“The worst Zoom ever”, as Scott Smith said on Twitter.

I have a lot of time for Benedict Evans. Along with Ben Thompson of Stratechery, John Naughton, and the more critical writings of L.M.Sacasas, he is is one of the places I turn to when I’m trying to figure out what’s going on in the world of tech.

But when he says, as he did recently, that the problem with trying to regulate tech is that everything is happening so quickly, he’s just wrong. He’s using the wrong model of change.

New kinds of problems

This is the section of his post that caught my eye, comparing the history of regulation of cars with that of technology:

To take this point further, cars brought many different kinds of problem, and we understood that responsibility for doing something about them sat in different places, and that solutions are probably limited and probably have tradeoffs…. ‘Tech’, of course, has all of this complexity, but we’re having to work this out a lot more quickly. It took 75 years for seatbelts to become compulsory, but tech has gone from interesting to crucial only in the last five to ten years.

Obviously, the reason for writing about this now is because a Congressional Committee has just got a lot of airplay by calling the ‘big four’ to testify. Mark Zuckerberg of Facebook, Jeff Bezos of Amazon, Sundar Pichai of Google, and Tim Cook of Apple all appeared. Maybe Microsoft’s Satya Nadella didn’t get the memo.

If you want to catch up with what happened at the hearings, The Verge can tell you more than you likely want to know.

Technological change

Regular readers of this blog will know that when it comes to thinking about change in technology, my go-to model is the work of Carlota Perez. Her model of technological change connects innovation and finance, and goes back to the Industrial Revolution. It suggests that dominant technology platforms, in each era, have a lifecycle of 50-60 years.

There have been five such technology ‘surges’ (her word). The first, based on cotton, canals and water power, started in 1771; the second (rail and steam) in 1829; the third (steel and electricity) in 1875; the fourth (cars and oil) in 1908; and the fifth (information and telecommunications) in 1971.

Distinctive innovations

Each surge follows a similar pattern. A first phase—the “installation” period—sees financial or investment capital investing in infrastructure; there’s a crash, when market expectations run ahead of take-up; and production capital then picks up the pieces as the technologies become embedded in everyday life and everyday commerce.

A surge starts with a distinctive technological innovation that then shapes what follows. For the first surge, of cotton and canals, it was Crompton’s Spinning Jenny. For the cars and oil surge, it was Henry Ford’s assembly line. For the ICT surge, it was the microprocessor, in 1971, and the associated invention of the internet protocols at around the same time.


As seen above, the whole cycle follows an S-curve. The crash is normally around half way through the cycle in terms of time, but earlier than that in terms of diffusion.

The dot.com crash, for example, is 30 years after the start of the information and telecommunications surge. At the time, the take-up of domestic internet use was of the order of 20-25%, even in leading markets.

And similarly for the autos surge. In 1929, despite Henry Ford’s best efforts, the car was still mostly the preserve of the rich and the upper middle-classes.

So, when Ben Evans airs a chart headed “Tech was very small until recently”, he’s actually telling us that Perez’ model is right. It’s an S-curve that has accelerated through the middle stage (and is probably about to slow down as it hits the ‘maturity’ phase).

Source: Benedict Evans

Let’s just play that again, taking the start date back to Perez’ 1971 and the invention of the microprocessor, and mapping on the four quarters of the S-curve—and the crash.

Source Benedict Evans, annotated by the next wave.

Business to business

But then again: even then, it wasn’t that small. Evans makes this claim a couple of times, but Microsoft’s market cap after its IPO in 1986 was more than $500 million. During the 1990s its software was on 90% of all PCs shipped—which even in a smaller market was still substantial. There was a reason why the US Department of Justice took Microsoft to court over anti-trust allegations.

And one more point about this. The first half of the technology surge was largely a story of business to business use, but even then its effects were clear enough to academics such as Manuel Castells, who wrote about this in his book The Informational City in 1989, published the same year as Tim Berners-Lee published the protocols for the World Wide Web. Claims about the novelty of the tech sector are always overblown.

Regulation catches up

In my own work with the Perez model, I have noticed that regulation tends to catch up in the final quarter of the S-curve. Penetration levels are high, and markets are saturated. There’s still noise about an exciting future, but innovation is running out of steam.

The reason for this is that with scale the external costs of the technologies in the surge start to become completely obvious. Constraints on car use—seat belts, parking meters, drink driving laws—all emerged in the 1960s, when the surge was mature: it has reached its 50s. Similarly, Jane Jacobs’ book on American cities—part of her campaign against Robert Moses’ Manhattan Expressway, appeared in 1961.

So what do we see when we look at the technology surge? It hits 49 this year, and the regulatory interest, which has been emerging for a while, is starting to get more serious. There is talk of controls and sanctions, even break-up. All this talk will become more serious.

Patterns of development

So why does an experienced and deeply well-informed tech commentator and investor such as Benedict Evans miss all of this? Part of it is the endless story that tech is different, that things are speeding up, etc, which has been part of the self-serving sales and investment hustle that has inflated the size of IPOs.

But part of it is that he’s over-focussed on the companies themselves, and not on the overall pattern of development. The majority of the big tech companies became visible only during the Production period. Amazon was founded in 1994; Google, 1998; Facebook, 2004.

Apple, which dates back to 1976, reinvented itself in the early 2000s, after the dot.com crash, after coming close to bankruptcy in the late 1990s. (It was saved from this when Microsoft made a $150 million investment in the company).

Modern management

But this isn’t so different from the car phase either. General Motors was also, largely, a creature of the Production phase: Alfred Sloan invented modern management in the late 1920s and after to impose clarity on the ramshackle collection of businesses that Billy Durant had assembled, overtaking Ford in the process.

Ford, despite having invented the mass production car industry, struggled in the Production phase until Robert McNamara was drafted in to run the business after World War II.

Obviously Benedict Evans is right when he says, in his post, that regulation is complex. That doesn’t mean that it isn’t necessary.

A thicket of products

As Kashmir Hill reported in the New York Times last week, she spent two months trying to live outside the systems created by the big technology companies—and found it impossible:

Critics of the big tech companies are often told, “If you don’t like the company, don’t use its products.” My takeaway from the experiment was that it’s not possible to do that. It’s not just the products and services branded with the big tech giant’s name. It’s that these companies control a thicket of more obscure products and services that are hard to untangle from tools we rely on for everything we do, from work to getting from point A to point B.

Her article quotes the US Representative Jerrold Nadler. He makes the kind of connection that Perez makes between different technology surges.

“By virtue of controlling essential infrastructure, these companies appear to have the ability to control access to markets,” Mr. Nadler said. “In some basic ways, the problem is not unlike what we faced 130 years ago, when railroads transformed American life — both enabling farmers and producers to access new markets, but also creating a key chokehold that the railroad monopolies could exploit.”

One of the things that happens in every technology surge is that the technology embeds itself in our culture and metaphors. This makes it harder to see what‘s happening as it runs out of energy. Taking a 250-year view, digital tech looks a lot like other technologies.