The Tragic iPad

From The Verge:

Steve Jobs stepped onstage 10 years ago today to introduce the world to the iPad. It was, by his own admission, a third category of device that sits somewhere between a smartphone and a laptop. Jobs unveiled the iPad just days after the annual Consumer Electronics Show ended in Las Vegas and at a time when netbooks were dominating personal computing sales…

Apple had an answer to the netbook: a 9.7-inch tablet that allowed you to hold the internet in your hands…Apple was also looking to create a third category of device that was better at certain tasks than a laptop or smartphone. The iPad was designed to be better at web browsing, email, photos, video, music, games, and ebooks. “If there’s going to be a third category of device it’s going to have to be better at these kinds of tasks than a laptop or a smartphone, otherwise it has no reason for being,” said Jobs.

Stratechery wasn’t my first (or second) blog; back in 2010 I had a Tumblr and I imported some of the posts to Stratechery, including this piece that I wrote when the iPad was announced:

What the iPad does is give Apple a product that offers a superior experience in every dimension of the mobile experience, namely, content creation, content consumption and mobility.

Apple's mobile device offerings

The reason this matters is that the vast majority of users are primarily content consumers. These are the people buying netbooks as their primary computers, or simply avoiding computers as much as possible. They simply want to go on Facebook, check their email, watch YouTube, and at most, upload pictures. Apple’s value proposition to these customers is: The iPad is a superior content consumption experience with sufficient creation capabilities to meet your needs. That is why iWork figured so prominently into the Keynote — it was reassurance that the iPad can pass as your only computer (more on iWork in just a moment).

The post holds up pretty well, if I might say so myself, but it is where it is wrong that is the most interesting.

The iPad Disappointment

John Gruber is disappointed in the current state of the iPad:

Ten years later, though, I don’t think the iPad has come close to living up to its potential. By the time the Mac turned 10, it had redefined multiple industries. In 1984 almost no graphic designers or illustrators were using computers for work. By 1994 almost all graphic designers and illustrators were using computers for work. The Mac was a revolution. The iPhone was a revolution. The iPad has been a spectacular success, and to tens of millions it is a beloved part of their daily lives, but it has, to date, fallen short of revolutionary…

Software is where the iPad has gotten lost. iPadOS’s “multitasking” model is far more capable than the iPhone’s, yes, but somehow Apple has painted it into a corner in which it is far less consistent and coherent than the Mac’s, while also being far less capable. iPad multitasking: more complex, less powerful. That’s quite a combination.

I could not agree more with Gruber’s critique. In my opinion, multi-tasking on the iPad is an absolute mess, and it has ruined the entire interface; I actively dislike using the iPad now, and use it exclusively to watch video and make the drawings for Stratechery. Its saving grace is that it is hard to discover.

What is fascinating — and, in my opinion, tragic, in both the literal and literary sense — is how the iPad arrived in its current state. That initial announcement featured Jobs reclining on a couch — it wasn’t very difficult to come up with the “content consumption” angle! Still, you could see the potential for something more. I wrote at the end of that piece:

It’s the long-term picture that is particularly fascinating, and gets back to my contention at the beginning of this post. For while the laptop has all but reached it’s potential — the consumption experience will never improve beyond what it is now — the creation experience on the iPad will only get better with time. In fact, I believe the iPad will be looked back upon as the pioneer of what will become the default way of interacting with computers just like the Macintosh.

Go back and watch the Keynote again, especially the iWork demonstration that begins 57 minutes in. The iPad doesn’t just let you create documents. It lets you create documents in a way that is simply impossible on a normal computer. It is so much more natural, so much more intuitive, that users accustomed to a keyboard-and-mouse will adapt quickly, and more importantly, users accustomed to multitouch will never understand the attachment to a mouse. I truly believe my two year-old daughter, who has already taught herself to use my iPhone, will never seriously use a mouse.

For the record, my now 12 year-old daughter still doesn’t use a mouse, but that is because she has a laptop and uses a touchpad. That was a clear miss by me. A year later, though, when Steve Jobs, in his second-to-last keynote, announced the iPad 2, the future I envisioned looked like it was right on track. The most amazing part of the launch was GarageBand:

This is the entire demo, but the most important part is Steve Jobs reaction to the demo — jump to 12:30 if you don’t have time or inclination to watch the whole thing:

Jobs look of wonderment says more than his words:

I’m blown away with this stuff. Playing your own instruments, or using the smart instruments, anyone can make music now, in something that is this thick and weights 1.3 pounds. It’s unbelievable…this is no toy. This is something you can use for real work.

GarageBand, even more than iWork the year before, was the sort of app that was only possible on an iPad. Sure, it shared a name with its Mac counterpart, but the magic came from the fact that it had little else in common.

