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.

S-curves

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).

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.

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.

In a recent post, I mentioned John Lanchester’s view that Facebook, and by extension other software giants represented a new form of surveillance capitalism. In his LRB article, in which he reviewed three recent books on things digital, he is pretty trenchant on this subject.

Facebook, in fact, is the biggest surveillance-based enterprise in the history of mankind. It knows far, far more about you than the most intrusive government has ever known about its citizens. It’s amazing that people haven’t really understood this about the company. I’ve spent time thinking about Facebook, and the thing I keep coming back to is that its users don’t realise what it is the company does. What Facebook does is watch you, and then use what it knows about you and your behaviour to sell ads. I’m not sure there has ever been a more complete disconnect between what a company says it does – ‘connect’, ‘build communities’ – and the commercial reality.

Although one of the elements of received wisdom about digital platform businesses such as Facebook is that they tend to monopoly because of their network effects, he can also see trends that might curb its growth, or even see it shrink. There are four, or perhaps five.

Disenchantment

The first is what Timothy Wu calls in The Attention Merchants “disenchantment effects”—that periods of rapid media growth produce a backlash, sometimes with accompanying regulation or legislation. This is why Paris still has tight control of display advertising. This extends beyond media. When significant technologies become mature, and their rate of growth slows, there’s a period of social stocktaking in which social side effects (external costs, in other words) start to be regulated. Think of ‘60s innovations such as seat-belts, parking meters, and drink-driving legislation, in response to road deaths and congestion.

Lanchester focuses on customer disenchantment, meaning advertiser disenchantment.

[A] lot of the clicks on these ads are fake. There is a mismatch of interests here. Facebook wants clicks, because that’s how it gets paid: when ads are clicked on. But what if the clicks aren’t real but are instead automated clicks from fake accounts run by computer bots? This is a well-known problem, which particularly affects Google, because it’s easy to set up a site, allow it to host programmatic ads, then set up a bot to click on those ads, and collect the money that comes rolling in. On Facebook the fraudulent clicks are more likely to be from competitors trying to drive each others’ costs up.

Click fraud

Indeed, Ad Week has estimated the cost of click fraud at $7 billion, representing one-sixth of the market. Some of the estimates of click fraud are a lot higher than that. These are not trivial sums, especially in a market that is facing serious cost pressures. Even legitimate views are, shall we say, enhanced by the way Facebook chooses to measure use.

A video is counted as ‘viewed’ on Facebook if it runs for three seconds, even if the user is scrolling past it in her news feed and even if the sound is off. Many Facebook videos with hundreds of thousands of ‘views’, if counted by the techniques that are used to count television audiences, would have no viewers at all.

Inevitably, Facebook has tried to claim that its ads have the most impact in the first two seconds: advertisers are sceptical.

But there might also be end-user disenchantment. We have seen generational change affect online use before, and Facebook might also be vulnerable to this. There’s some evidence that younger people are sharing less on more open platforms, and using closed platforms instead, such as WhatsApp. Even though Facebook also owns WhatsApp, there is less data to be seen and used in a closed end-to-end platform. People don’t need to stop using Facebook to erode much of its value; they just need to use it less.

Unhappy

Another source of end-user disenchantment might be the emerging evidence that Facebook use is bad for well-being, given that wellbeing is one of the strong long-term shifts in social values. Lanchester notes that “the more people use Facebook, the more unhappy they are.” It is possible that this is a correlation, not causation, but I expect to see more on this issue.

Second, there could be more formal regulation, and this might be related to advertising, particularly political advertising. Every time it is clear that Facebook’s algorithms actually allow people to advertise to “Jew-haters,” as ProPublica recently discovered, or that Russian interests have paid Facebook for American political ads during the 2016 Presidential campaign. The fact that Facebook profits so handsomely from “fake news” that it has no incentives to do anything about it —what’s good for Facebook is bad for America—might also spur some kind of regulation.

And right on cue: [Update 1] the EU has just warned Facebook and Twitter that they will regulate them unless they take steps to remove hate speech in a much more timely way.

And, [Update 2] Congress is actively concerned about the Russian ads issue, as Buzzfeed reports:

The Russian ad scandal has captured lawmakers’ attention in a way Facebook’s previous political crises — from allegations of bias in its Trending column to its role in spreading fake news — have not. It has crystallized a trio of individual fears — Facebook is too big, has too much influence, and cannot effectively monitor itself — into one big expression of all of them.

