But no, there are no alternatives to the services. Elections are no longer to be trusted.
And the other one is that we’ve always had multiple networks and multiple ways of getting information. We had radio, which was different from TV, which was different from books, which was a different way of getting information, and moreover, we had different channels even within TV or radio. On television, there were basically three networks. So, that meant that everybody in the country got the same information, and that brought the country together. You had to listen to things that maybe you disagreed with. But today, [the internet] allows us to read only things we agree with. And if you only hear agreement, you never really use your mind, you never consider alternatives, you never understand other people’s points of view.
So yes, I am very concerned. Now, I haven’t told you anything in that whole long list, that diatribe, that others haven’t been saying. Which actually is a good thing, not a bad thing. Because it means that it isn’t just my private opinion, but it’s shared by many, many people.
X: Putting on a futurist hat, what are the most important forces likely to determine what the internet looks like 50 years from now?
DN: It’s really hard to predict. I suppose we could have imagined Facebook in the ARPANET days, but we would never have thought they would reach a billion people. That number would be inconceivable. The word “giga” didn’t exist. I mean, it existed in some exotic dictionary, but it’s not in anyone’s normal vocabulary. We talked about kilobytes; not about even megabytes, let alone gigabytes or terabytes. So, I don’t think it’s predictable.
My friend Herbert Simon, a Nobel laureate and all that, once made this wonderful statement which I love, which is, “It’s really easy to predict the future. People do it all the time. The hard part is getting it right.” And the problem is that we are talking about multiple orders of magnitude of changes in the technologies. It’s hard enough to predict what a 10 times change will do, but to try to predict a thousand times or a million times change?
I mean, neural networks were invented in the AI Lab at MIT. I remember we had cognitive science postdocs in the early days, and I remember I was once teaching them about perceptrons and why they were early computational devices, but they were very limited. And I was talking about the work of [Marvin] Minsky and [Seymour] Papert showing their limitations. And one of the new postdocs said I was wrong. And he stood up and took the chalk away from me and started showing some of the new work that was being done in England. And that was Geoffrey Hinton. And from that conversation he and then Dave Rumelhart, my colleague, developed what’s called the hidden layer in neural networks, which made them dramatically more powerful. Of course, Minsky and Papert never imagined that you could do this with their technology, so their analyses didn’t apply anymore.
Everybody liked neural networks for a while, but then they sort of died away because they weren’t very powerful. But today they are. Today, we have a thing called deep learning, which actually Geoff helped invent. And he now works for Google. I met him there and I said, “So, what was the great theoretical advance that made deep learning possible?” He said, “Nothing. The great advance was that in the last 20 years, computers have gotten a million times more powerful.” We never would have thought that we could get such huge databases and hold them and store them and then use them effectively in our computational devices. That wasn’t possible, and we could not have predicted that.
X: At what point, if ever, will we reach 100 percent internet penetration? Will it happen before 2069?
DN: Communication networks will soon hit the entire planet. I suspect it’s only people who deliberately do not wish to be connected or people who are living in remote, difficult places that won’t do it. But that isn’t the real issue. I think the real issue is, we may well find nations that will cut it off—nations that will close, that restrict themselves from interacting with others. We already see that happening.
And look how difficult it is in the United States to get onto the internet. And why is that? Well, because—and this is a completely different topic now—it’s our economic system and the way that our version of capitalism has worked. And it’s not just us, it’s basically Europe and the United States and maybe most of the developed nations of the world.
And we’ve gotten to the point where—again, I’m not going to say anything that’s new—but Milton Friedman is, in my opinion, the enemy. It’s not just him. But he codified it. He said that the business of a corporation is basically to make profit, and it owes its loyalty to the stockholders as opposed to the community in which it’s embedded, as opposed to its employees, as opposed to its customers. And what this has led to in the end has been the short-term emphasis on quarterly profits and the obscene rewards that senior executives in large companies are getting and a complete disregard to what it does to the environment, what it does to the community, what it does to your customers, and what it does to the people who work for you. And that’s evil and immoral in my opinion, and more and more people are starting to say that in business schools. Milton Friedman took the statement as law, but it is not a law, it’s an opinion. And now, starting to back off and say it differently. But this has governed the whole world, and now it’s very, very difficult to back out of the infrastructure that has resulted.
