Any discussion of monopolization of the web is bound to include the term "network effects," and its constant companion, "natural monopolies." This econojargon is certainly relevant to the discussion, but really needs the oft-MIA idea of "switching costs."

A technology has "network effects" when its value grows as its users increase, attracting more users, making it more valuable, attracting more users.


· · Web · 2 · 9 · 8

The classic example is the fax machine: one fax is useless, two is better, but when *everyone* has a fax, you need one too.

Social media and messaging obviously benefit significantly from network effects: if all your friends are on Facebook (or if it's where your kid's Little League games are organized, or how your work colleagues plan fun activities), you'll feel enormous pressure to join.


Indeed, in these days of Facebook's cratering reputation, it's common to hear people say, "I'm only on FB because my friends are there," and then your friends say, "I'm only there because *you* are there."

It's a form of mutual hostage-taking.

That hostage situation illustrates (yet) another economic idea: "collective action problems."


There are lots of alternatives to Facebook, but unless you can convince everyone on Facebook to pick one and move en masse, you'll just end up with yet another social account.

This combination of network effects and collective action problems leads some apologists for tech concentration to call the whole thing a "natural monopoly" - a system that tends to be dominated by a single company, no matter how hard we try.


Railroads are canonical "natural monopolies." Between the costs of labor and capital and the difficulty in securing pencil-straight rights-of-way across long distances, it's hard to make the case for running a second set of parallel tracks for a competing company's engines.

Other examples of natural monopolies include cable and telephone systems, water and gas systems, sewer systems, public roads, and electric grids.


Not coincidentally, these are often operated as public utilities, to keep natural monopolies from being abused by greedy jerks.

But the internet isn't a railroad. Digital is different, because computers are *universal* in a way that railroads aren't - *all* computers can run *all* programs that can be expressed in symbolic logic, and that means we can almost always connect new systems to existing ones.


Open up a doc in your favorite word processor and choose "Save As..." and just stare in awe and wonder at all the different file-formats you can read and write with a single program. Some of those formats are standardized, while others are proprietary and/or obsolete.

It's easier to implement support for a standard, documented format, but even proprietary formats pose only a small challenge relative to the challenge presented by, say, railroads.


Throw some reverse-engineering and experimentation at a format like MS DOC and you can make Apple Pages, which reads and writes MS's formats (which were standardized shortly after Pages' release, that is, after the proprietary advantage of the format was annihilated).


This is not to dismiss the ingenuity of the Apple engineers who reversed Microsoft's hairball of a file-format, but rather, to stress how much harder their lives would have been if they were dealing with railroads instead of word-processors.

During Australia's colonization, every state had its own governance and its own would-be rail-barons.


Each state laid its own gauge of rail-track, producing the "multi-gauge muddle" - which is why, 150+ years later, you can't get a train from one end of Oz to the other.

Hundreds of designs for interoperable rolling stock have been tried, but it's proven impossible to make a reliable car that retracts one set of wheels and drops a different one.

The solution to the middle-gauge muddle? Tear up and re-lay thousands of kilometers of track.


Contrast that with the Windows users who discovered that Pages would read and write the thousands of documents they'd authored and had to exchange with colleagues: if they heeded the advice of the Apple Switch ads, they could buy a Mac, move their files over, and voila!


Which brings me to switching costs. The thing that make natural monopolies out of digital goods and services are high switching costs, including the collective action problem of convincing everyone to quit Facebook or start using a different word-processor.

These switching costs aren't naturally occurring: they are deliberately introduced by dominant firms that want to keep their users locked in.


Microsoft used file format obfuscation and dirty tricks (like making a shoddy Mac Office suite that only offered partial compatibility with Windows Word files) to keep the switching costs high.

By reverse-engineering and reimplementing Word support, Apple obliterated those switching costs - and with them, the collective action problem that created Word's natural monopoly.


Once Pages was a thing, you didn't have to convince your friends to switch to a Mac at the same time as you in order to continue collaborating with them.