Pricing algorithms

Third, one of the trends implied by surveillance capitalism is individualised pricing. As Lanchester observes, the big internet enterprises have so far escaped competition investigation, because—in the US—this typically focuses on consumer pricing and consumer benefit. Because Facebook and Google are free, in the sense that they don’t cost money (we’re paying them with our data, which they sell on), they have largely escaped competition scrutiny in the US, though not, of course, in Europe.

But the point at which their activities lead to individualised pricing, and especially if this pricing is driven by algorithms that can’t be independently tested or evaluated, is the point at which this may well change. Even without that, there will surely come a point where a competition economist decides that algorithms that increase the price of an airfare when you return for a second look are anti-competitive.

Cookie wars

The other issue is that not all of the companies that have grown rich on the digital boom share the same interests. Unlike Facebook and Google, Apple’s customers are end-users who pay them for their products (Amazon sits somewhere between). It’s better for Apple’s business model to position themselves as being on the side of their customers; Apple’s software is increasingly designed to reduce intrusion.

[T]he new Safari feature uses a “machine learning model”, Apple says, to identify which first-party cookies are actually desired by users, and which are placed by advertisers. If the latter, the cookie gets blocked from third-party use after a day, and purged completely from the device after a month, drastically limiting the ability of advertisers to keep track of where on the web Safari users visit.

And it’s worth noting the extreme response by the ad industry to this relatively modest restriction on what advertisers can do with end-user data:

Apple’s unilateral and heavy-handed approach is bad for consumer choice and bad for the ad-supported online content and services consumers love. Blocking cookies in this manner will drive a wedge between brands and their customers… As organizations devoted to innovation and growth in the consumer economy, we will actively oppose any actions like this by companies that harm consumers by distorting the digital advertising ecosystem and undermining its operations.

Well, blah, blah, blah. If controls such as those that Apple is proposing are going to be so damaging, then clearly the “digital advertising ecosystem” is a lot more fragile than we thought it was.

Pushing monetisation

It is worth putting all of this in a wider context. Facebook’s expansion has been driven by a simple model:

Growth (in numbers) x Monetisation (of each user).

There are strong grounds to think that user growth has run of road, certainly in mature markets, which really only leaves monetisation. What monetisation means is getting advertising in Facebook and Google to work harder, which means, for example, inserting more ads in feeds and being more unscrupulous with user data. We know from experience that end-users dislike both of these things: in other words, the pursuit of monetisation could turn into a vicious circle.

One final note here, from Nick Srnicek’s book on Platform Capitalism. He points out that advertising markets are more fragile than they seem, especially to recession. The last time, ad-based platforms like Facebook and Google were able to ride out the recession through growth in numbers, which was still strong.

He doesn’t add that they were also insulated at the time from profit/performance pressures from shareholders and investors because both were privately held at the time. Both are now at least partially listed on the stock exchange.

Recession, in short, could be a catalyst that makes the combination of pressures outlined here start to have an effect on how these surveillance businesses work. We could be close to peak Facebook.

The image at the top of the post is by Marc Smith, and is published here under a Creative Commons licence. It was also used to illustrate an article on The Conversation on the subject of surveillance capitalism.

Chaos Monkeys takes us from Goldman Sachs to Facebook, and out the other side, via a digital ad business called Adchemy,  the start-up ‘school’ Y Combinator, and the startup, AdGrok. Mostly Martinez is a good guide; especially to the world of venture capital and startups. Martinez persuades a couple of engineer colleagues at Adchemy, which is clearly struggling, to pitch to Y Combinator, the start-up hothouse then run by Paul Graham. Going in, they have a product concept that probably isn’t going to work, but that doesn’t matter because Y Combinator, like much of the rest of Silicon Valley, is as interested in teams as it is in products.

Miracles

Why isn’t their product going to work? This is a flash of innovation insight that’s worth the price of the book and quite a lot more.

When confronted with any startup idea, ask yourself one simple question: How many miracles have to happen for it to succeed?

If the answer is zero, you’re not looking at a startup, you’re just dealing with a regular business… To be a startup, miracles need to happen. But a precise number of miracles.