You can predict a horrible, dysfunctional society. Or you could predict an optimistic and altruistic society. And quite often, the difference between these two is the butterfly flapping their wings in Brazil. It could be very small events that tip you one way or another way.
When radio was first developed, it was supposed to be the democratizing new technology that allowed people themselves to broadcast to others. And it was all going to be university and public-spirited broadcasting stations. But when the commercial value started to be recognized, companies, in particular RCA, simply lobbied the US government and helped develop the Federal Communications Commission to develop radio as a restricted property. They changed the whole course [of broadcasting]. It didn’t have to be that way. A different president might have changed that, or a different group of people. As a result, all of our information sources have become commercialized to an extreme extent.
I think we have to start over. And we may need very separate networks. The notion that there is one network that is for everything maybe is wrong. We used to have separate networks. They were, in fact, determined by the technology. Hence radio was different than telephones, which was different than television, which was different than printed books. And today we say, “Nah, it’s all information,” and that this isn’t that neat, because on the internet you can do all of that. Well, OK. But the content is really what’s important, not the technology or the way it’s distributed. And I would love to find a different scheme where people are controlling their own data.
Right now, it’s really hard, in fact it is impossible for me to just subscribe to the internet without paying a fortune. Why is it so expensive? Because the people providing the internet are big monopolies, basically, and they say, “You can only get the internet if you also take these other services.” I don’t want any other services. I don’t need them anymore. But that isn’t good for their profits.
So, where am I going in all of this? Well, the truth is I don’t know where I’m going because, in fact, you’re asking me to predict the future 50 years from now. I’m trying to predict the future five years from now. I see where the issues are, and I’m in contact with large groups of people, in various kinds of digests and mailing lists and conversations, who are very concerned about these issues and are trying to move it in different ways. You know, looking at the morality of artificial intelligence, looking at the way that we use data, and trying to see how we can make it so the people own their own data and have a say, but without overwhelming them. Because if I have to say how each of my little pieces of data were used in every single instance, I’d be so bombarded with it, it would be like trying to read the horrible license agreements. I’d say, “The hell with it, just do it.”
X: It would be like those cookie disclosures on every website now, thanks to the European General Data Protection Regulation.
I want to remind you that when the automobile was invented, back in the early days when they were just starting to become popular, one of the great virtues of the automobile was it would reduce pollution in cities. The streets were all littered with horseshit. And that prediction turned out to be true. But nobody predicted the vast numbers of automobiles that would result, and the vast pollution of the air that would result. And, for that matter, the paving over of vast amounts of our territory to hold these automobiles, most of which are idle most of the time. And the fact that people suddenly became second-class citizens because the cars dominated the streets and not people. Why was that? It didn’t have to be.
So, I don’t want to predict the future of the internet. What I would like to predict is the future of our lives, and how we live them. There will be an infrastructure beneath it that helps and supports it. And whether that’s called the internet or whether we actually have all sorts of multiple services, that isn’t so important. What I want is to worry about what the goals are.
Let me take an example of a really positive thing that has resulted. The conversation you and I are now having, this video conference, couldn’t have been done a number of years ago, and it was not predicted, again. People have talked about video conferencing for a long time, and the phone companies tried it forever and ever and ever. And I had friends who work in the high-tech industry telling me, “Forget it, it is never going to happen. People don’t want to do it.” Well, here we are doing it. And not only is the sound quality really excellent, better than the telephone system provides, the image quality is really fantastically good. I can almost read the titles of the books in your bookcase.
