Discussion: (0 comments)
There are no comments available.
A public policy blog from AEI
The latest on technology policy from AEI, published daily.
More options: Share,
View related content: Technology and Innovation
Network effects, Metcalfe’s law, or demand-side economies of scale — whatever you choose to call it, this phenomenon fueled the explosive growth of our digital world. When someone else joins the network, my device or service becomes more valuable. And when the network becomes larger and more valuable, the incentive you have to join the network becomes greater, too. By linking nodes, devices, and people together, this upward spiral created today’s hyper-scale networks . . . and hyper-scale companies.
The measurable value of the internet for consumers is immense and so are the tougher-to-measure consumer surpluses — all those hidden free benefits that don’t appear on any ledger.
In the past few years, however, many began to worry about the downsides of hyper-scale entities. Nicholas Carr called it The Shallows. Tim Wu’s new book is The Curse of Bigness. Farhad Manjoo calls the Silicon Valley giants the “frightful five.” For most people, it’s just Big Tech. The bigger these companies get, the theory goes, the bigger yet they can and will get, until they swallow everything.
Surely, then, depending on your political vantage, it’s time for net neutrality regulation, stepped-up antitrust enforcement, strict privacy laws, online speech regulation, or even a new “technology tribunal,” no? Only government can check the unrivaled growth of hyper-scalable networks, right?
Well, maybe not. Facebook is finding that most people don’t want two billion friends but perhaps instead prefer smaller groups with private encrypted communication. If my experience and that of my friends is any indication, Amazon is having real problems controlling the product and fulfillment quality of its third-party sellers. And Google is finding it difficult to manage tens of thousands of smart but unruly employees with political opinions.
These are curses of bigness, but not necessarily the ones Wu contends. Network effects can work both ways. Larger entities are juicier targets for viruses or data breaches, for example. Remember during the PC-era how many viruses targeted the dominant Windows OS, versus the much smaller Apple Mac? Or how about fax machines, where network effects ordained that every business own one — until suddenly no one did?
I speculated years ago that Facebook’s size would, in some ways, become a liability and that people would look for smaller online groups and perhaps gravitate to more decentralized networks, such as Blockstack. In the last couple years, in fact, smaller sub-groups for neighborhoods, churches, and clubs have indeed become popular, and now Facebook is retooling its whole business based in part on this observation.
Most networks thus do not grow forever without complication, and we cannot simply assume hyper-scale businesses are infinite monopolies.
A new paper in the journal Nature Communications, for example, shows that networks are highly heterogeneous and mostly do not scale without limit. In fact, in “Scale-free networks are rare,” Anna Broido and Arron Clauset show that while many technological networks are “scale-free,” many social networks are not.
Many of today’s problems may thus arise from this mismatch between hyper-scalable information networks and far more nuanced and limited social networks.
Think about recent attempts by insular groups on college campuses to impose highly particular speech codes on all of their fellow students and professors. Something similar is now happening, on a far larger scale, on the social network and content platforms.
Twitter, in just one of many examples, has been suspending small-time users who tweet the snarky but innocuous jab, “learn to code,” at journalists. Twitter contends this is part of a “campaign of harassment.” And yet Twitter saw nothing wrong with — and in fact promoted — a fire hose of defamation by Hollywood celebrities and major news outlets, such as CNN and the Washington Post, against the students of Covington Catholic.
This is a mismatch between Twitter’s hyper-scalable network and the comical inability of the company’s clumsy algorithms and human employees to police speech in this hyper-social sphere. In the last sentence, I was tempted to write “out-of-touch and politically motivated human employees,” but of course from their point of view it is the users they are suspending who are out-of-touch and politically motivated. Which only highlights the problem: the inability to write universal speech codes for a hyper-diverse population on a hyper-scale social network.
When any small, sheltered group, whether inside a university or a technology firm, attempts to impose its speech code on the larger world, it does not compute. In many cases, it is merely that each side literally does not understand the other side’s jokes. That’s why we need alternative and nested communities within the larger platforms. To police their own, develop norms, and defuse anger and agitation that is only inflamed when top-down authorities get it wrong.
For dynamic, complex systems, the best top-down rules tend to be those which ban bad rules. States may not erect parochial barriers to interstate commerce, for example, and “Congress shall make no law…abridging the freedom of speech.”
The Constitution got this mostly right 230 years ago. In order to promote pluralism, diversity, and innovation across a new hyper-scale institution — the United States — it could not impose many specific top-down rules, which would inevitably seem obnoxious or even abominable to some region or faction. It would guarantee basic human rights and ban bad rules but otherwise allow social evolution. Today’s technology platforms would be wise to do the same. If, however, they try to regulate dynamic and nuanced human behavior, they will not only tie themselves in knots but also send an authoritarian signal and invite bad rules from Washington.
There are no comments available.
1789 Massachusetts Avenue, NW, Washington, DC 20036
© 2019 American Enterprise Institute