Most successful startups depend on one miracle only. For Airbnb, it was getting people to let stragers into their spare bedrooms and weekend cottages. This was a user behavior miracle. For Google it was creating an exponentially better search sevice than anything that had existed to date. This was atechnical miracle. For Uber or Instacart, it was getting people to book and pay for real-world services via wesites or phones. This was a consumer-workflow miracle…

The classic sign of a shitty startup is that it requires at least two (or more!) miracles to succeed. (P50-51)

Y Combinator is particularly good at getting its progeny in front of investors, but even allowing for that the commitment, rat-like cunning and persistence needed to get it across the line with investors and an eventual acquisition is more or less complete. As he says when he has the chance to pitch to a potential buyer and has a computer glitch, rule #1 of the start-up is “Always be closing”.

VCs at work

The history of AdGrok, and its eventual acquisition following some initial investment, is almost a case study in VCs at work, and Martinez is pretty good at explaining the mechanics of how the VC business works (and the different and sometimes conflicting things that different types of VCs want from the eventual deal), even if you sometimes have to go back and check on some of the technicalities. This focus on the mechanics of the start-up reminded me of Jerry Kaplan’s 1990s classic Startup.

I hadn’t realised before I read the book of the extent to which the big Silicon Valley companies use startup acquisitions as a way of acquiring talent rather than software or patents. The rationale is that acquiring good product managers and engineers is difficult, and they come expensive, so using discussions about their product is a good way of assessing cultural fit. And bringing in people who have been through the startup mix, rather than looking first for a long-run career at a big company, injects some energy into the culture, and that energy is worth paying a premium for.

Breakfast, lunch and dinner

In the end (spoiler!) Martinez has the chance to go to Twitter or Facebook. He chooses Facebook because of the full-on nature of the demands that the business makes of its staff, compared to Twitter, and this was partly learned from a casual remark from a Twitter engineer, when Martinez asked him what he liked most about Twitter.

“Well, you know, in companies like Facebook and Google, they serve you breakfast, lunch and dinner. Here at Twitter, they only serve you breakfast and lunch.”

I cringed inwardly. So the big selling point was that nobody worked late into the night…? (P237)

And after two years (more spoilers) after his live adserving product FBX loses out in an internal power struggle, he’s gone again:

I could barely remember what my life was like before Facebook, and there was a trail of destruction I had caused by spending m entire life there: two children neglected, two different women whose worthy love I’d spurned, two boats rotting in neglect, and anything like an intellect or a life outside campus rotting in neglect due to indifference and my dedication to the Facebook cause. (P458)

If he’d stayed with the rest of the AdGrok team and gone to Twitter, he’d also have been about $1m better off and would not have burned his bridges irrevocably with one of his investors.

And no: nowhere in the book does he make the connection between these outcomes and his disdain for Twitter’s failure, at the time, to serve dinner.

If he were chocolate

There are things not to like about Chaos Monkeys. It’s readable, but it’s too long; a competent editor could have taken 80 pages out of it without blinking, and improved it a lot. For example, I didn’t need to know about an inconsequential fling with a Facebook colleague in a company stationery cupboard, nor his streetrace in a carshare Tesla; I already knew he was reckless and competitive by then. He tries not to be, but he’s too pleased with himself a lot of the time. The lawyer who bailed AdGrok out of a hole may have agreed to a terrible deal (he accepted equity in the startup with no risk premium) but I’m not sure why Martinez needs to rub his face in it.

There are too many footnotes, all too long. He tips his hat in a look-at-me sort of way to Michael Lewis’ Liar’s Poker without really understanding what made that such a fine book (well written, aware of his own shortcomings as a bond salesman, telling anecdotes, watching the start of deep transformation in the way investment banks work.) Martinez is a good observer, but if he were chocolate he would eat himself.

Saving Facebook

The reason that the advertising market and its monetisation became so important to Facebook in that period was because of its IPO–its revenues were flat and it needed to show growth to its new shareholders. The solutions that were attempted were basically a develop of Facebook’s existing, but closed, ad platform, and FBX, Martinez’ project, which was an open platform. Facebook’s culture and model is about closed systems, and that approach commanded most of the resources. The FBX platform, with its small team and limited resources, was essentially an insurance policy, although Martinez, for all his smarts, never seems to realise this.

But the thing that saved Facebook’s financial position was neither of these things. It was the rapid shift in 2013 towards mobile and device use. If the shift was expected, its timing and speed were not. This allowed Facebook to insert ads profitably into user feeds on mobile.