The point is technology’s been wonderful. It has really enhanced our lives in many, many different ways. At dinnertime we’ve had to be careful because it’s so tempting when we’re having a discussion and we don’t remember something, to turn on the computer and get the answer. You get hooked. Actually, we do say, “Alexa, what is the whatever.” Well, these are good things. I don’t care what the underlying technology is. I’d be happy to have multiple technologies that do it or maybe different networks that do it—as long as I can avoid the problems with privacy we’re having today.
Private conversations are reasonably private now. Can somebody sneak in and listen? Yeah. But actually it’s not easy. And that’s the way privacy, in some sense, has always worked. Can somebody steal my car? Yeah, but it’s not easy. Can someone break into my house? Yeah, but it’s not easy. We put in enough hurdles so that only the professionals can do it, and not the average everyday person. The problem with the internet now is that you don’t have to be a real professional to break in. The tools are available to anybody.
So, how can we get the benefits without the downsides? I think it may require having separate networks [for different purposes]. I don’t know. But I do know there are many groups of technology experts thinking about what it’s going to take. Technology and policy and legal experts who are banding together to try to find answers.
X: Are these people mainly from academia, or NGOs and think tanks, or are they in startups, or big corporations, or all of the above?
DN: They’re from all of the places you mentioned. The ones from large companies are mostly from research, because as I’ve always commented, the research divisions of a company like Microsoft—and Apple in the day, Google, any company you can imagine—they interact more with academics than they interact with the product people in their own company.
So, I think that probably things like Skype and Zoom and the competition are good enough, they’re private enough. But there is a lot of metadata that’s being collected, and it’s possible for the companies to be recording the entire conversation. And that’s what I worry about. But those are not technology issues. Those are policy issues.
When it comes to the internet, the technology is the major problem because it was invented so many years ago. It’s time to upgrade it. Upgrading is really hard. It may not even be possible.
Another example of bad technology is the electronic medical record systems. And they are actually a major cause of the physician burnout. They increase the workload. But they also have benefits. The problem is that they were developed many years ago, decades ago, and they were primarily designed for insurance purposes, for billing purposes. And they use technologies and computer languages that are no longer really in fashion. But they are so huge that all the major companies I work with, executives of the major electronic medical record companies, they know the deficiencies, the problems in their systems. But they’re very, very difficult to change when you have this huge massive infrastructure. And the only solution may be to completely start over again.
X: Right, you just need to replace the software in every physicians’ office, every emergency department, every hospital back office.
DN: Exactly. As we like to say, “It’s just a simple matter of programming.” Well, I was at Apple, where for many, many years we tried to get a more modern operating system. We knew our operating system was crappy. It was originally written by amateurs, by young kids, and built on over and over and over for many years. There was so much of that at Apple we no longer even knew the source code of everything we were running because it got lost. And it was called “spaghetti code,” and we had really world-class computer scientists who just hated it, but it was really difficult to change. It took a long time to change.
And one of my favorite stories is how we took a part of the operating system that was flawed and buggy and slow and huge, and we said, “This one is the most disgraceful of all, so let’s change it.” And it was rewritten so that it was small and elegant and sophisticated and had no more bugs, no more errors in it. And we introduced it. And guess what? All sorts of people around the world complain that their programs were no longer running because the people who did their programs knew about all the errors and had devised all sorts of ways of getting around them. And now that we had a new, efficient system, it caused their programs to fail. And so we had to put back the flawed system, so that our customers could keep running their programs. And that just shows you the complexity of all of this.
X: It strikes me that these complex systems do get replaced over time, just not all at once. It’s like rebuilding a jet engine while the plane is in flight. Or it’s like the way all the cells in your body, except maybe some brain cells, get replaced over months, years, or decades. You are still you, but you consist of different matter. Is that the way the internet will get replaced by something better?
DN: My answer is very clear: yes, maybe, no. Take a look at the automobile industry. It certainly has evolved in a very interesting way. In the beginning it was three kinds of vehicles. They were powered by electricity, by steam, or by internal combustion engine. In all my readings of the history of technology of those three, the internal combustion engine was considered the least likely to succeed. And then there are all sorts of arguments about how it did succeed. I won’t repeat them, but in the early days, if you went for a drive you basically had to either be a mechanic or you took your mechanic with you. And quite often you had to stop or repair things, and you certainly had to change your tires frequently or fix flats.
