[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.]
It’s a good thing journalists, pundits, and consultants can’t be held liable for the predictions we get wrong and the outcomes we fail to anticipate. Because if we could, we’d all be bankrupt.
Back in 2014, on the occasion of the 25th anniversary of Tim Berners-Lee’s seminal paper proposing the World Wide Web, the Pew Research Center asked an array of technology experts to imagine the state of digital life in the year 2025. They predicted gains in education and peaceful political change, à la the Arab Spring. And they celebrated the power of global connectivity to make people smarter and more empowered. “I expect the miasma of myth and ignorance and conspiracy theory to recede to dark corners of the discourse of civilization, where nice people don’t go,” one veteran engineer ventured.
No one imagined—how could they?—that the Arab Spring would soon settle into the Arab Winter. Or that in 2016, data firms and political campaigns would illicitly obtain raw profile data of tens of millions of Facebook users and would use that data to target them with polarizing ads and posts—posts that, with the aid of misinformation posted by Russian-sponsored hackers, may have ultimately helped to swing the US electorate toward Donald Trump. Or that in 2019, cases of measles, an infection once virtually eradicated in this country, would reach a new 27-year high, as misinformed zealots pushed false claims about vaccines on social media.
The point is that it’s almost impossible to guess which technologies will be most important in the future, or how specific actors will use or abuse them. It turns out that the same platforms that help organizers stage democratic uprisings can be used to repress them. Fundamental design problems, such as the tendency of Facebook’s personalized news feed or YouTube’s video recommendation engine to amplify misleading and divisive messages, can act like unexploded ordnance, staying hidden until it’s too late.
As the saying goes: It’s easy to predict the future. The hard part is getting it right.
And yet here we are on the edge of another anniversary. On October 29, 2019, it will have been 50 years since engineers sent the first message over ARPANET, the US military-funded academic network that established the control and communications protocols behind today’s internet. At such moments, there’s an irresistible temptation to gaze both backward, asking how we got here, and forward, asking what the global network’s next half-century might be like.
We pundits know in our hearts that we can’t speak with much authority about how the internet will work, what its impact might be, or even whether it will still exist in the year 2069. But we understand that the internet and its expanding cloud of endpoints, especially our 2.5 billion smartphones, are the most important inventions of the last half-century, and that they will continue to alter the tenor of our lives. So, we spin our forecasts anyway—and may the gods of wisdom forgive us.
A Stubbornly Persistent Infrastructure
The first thing to be clear about is definitions. Here I’m not talking about “digital life” in all its ramifications, the way the Pew Research Center did. I’m talking about the internet: the set of interconnected computing devices that communicate using TCP/IP (Transmission Control Protocol/Internet Protocol).
That’s the feature that ties today’s internet back to ARPANET, where TCP/IP was developed and tested. Ingenious in its basic design, TCP/IP allowed any network with a “router” to exchange data packets with other networks, regardless of each network’s internal architecture—hence the term internet, for interconnected network. Routers concerned themselves only with sending packets of information through the network, hop by hop, and reassembling them at the right destination, at unique numerical addresses like 63.246.24.145. All of the real smarts in the network could stay in the computers at its edges.
Counting just four nodes at the outset in 1969—at Stanford University, the Los Angeles and Santa Barbara campuses of the University of California, and the University of Utah—ARPANET grew steadily, to about 40 nodes in 1973 and more than 200 by 1981. On top of this nascent network, engineers would introduce all sorts of new computing and networking protocols and languages, for things like email (SMTP), file sharing (FTP), local networking (Ethernet and the Spanning Tree Protocol (STP)), web pages (HTTP, HTML, and URLs), and domain names (DNS).
But the real acceleration came after 1985, when the National Science Foundation grafted several new “backbone” networks onto the existing network and opened them to public commercial use. By the time the original ARPANET was decommissioned in 1990, Berners-Lee at CERN had already written the first web browser, and the kindling was in place for the first dot-com boom, ignited by Netscape’s IPO in 1995.
Today’s internet has grown vastly in scope, connecting more than 3.9 billion people—51 percent of the global population—and 17 billion devices, according to data from the International Telecommunications Union and market research firm IOT Analytics. (The term “Internet of Things” cropped up when it became clear that the devices using the network would outnumber the people.) But all of this is built on the same remarkably robust technical foundation. Which leads to our first couple of predictions.
Assuming that civilization itself hasn’t been decimated by war, plague, or climate upheavals, then by 2069, everyone on the planet will have internet access. Or virtually everyone.