News Feed Ads rode to success atop a tsunami-esque wave nobody had predicted, or at least hadn’t predicted to arrive right then and so quickly. In this case, that wave was mobile, which in the space of a few months in 2013 suddenly constituted the majoriy of Facebook usage… [D]ata and high quality formats and placements, meant that Facebook dominated mobile like few oter incumbents had managed to. (P483, 487-8)

Carthage must be destroyed

Despite being fired by them, Martinez clearly remains a fan of Facebook. He defends its position on privacy vociferously (they’re not a threat to your privacy, he says) and he admires the focus that Zuckerberg has on avoiding complacency. When Facebook moves into Sun’s former buildings in Menlo Park, Zuckerberg left some of the old Sun signs up, as a reminder, as it were, that this too could pass. When Google launched Google Plus, a clear bid for Facebook’s market space, the phrase Carthage delenda est (“Carthage must be destroyed”) became a company catchphrase.

John Lanchester, on the other hand, is not such a fan. He sees Facebook as the exemplar of a new form of surveillance capitalism. More on that in another post.

The short book falls into three parts. There is a context-setting chapter on ‘The long downturn,’ which will surprise no-one who has read, for example, Wolfgang Streeck’s book and articles on the end of capitalism. The second chapter analyses different types of platforms and their strengths and weaknesses. A final chapter, ‘Great platform wars’, looks at the prospects.

Official futures

One of the strengths of this approach is that it provides some valuable context. One early observation is that the tech sector remains small by value and even more by employment. In the US it represents 7% of the value added by private companies, and 2.5% of the employment. However, it is also a pervasive technology, both systemically and ideologically.

The ‘official future’ (my phrase) of the digital economy, as he observes, is that

cities are to become smart, businesses must be disrupted, workers are to become flexible, and governments are to become lean and intelligent. (p5)

The critical view, in contrast:

[W]ith a long decline in manufacturing profitability, capitalism has turned to data as one way to maintain economic growth and vitality in the face of a struggling production sector. (p6]

Platforms as data engines

This is where platforms come in. If data has become a massive new raw material for capitalism, then platforms are the engines that allow it to process this data. Srnicek identifies five types of platform (p49), described below. For me this is a valuable piece of analysis, since it allows greater clarity about their differences and their strategies, rather than bundling them together as if they are a single phenomenon. What they do have in common, though, is that “collecting massive amounts of data is central to the business model.” (p89)

• Advertising platforms (e.g. Google, Facebook), which extract information from on users, analyse it, and use the outcome of that process to sell advetising
• Cloud platforms (e.g. AWS, Salesforce), which own the hardware and software used by digitally-dependent businesses and rent them as needed
• Industrial platforms (e.g. GE, Siemens), which build the hardware and software necessary to transfer traditional manufacturing into internet-connected processes
• Product platforms (e.g. Rolls-Royce, Spotify), which generate revenue by transferring a traditional good into a service
• Lean platforms (e.g. Airbnb, Uber), which seek to connect buyers and sellers of a service while maintaining a minimum of assets. (I have adapted Srnicek’s definition slightly here to emphasise the *market-making* nature of such platforms.)

Questions of scale

For the detail of these you’ll need to turn to the book, but pulling them apart like this helps understand their differences. Advertising platforms have been hugely profitable, to the point where they don’t know what to do with their cash, but are running out of road in terms of growth and monetisation. Cloud platforms generate good returns (AWS is by far the most profitable part of Amazon) but require significant scale and investment, as well as robustness. Industrial platforms are specialist versions of cloud platforms; they need scale, and to be able to handle prodigious amounts of data. (Srnicek reports that GE’s natural gas platform alone is collecting as much data as Facebook.)

Product platforms, perhaps better thought of a service platforms, own the asset that they lease to the end user (this is the difference between Zipcar and Uber), although in the case of Spotify that “ownership” takes the form of a licence from the underlying rights holders, which is a problem for its business model. There’s a useful discussion here of the aero-engine market, now moved more or less completely to a service model: effectively this allows the engine companies to integrate higher margin maintenance with the lower margin product by connecting them through their data.

Venture capitalist welfare

Finally, lean platforms need to own the relationship with both buyers and sellers. They are lean because they outsource everything else. “It’s hard,” writes Srnicek, “not to regard the new lean platforms as a retrogression to the earliest stages of the internet-enabled economy.” Or further, I’d say, to the piecework of early industrial capitalism (pdf). As Srnicek notes, “the most notorious part of these firms is their outsourcing of workers.” (p75,76).