Once you get an email-to-fax program, you can discard your fax machine without convincing everyone else to do the same.


Interoperability generally lowers switching costs. But *adversarial* interoperability - making something new that connects to something that already exists, without its manufacturer's consent - specifically lowers *deliberate* switching costs.

Adversarial interoperability (or "competitive compatibility," AKA "comcom") is part of the origin story of every dominant tech company today. But those same companies have gone to extraordinary lengths to extinguish it.


Just as a new company likes standardization when it's trying to attract customers who would otherwise be locked into a "ecosystem" of apps, service, and protocols, so too do new companies endorse reverse-engineering and comcom to "fix" proprietary tech.

But every pirate wants to be an admiral. Once companies attain dominance, they start adding proprietary extensions to the standard and fighting comcom-based interoperability, decrying it as "hacking" or "theft of intellectual property."


In the decades since Microsoft, Apple, Google, and Facebook were upstarts, luring users away from the giants of their days, these same companies have labored to stretch copyright law, terms of service, trade secrecy, patents and other rules to ban the tactics they once used.

This has all but extinguished comcom as a commercial practice.


Today's comcom practitioners risk civil and criminal liability and struggle to get a sympathetic hearing from lawmakers or the press, who have generally forgotten that comcom was once a completely normal tactic.

The obliteration of comcom is why network effects produce such sturdy monopolies in tech - and there's nothing "natural" about those monopolies.


If you could leave Facebook but still exchange messages with your friends who hadn't wised up, there'd be no reason to stay.

In other words, the collective action problem that the prisoners of tech monopolies struggle with is the result of a deliberate strategy of imposing high technical and legal burdens to comcom, in order to impose insurmountable switching costs.


I wrote about this for Wired UK back in April, comparing the "switching costs" the USSR imposed on my grandmother when she fled to Canada in the 1940s to the low switching costs I endured when I emigrated from Canada to the UK to the USA:

Today, there's a group of tech monopoly hostages who are stuck behind their own digital iron curtain, thanks to Facebook's deliberate lock-in tactics: the users of Whatsapp, a messaging company that FB bought in 2014.


Show newer

@pluralistic Hal Varian (now at Google) and Carl Shapiro literally wrote the book on this in the 1990s, Information Rules:

About vendor lock-in as a strategy, and strategies (for both vendors and buyers) at strengthening or weakening the lock-in dynamic (respectively).

@pluralistic I'm all but convinced that every case of a "natural monopoly" can be tied to a network structure, either physical or conceptual.

More to the point, a network with either a central node or a least-cost path which is controllec by a specific entity.

Physical networks are transport (ground, rail, canals, sea, air), commerce, distribution (water, sewerage grid power, gas and oil pipelines, flight networks, shipping routes). Real estate (literally, control over nodes in a landscape). Extractive natural resources (control over high-yield nodes).

Virtual ones are relational, but also follow the same general structure: nodes, links, flows (or relations), and control points. Wholesale and retail distribution networks. Business contracts. Communications networks. Knowledge itself, which can be conceived of as a web. Education systems (access to and certification in knowledge or skill). Business conglomerates or chaebol are relational networks of control and distribution. Labour unions or guilds create a gateway around labour or skills resources. Patents and copyrights are government-instituted gateways, which through networks of owned properties and/or ownership give monopoly power.

Network structures create monopolies, monopolies are networks.

(I'm still looking for exceptions, I've not found any yet.)

Public ownership or control of the controlling node or least-cost path seems to be one of the few countermeasures.

#networks #monopoly #MonopoliesAreNetworks

@pluralistic Oh, and standards as well.

And softwar. Programs are networks of functions and/or methods interacting with data (streams, structures, inputs, outputs).

Sign in to participate in the conversation
La Quadrature du Net - Mastodon - Media Fédéré est une serveur Mastodon francophone, géré par La Quadrature du Net.