“We’ll never get all the way; it’s asymptotic,” says Bob Metcalfe, co-inventor of the Ethernet local-area networking standard that carries internet traffic to virtually every computer in every office. “It reached about halfway in just 50 years, and assuming a certain symmetry to the adoption curve, that would say [it’ll take] another 50 years to complete the cycle.”
And the internet of 2069 will still bear the visible imprint of our internet, just as North America’s telephone system still relies on a 10-digit dialing system rolled out in 1947. That’s partly because, in systems with so many stakeholders, even the simplest changes can take decades. Engineers anticipated a shortage of Internet Protocol addresses as early as 1996. Yet the ongoing transition from traditional “IPv4” addresses to longer “IPv6” addresses, expanding the possibilities by a factor of 2 to the 96th power, is expected to drag on for many more years.
But it’s also because the internet’s ruling idea—that you can make networks interoperable by abstracting different communications functions into “layers,” each layer serving the one above it—is so sound. “There is this underlying infrastructure that knows how to signal a bit on the wire, which is Layer 1,” explains Radia Perlman, a fellow at Dell EMC who literally wrote the book on networking protocols, a 1992 classic called Interconnections. “And then Layer 2 says, ‘How can I send the whole message to my neighbor?’ And Layer 3 finds the whole path. I think that basic premise is flexible enough that it’s unlikely the internet will ever change.”
At the same time, Perlman admits she’s surprised that elements such as STP—a key networking concept she invented at Digital Equipment Corporation in 1985 to prevent loops inside Ethernet networks, thus allowing them to grow arbitrarily large—are still in place today. “It was a hack that I thought would live for, like, six months,” she says. “I had no idea how long it would last.”
Forces of Change
After this point, forecasting gets trickier. Will the internet backbone of 2069 still be made up of a web of fiber optic networks crisscrossing the continents and oceans? Yes, but probably with the addition of satellite-based networks and ultra-broadband wireless coverage blanketing every population center.
Will there be an even larger cloud of automated server farms at the network’s edges, each sucking down massive amounts of electricity? Probably, but with luck we’ll find low-carbon ways of powering them.
Will people in 2069 connect to the internet via exotic new devices, like augmented-reality and virtual-reality glasses? Sure, why not. Or maybe by then everyone will have moved on to neural implants.
The truth is that we can’t describe the internet of 50 years hence in precise detail, any more than ARPANET’s builders could have imagined Grumpy Cat or Pokémon Go. It’s probably more instructive to look at the major exogenous forces that could shape the internet’s growth, and then try to imagine scenarios in which those forces diminish, or become overpowering, or end up somewhere in between.
One of the forces is plain old innovation. The internet reaches billions of devices and provides fixed broadband speeds above 100 megabits per second in many countries thanks to Moore’s Law, the pattern under which available computing power has doubled every couple of years, while dropping in cost. The semiconductor industry has managed to keep up that pace of improvement for more than 50 years, but some experts believe we’re approaching the end of the line: the point at which we’ll run out of tricks for making transistors smaller and packing more processors into each device.
If Moore’s Law fizzles out, that could put a limit on the number of devices connected to the internet (since those devices won’t be getting continuously cheaper) and on the speed at which they exchange data (since bandwidth is partly a function of router processing speed). But it’s possible that some new, hard-to-anticipate event, such as an advance in quantum computing, will keep Moore’s Law going—or even accelerate it.
Another force is the eternal dance of competition between technology corporations, the monopoly power that usually afflicts markets when one company gets too far ahead of its competitors, and regulation by governments. Monopolies rise through innovation but always seem to end up impeding it. Arguably, it was the federal government’s antitrust cases against computing-industry giant IBM (NYSE: [[ticker:IBM]]) and telecommunications monopoly AT&T (NYSE: [[ticker:T]]) in the 1980s that cleared the way for the explosions in personal computing and internet commerce in the 1990s.
But then we entered a long era of listless antitrust enforcement, and now Google (NASDAQ: [[ticker:GOOGL]]) has a near-monopoly in the search business, Amazon (NASDAQ: [[ticker:AMZN]]) dominates e-commerce, and Facebook (NASDAQ: [[ticker:FB]]) defines social networking. These companies could grow even more monolithic, powerful, and exploitative, perhaps abetted by further advances in artificial intelligence. Or, with a swing in the political breeze—say, toward a President Elizabeth Warren—they, too, could be reined in or swept away.
Corporations usually aren’t out to screw us—not consciously, anyway—but a lot of bad actors are. Cybercrime in the form of botnets, darknets, phishers, and fraudsters, not to mention state-sponsored hackers, will be yet another major force shaping the internet. It’s not surprising that scammers have colonized the internet, just as they have every other medium. The problem is that