It’s worth taking a small detour here on lean platforms, party because Srnicek suggests that despite the hype about them, not to mention that tedious conference trope, which was apparently created by IBM, they are the most minor of these five.

The profit-making capacity of most lean models likewise appears to be minimal and limited to a few specialised tasks. And even there the most successful of the lean models has been supported by VC [venture capitalist] welfare rather than any meaningful revenue generation…. these models seem likely to fall apart in the coming years. (p87-88)

Building on this, Airbnb appears sustainable, because it can charge a larger price per unit than the other two, and therefore take a larger cut itself, the units have lower running costs, and often do represent genuinely marginal capacity, and the balance of power between seller and platform is more equal. Uber, on the other hand, has relied completely on “VC welfare” to undercut conventional taxi prices—to the point where the scale of its price advantage matches almost exactly the losses per trip funded from its reserves. I’m always surprised that competition authorities haven’t paid closer attention to this aspect of Uber’s business model.

Platforms need to expand

In the final chapter, he uses an analysis of the ‘tendencies’ within platform capitalism to identify some of the long-run trends. Since platforms are based on the extraction of data and the creation of network effects, they also have certain dynamics: “expansion of extraction, positioning as a gatekeeper, convergence of markets, and enclosure of ecosystems.” (p98)

Taking these in turn, platforms need to expand because they need more data. Because they need to expand, they tend to encroach on the same markets and the same types of data. One can see this most clearly in the competition between Google and Facebook, for example. Srnicek asks whether this tendency goes beyond competitors around similar platform spaces (e.g. Amazon and Alibaba) to platforms in general. I suspect this is unlikely. The skills and competencies needed to run an industrial platform like GE’s Predix, not to mention the customers and sales channels, seem to me to be fundamentally different to a consumer platform like Spotify or an advertising platform like Google.

Locking in customers

The final tendency, about the enclosure of ecosystems, is because the acquisition of data does not, finally, ensure a relationship with the customer, and so they have to be locked in, as much as this is possible without irritating them. Facebook has been effective as this; Apple too. In the industrial space, it is even more acute; it is possible to migrate your sales and customer data from Salesforce, but it is not something you would do lightly. Apps reinforce this tendency towards closed systems.

There are challenges here. Srnicek mentions the limits of outsourcing, a particular problem for lean platforms, the dependence on loose finance capital, and, for advertising platforms, a dependence on advertising growth. Platforms do continue to grow, but to do this, they have “to extract rents by providing services.” On this basis, “Amazon is more the future than Google, Facebook, or Uber.” (p126)

What is to be done? Srnicek suggests that in the short term national governments (and by extension, cities) might take more interest in platforms and their anti-social side effects:

Antitrust cases can break up monopolies, local regulations can impede or even ban exploitative lean platforms, government agencies can impose new privacy controls, and co-ordinated actions on tax avoidance can draw capital back into public hands. (p127)

Beyond that, there might be scope for publicly owned and publicly controlled platforms, creating new public utilities. There are co-operative platforms out there already.

Reaching limits

My own take on this is that while there should certainly be more controls on what platforms do, they might, nonetheless, be reaching their limits. Given that they need data to fuel their expansion, this comes from two places: growth in the number of customers, and second, growth in how much you can get those customers to pay for (monetisation).

In most markets, growth in customer numbers is plateauing, or close to it, in both consumer and B2B markets. And in a world of long-run slow economic growth, the prospects for further monetisation are slowing too. This may lead to some unpleasant effects as they try to maintain growth by gouging customers or playing faster or looser with their data. This also suggests that platforms that do not already have a sustainable business model are unlikely to get to one.

This means, in turn, that the loose finance capital (e.g. venture capital money) that has been keeping such platforms afloat are likely to cut their losses and move on to other areas of investment, probably not even in the digital sector.

Platform Capitalism is published by Polity Press. The image at the top of the post is by Andrew Curry, and is published here under a Creative Commons licence.

There’s an interview in Fast Company with the former CEO of Groupon,Andrew Morton, who was forced out of the company after some disastrous results. About halfway through he tells his interviewer, “the moment a company goes public the conversation shifts from how they’re trying to change the world and the product they’re building to how they’re making money.” Of course: we all want to change the world, but that’s not the reason investors put money into an IPO

But it’s maybe a thought that should be on Mark Zuckerberg’s mind as well these days. Zuckerberg holds onto his belief that Facebook is on a mission to change the world, but the numbers aren’t looking good. It’s hard to avoid the notion that the company’s IPO caught the peak of Facebook sentiment, and the only way is down. The company’s problem is exemplified by its Home phone, previewed this month, and summarised by my Futures Company colleague Chloe Cook as “Essentially, Facebook gets wallpapered over the inside of your phone.” It effectively locks its users in to Facebook, which led the British commentator John Naughton to describe the company as a “pathogen”.