And then Jobs died, and I’ve never been able to shake the sense that this particular vision of the iPad died with him.

iPad’s Missing Ecosystem

There was one final part of that GarageBand introduction that, in retrospect, was an inauspicious sign for the future:

GarageBand for iPad's launch price

It’s tempting to dwell on the Jobs point — I really do think the iPad is the product that misses him the most — but the truth is that the long-term sustainable source of innovation on the iPad should have come from 3rd-party developers. Look at Gruber’s example for the Mac of graphic designers and illustrators: while MacPaint showed what was possible, the revolution was led by software from Aldus (PageMaker), Quark (QuarkXPress), and Adobe (Illustrator, Photoshop, Acrobat). By the time the Mac turned 10, Apple was a $2 billion company, while Adobe was worth $1 billion.

There are, needless to say, no companies built on the iPad that are worth anything approaching $1 billion in 2020 dollars, much less in 1994 dollars, even as the total addressable market has exploded, and one big reason is that $4.99 price point. Apple set the standard that highly complex, innovative software that was only possible on the iPad could only ever earn 5 bucks from a customer forever (updates, of course, were free).

This remains one of Apple’s biggest mistakes; in 2015, when Apple first released the iPad Pro, I wrote in From Products to Platforms

When it comes to the iPad Apple’s product development hammer is not enough. Cook described the iPad as “A simple multi-touch piece of glass that instantly transforms into virtually anything that you want it to be”; the transformation of glass is what happens when you open an app. One moment your iPad is a music studio, the next a canvas, the next a spreadsheet, the next a game. The vast majority of these apps, though, are made by 3rd-party developers, which means, by extension, 3rd-party developers are even more important to the success of the iPad than Apple is: Apple provides the glass, developers provide the experience.

That, then, means that Cook’s conclusion that Apple could best improve the iPad by making a new product isn’t quite right: Apple could best improve the iPad by making it a better platform for developers. Specifically, being a great platform for developers is about more than having a well-developed SDK, or an App Store: what is most important is ensuring that said developers have access to sustainable business models that justify building the sort of complicated apps that transform the iPad’s glass into something indispensable.

That simply isn’t the case on iOS. Note carefully the apps that succeed on the iPhone in particular: either the apps are ad-supported (including the social networks that dominate usage) or they are a specific type of game that utilizes in-app purchasing to sell consumables to a relatively small number of digital whales. Neither type of app is appreciably better on an iPad than on an iPhone; given the former’s inferior portability they are in fact worse.

A very small number of apps are better on the iPad though: Paper, the app used to create the illustrations on this blog, is a brilliantly conceived digital whiteboard that unfortunately makes no money; its maker, FiftyThree, derives the majority of its income from selling a physical stylus called the Pencil (now eclipsed in both name and function by Apple’s new stylus). Apple’s apps like Garageband and iMovie are spectacular, but neither has the burden of making money.

The situation has improved slightly since then, primarily with the addition of subscription pricing for apps. Still, that is far inferior from a customer perspective to the previous “Pay for Version 2” model that sustained developers on the Mac for decades; we never did get upgrade pricing or time-limited trial functionality for regular paid apps.

Instead, as Apple is so wont to do, it tried to fix the problem itself, by making the iPad into an inferior Mac. Thus the multi-tasking disaster Gruber decries, which not only is hard-to-use for consumers, but also dramatically ups the difficulty for developers, making the chances of earning a positive return-on-investment for an iPad app even more remote. Indeed, the top two developers making in-depth iPad apps are Microsoft and Adobe, in service to their own subscription models; the tragedy of the iPad is that their successors were never given the space to be born, which ultimately has limited the iPad from truly succeeding the Mac.


To be fair, would that we all could “fail” like the iPad; it was a $21 billion business last fiscal year, nearly as much as the Mac’s $26 billion.1 That, though, is why I did not call it a failure: the tragedy of the iPad is not that it flopped, it is that it never did, and likely never will, reach that potential so clearly seen ten years ago.

  1. This sentence originally included a revenue number that was totally wrong due to a complete brain fart on my end []

Visa, Plaid, Networks, and Jobs

Before the network, there is the job.

In the case of Visa, or, to be more exact, the Bank Americard that would eventually be spun out and renamed Visa, the job for consumers was obvious: instant credit for anything, without the need for a merchant-specific account or a visit to the bank for a personal loan. And so, when Bank of America dropped 60,000 Bank Americards on its customers in Fresno, California, in 1958, they had an immediate reason to give this new-fangled financial product a try.

