As the 1990s began, the Internet had a few hundred thousand hosts and was used by a narrow community of academics and computer nerds. By the end of the decade, there were tens of millions of hosts, and in 2005 the Internet added its billionth user. Today, the Internet is used by almost everyone in the developed world, and is rapidly being adopted in the developing world.
This is probably the most spectacular (and literal) example of a network effect. Network effects occur any time the value of a network or system—per user—grows with the number of users. This phenomenon is easy to see in the case of the Internet. It would have been hard to build a profitable Internet startup with fewer than a million potential users. There would have been little point in starting a blog like this one in 1990 because the universe of possible readers was tiny. And, for that matter, it would have been hard to send an email to your grandmother in 1990 because she almost certainly wasn’t on it then. And this, in turn, meant that the Internet was a (relatively) uninteresting place for ordinary users. But as the number of Internet users grew by four orders of magnitude, the number of opportunities—articles to read, friends to chat with, useful services, etc.—grew accordingly.
It’s not a coincidence that the Internet has a bottom-up design. No one controls access to the Internet, and any node can (more or less) communicate with any other node. This is in contrast to competing networks of the late 1980s and early 1990s, such as AOL or France’s Minitel network, which kept a tighter leash on who could offer services on their networks. Network owners that tried to anticipate users’ needs and optimize their networks for those specific use cases were incapable of exploiting the full power of network effects, because they made assumptions that precluded uses (like the Web) that turned out to be extremely valuable in the long run.
The value of any social system flows from making the right connections between groups of people. Social systems help connect friends, co-workers, and significant others. They connect writers with readers, entrepreneurs with investors, businesses with customers, teachers with students, programmers with users, and so forth. Bottom-up systems are good at exploiting network effects because it’s hard to predict in advance which connections will create the most value. In top-down systems, decision-makers have to anticipate in advance how best to organize people. As the system gets larger, the decision-makers will do worse and worse at this, because the number of potential connections grows faster than the size of the group. In contrast, bottom-up systems scale gracefully, because each participant pursues potentially-valuable relationships independently.
Network effects are critical to the success of disruptive innovation. Disruptive technologies are cheap technologies, and cheapness accelerates innovation in two ways, illustrated by the success of the microcomputer. First, more users means more people like Dan Bricklin inventing software like the spreadsheet. Second, more users meant more potential customers for those innovations once they were invented. Hence, a virtuous circle: VisiCalc made the Apple II more valuable, which caused more people to buy it, which further increased the incentive to create innovative products like VisiCalc. Multiply that by hundreds of other examples, and you have a good explanation for why so many cheap, simple technologies so often triumph over more complex and expensive technology that, on paper, seem more sophisticated.
Serendipity is key to this process. Theoretically, DEC could have given Bricklin a PDP-11 in 1970 and paid him to write a spreadsheet for it. But DEC didn’t know Bricklin would be able to create a hit product—indeed, Bricklin himself probably couldn’t have predicted how popular his spreadsheet would be. So VisiCalc became possible only after Apple IIs became cheap enough that Bricklin (and thousands like him) could buy one for his own use.
In other words, disruptive technologies lower the costs of experimentation, which makes bottom-up organization possible where it wasn’t before. The cheaper experimentation is, the more feasible it is for small firms, or even individuals, to cover the costs out of pocket. That allows them to try experiments on their own dime that they wouldn’t be able to convince anyone else to finance. And more experimentation allows more effective exploitation of the opportunities created by network effects.