Monetisation and intrusion

So it’s worth unpacking Facebook’s problem a little. In a line, it goes like this: the only way to monetise the site is to be more intrusive into your users’ lives, and it turns out that users don’t like that. People in mature markets aren’t leaving the site in droves, but they’re dialling down their level of engagement, visiting less often and sharing less stuff. All of which means less opportunity for Facebook to make money.

In the mobile world, this is more of a problem, since people are more aware of intrusion as it gets closer into them, and Facebook has always been behind the curve on mobile. But while it’s easy to spot the benefit to Facebook of a phone that locks you into Facebook, it’s hard to see the user benefit. The world’s not exactly short of alternative phones.

But there’s a deeper problem for Facebook. There’s a generational curve on social networking – the next generation of teens doesn’t want to do the same things as the previous one. And Facebook seems to have hit that moment. Some fascinating qualitative research with British teens was unveiled at a conference here recently, and some of the comments should be ringing alarm bells at Facebook head office..

“I don’t use it half as much as I did. My Mum’s got more friends than I have.””I just use it to upload family photos – so my Gran can see them.”

“I just use it to upload family photos – so my Gran can see them.”

“I use it to message friends. I haven’t updated my status for ages – because you’re judged by what’s posted, so I don’t post.”

In short, if you’re in business to change the world, it helps if make a product or service that your customers want to use and your suppliers don’t feel abused by. Locking people in so you can screw them over has never been a compelling business model.

The picture is taken from Mashable’s excellent ‘what is Facebook Home‘ article, and is used with thanks.

Edited with BlogPad Pro

For anyone who’s missed the story, Wired probably has the best summary: Facebook realised sometime last year that it had enough users and knew enough about them to transition itself from a social networking site to the place where its users kept their online identities – clearly a more valuable commercial proposition. But to do that, it needed to turn personal data into public profiles:

So in December, with the help of newly hired Beltway privacy experts, it reneged on its privacy promises and made much of your profile information public by default. That includes the city that you live in, your name, your photo, the names of your friends and the causes you’ve signed onto. This spring Facebook took that even further. All the items you list as things you like must become public and linked to public profile pages. If you don’t want them linked and made public, then you don’t get them — though Facebook nicely hangs onto them in its database in order to let advertisers target you.

There’s a great visual summary of how quick and how dramatic this change has been by the IBM researcher Matt McKeon. And at the same time, it’s become much more complex for users to manage their privacy settings; it’s one of those things that’s mostly possible in theory, but difficult and time-consuming in practice.  It’s worth checking out an application called openbook which searches profiles and demonstrates how people saying unwise things in their status updates are – probably unwittingly – broadcasting them to the internet.  As openbook puts it:

Facebook helps you connect and share with the people in your life. Whether you want to or not.

Facebook’s been playing loose with its users’ privacy settings pretty much ever since it moved out of its original student user base. The company’s publicly stated view is that attitudes and values about privacy are changing and it is merely responding to these changes (but then, it would say that, wouldn’t it?). Its founder Mark Zuckerberg made a speech in January this year in which he said that “Privacy is dead”. (Whether he believes it or not, and there’s some evidence that he might, it’s a convenient point of view if billions of pounds are riding on it). And Zuckerberg has played fast and loose, pretty much with everyone, ever since he started along the Facebook route, as Jason Calcanis has pointed out.

The internet and privacy

It’s true that the internet does test ideas about privacy. Google’s had a number of problems here as well, most recently when it launched Google Buzz without thinking it through. At the SXSW conference this year, the researcher danah boyd picked up on both companies in a presentation. As she wrote later, in a blog post that’s the best single introduction to the Facebook privacy question, the response was instructive:

My SXSW used a bunch of different case studies but folks focused on two: Google and Facebook. After my talk, I received numerous emails from folks at Google, including the PM in charge of Buzz. The tenor was consistent, effectively: “we f*cked up, we’re trying to fix it, please help us.” What startled me was the radio silence from Facebook, although a close friend of mine told me that Randi Zuckerberg had heard it and effectively responded with a big ole ::gulp:: My SXSW critique concerned their decision in December, an irresponsible move that I felt put users at risk. I wasn’t prepared for how they were going to leverage that data only a few months later.