Today, cars are sold without replacement tires. We don’t expect our car ever to fail. Most people don’t even think about that anymore, and cars have really, dramatically improved in many, many ways. And they’re also much more efficient. And we’re now switching to different things. Pretty soon, all cars will be hybrid or fully electric. Business models are changing, with ride sharing and with more and more automation, and someday autonomous driving. We won’t have fully autonomous vehicles, I claim, for 30 or 40 years, but I believe we will have them. But there is an example of what you’re talking about, how the whole infrastructure and business model is going to change. It will take a long time, but it will happen.
Now, in the computational arena, software is slightly more complex than anything else we have. But we are changing the software slowly. The operating systems of our computers are changing. We have new types of devices. I’m wearing one on my wrist right now and which is really pretty powerful, with lots of sensors built in.
Microsoft has dramatically changed the nature of their computer; we had tablets and cell phones and desktop computers, and Microsoft now combines them all into one, basically. And, in fact, I myself have switched to the Surface because I find that this way I can draw, and take notes, and I can make it into a tablet and read it when I want to read things. And yet it has all the power of a major computer. And it looks like Apple is about to move in that direction because they just announced a new operating system for the iPad which is no longer going to be just the same as in the iPhone. It’s moving us more directly where they someday won’t sell computers, except for professionals who have real need for huge computational demands. But for the average person it’ll be the iPad, which will have a keyboard and stylus.
And so, everything is changing, but it’s slow. But the problems that I talked about in the internet—the long list I gave you—if you think about it, all of them are policy issues, not technology issues. Policies that enhance companies’ profitability and also keep out the competition are harder to change. Although, companies don’t last that long. There is still one company that has lasted longer than any other, and even that one is almost dead now. And that’s General Electric.
X: It does seem like even small changes in the internet infrastructure take a long time. The transition to IPv6 addresses still isn’t finished, for example.
DN: IPv6 was introduced in 1994. First of all, it is a relatively small change. It was not just increasing the number of addresses, but it also changed some of the infrastructure. But look how long it took to be adopted. It still isn’t fully in use. And that’s kind of what I’m getting at. It’s not that easy even for what appeared to be a fairly simple problem—we’re running out of addresses. The problem is, we get workarounds. We now have subdomains. So, instead of a big company needing thousands of IP addresses, they just need one and then they have their own internal system.
X: So, will the next internet just be an iteration of the one we have, or will it work on completely different principles? I’m thinking of ideas like content-centric networking, where the emphasis is on what people are looking for, rather than where it’s stored.
DN: I don’t know. Let me point out your example of the human body, the example of the airplane, and even the IPv4 vs IPv6. Those are incremental and not radical changes. You’re going to change the jet engine in flight. You may get a far better jet engine, but it is still a jet engine, and the airplane is still the same, and the engine is still in the same place, it’s just that it is now better. And IPv6 is still the same internet, it’s just slightly better. And we have lots of stuff in our human body that nobody understands and we think is archaic, and it’s because of the way evolution works. It’s always trying out new things.
So, how do you ever get a completely different organism? You cannot get a completely different organism by incremental changes. And in the world of products where I talk about this more often—I even have a paper on this, Radical and Incremental Innovation, where I talk about how evolution is a hill-climbing device, really. It makes little changes, and if it’s better, hey, it keeps going, and if it’s worse, that organism dies out. And to do radical innovation, well, that has happened certainly in the evolution of animals and plants, but it has taken millions and millions of years. And often the biggest innovations come when some event happens that destroys half the life on the planet, and new life comes out. We have meteors striking or volcanoes or whatever. That’s not a good way to change. And so, a lot of what you’re talking about are incremental improvements, whereas what we need is a radical improvement.