When you’re in a hole, I suppose, the best tip is to stop digging. And we don’t yet know how this will play out. Facebook has got huge critical mass; when I talk to some of my younger colleagues, they feel locked in. They’re aware of Facebook’s behaviour and increasingly bad reputation, but their life and their social networks are mediated through Facebook. They are the victims of bad timing. Nancy Baym has a good post on her mixed feelings about this; a desire for an ethical Facebook. [And, update, 21/05, O’Reilly has just asked on its blog about people’s “Facebook tipping point”.]

So here are my six lessons from Facebook.

1. If you make changes rapidly, you will cause social contention. It’s never good for business in the short-term. Despite the online noise, Facebook is probably gambling that because people are effectively locked in, most of its users will stay put. The financial prize is worth riding a storm for. I think this is short-sighted; all social networks go through social cycles (younger kids, for example, don’t want to be where their older siblings are), and people do leave. A bad reputation creates opportunities for competitors.
2. Online businesses become monopolies much more quickly than offline businesses, because of the economics of “network effects”.  We’ve seen this already with Microsoft and Google. Facebook’s privacy changes have already prompted a complaint (opens pdf) about the company’s (alleged) “deceptive trade practices” to the US Federal Trade Commission by the Electronic Privacy Information Center and others. EU data protection regulators are also alarmed. As Microsoft discovered, it becomes increasingly difficult to do business once regulators are monitoring you closely. One of the more unexpected developments was that the US activist site Move On joined the debate, setting up an online petition about privacy, although it has little history of engagement with digital issues. (The picture at the top of the post comes from them). It’s not the sort of organisation you want on your tail.
3. The notion that ‘privacy is dead’ in the digital world is fashionable but not borne out by an analysis of social trends. In client work I’ve done at the Futures Company we’ve found that there’s a division between “open engagement” and “reluctant engagement”: the second group are concerned about the negative aspects of the web, whether fraud, identify theft, or undesirably social consequences. The continuing rise in spam and phishing attacks are understood by users. One of the consequences is that companies whose future is bound up with the integrity of the web need to defend it – indeed Nick Carr argued in January that this was the reason that Google had decided to pull out of China. If this is true, Facebook’s strategy – short-term advantage against long-term corruption of the overall operating environment – is a classic example of killing the golden goose.
4. There are complex questions are about the nature of the types of public, or publics, which are created in online spaces. The Facebook model is of one large public, but in practice each of us is a member of different overlapping publics. We don’t show all of ourselves to everyone. There’s not space to go into this in much detail, but Jeff Jarvis argued two weeks ago that Facebook confused the notion of a single public – even a ‘public sphere‘ – with the different publics, or communities, that we choose to create for ourselves. Users see Facebook (and Twitter) as a vehicle for the latter, not the former. [Update: And arguably, the “public sphere” can only be created in a public space, not by a private or commercial organisation.]
5. We will see new entrants into the social networking space with clear commitment to privacy. One which is getting a lot of coverage at the moment is Diaspora*, an open source network which seeks to place control of networks and content in the hands of its users, and which has raised $180,000 of crowdsourced development finance on Kickstarter – against an target of $10,000. Over at Shareable, Neal Gorenfio wrote what amounted to a manifesto for ‘next generation civic networks’. This is a battle which goes back to the heart of the competing stories (and histories) about the internet, and recurs constantly. But it’s also worth remembering Kim Cameron’s Laws of Identity, which suggest that users engage much more freely in online spaces if they can manage their own privacy settings. This isn’t a tussle of idealists against merchants; it is a central question for online business strategy.
6. The web is acquiring the attributes of a utility. We’re moving to a phase in the web’s development where it is increasingly pervasive, moving from ‘installation’ to ‘deployment’. And, certainly in Europe and elsewhere, utilities tend to get regulated to prevent them exploiting their customers. We haven’t yet found reliable methods for managing dominant online entities; mostly, so far, it’s been done through a combination of trade law and competition law. This works, up to a point, but it is a bit ad hoc. But every time a rapacious online player abuses its market position, it makes new regulatory intervention more likely.

My bet, for what it’s worth, is that Facebook has reached the limits of its size and influence. Suppliers and partners are as suspicious of it as its users. The way from here is likely to be down. But it will be a long time falling.