What may be less obvious is why Fresno’s merchants might have been interested, particularly since Bank of America planned to charge them 6% of sales. Remember, this is before the network: it was not at all obvious, as it is today, that the increase in sales enabled by the convenience of credit cards would more than make up for credit cards’ attendant fees. However, it turned out that for small merchants in particular there was a major job-to-be-done; Joe Nocera explained in his 1994 book, A Piece of the Action:

Fresno’s shop owners knew for a fact that, on the day the program began, some 60,000 people would be holding BankAmericards. That was a powerful number, and it had its intended effect. Merchants began to sign on. Not the big merchants, like Sears, which had its own proprietary credit card and saw the bank’s entry into the credit card business as a form of poaching. Rather, it was the smaller merchants who first came around. Larkin remembers visiting a drug store in Bakersfield, hoping to persuade its owner to accept BankAmericard. “When I explained the concept of our credit card,” he says, “the man almost knelt down and kissed my feet. ‘You’ll be the savior of my business,’ he said. We went into his back office,” Larkin continues. “He had three girls working on Burroughs bookkeeping machines, each handling 1,000 to 1,500 accounts. I looked at the size of the accounts: $4.58. $12.82. And he was sending out monthly bills on these accounts. Then the customers paid him maybe three or four months later. Think of what this man was spending on postage, labor, envelopes, stationery! His accounts receivables were dragging him under.”

A store owner who accepted the credit card was, in effect, handing his back office headaches over to the Bank of America. The bank would guarantee him payment — within days instead of months — and would take over the role of collecting from the customers. As for the bank, in addition to taking its 6 percent cut, the card was a way to get its hooks into businessmen who were not yet Bank of America customers.

It’s easy to forget just how many things a business that takes credit cards does not need to do: it does not need to extend credit, it does not need to collect payment, it does not need to handle excess amounts of cash. It does not, as Nocera noted, need to have much back office functionality at all. Instead banks provide the credit, Visa provides the infrastructure, and merchants pay around 3% of their sales.

Visa’s Network

Not that merchants have much choice in the matter these days. Credit cards are perhaps the best possible example of the power of a multi-sided network; Visa sits in the middle of banks, consumers, and merchants:

The Visa network

Everyone benefits from each other:

Customers and Banks:
– Customers want to have always-available credit
– Banks want to provide credit with high fees and interest rates

Customers and Merchants:
– Customers want to have a card that works everywhere
– Merchants want to be able to accept payments from anyone

Merchants and Banks:
– Merchants want to be able to sell on credit while still getting their money immediately
– Banks want to collect a fee on every purchase in exchange for managing credit and pooling risk

Visa and Mastercard, the other major credit card network,1 sit in the middle of each of these relationships, across billions of customers, millions of merchants, and thousands of banks, collecting a network fee — on top of the interchange fee paid to banks — on every purchase (about 0.05%). The total revenue collected — $20.6 billion in 2018 — is rather small, particularly relative to Visa’s market cap of $420 billion, but that multiple is a testament to just how durable Visa’s position is in the network it created.

Plaid’s Network

There are some obvious parallels to be drawn between Visa’s network, particularly in its earliest days, and Plaid, the fintech startup Visa acquired yesterday. From the Wall Street Journal:

Visa Inc. said Monday it would buy Plaid Inc. for $5.3 billion, as part of an effort by the card giant to tap into consumers’ growing use of financial-technology apps and noncard payments. More consumers over the past decade have been using financial-services apps to manage their savings and spending, and Plaid sits in the middle of those relationships, providing software that gives the apps access to financial accounts. Venmo, PayPal Holdings Inc.’s money-transfer service, is one of privately held Plaid’s biggest customers.

Visa is the largest U.S. card network, handling $3.4 trillion of credit, debit and prepaid-card transactions in the first nine months of 2019, according to the Nilson Report. Its clients are largely comprised of banks that issue credit and debit cards, but the company is looking to expand its presence in the burgeoning field of electronic payments, where trillions of dollars are sent by wire transfer or between bank accounts globally each year.

Plaid has its own three-sided network, but it operates a bit differently than Visa’s:

The Plaid network

The benefits to some parts of the network are more obvious than others:

  • Developers are able to immediately connect to their customers’ bank accounts without having to implement custom integrations with thousands of banks or waiting several days for traditional verification methods (making two deposits of less than a dollar and having the customer report how much).
  • Consumers are able to use new fin tech apps like Venmo immediately, without having to wait several days.
  • Banks…well this is where it gets messy.

Plaid’s Product

To understand how the banks fit in this network it is important to understand what exactly Plaid does. Many banks in the U.S. do not have APIs (Application Programming Interfaces) that offer a programmatic means of accessing a particular account; those that do are not consistent with each other in either implementation or in features. Plaid gets around this by effectively acting as a deputy for consumers: the latter give Plaid their username and password for their bank account, and Plaid utilizes that to basically log in to a bank’s website on the user’s behalf.

If this sounds a bit shady, well, it kind of is! Bank login information is among the most sensitive credentials consumers have, and apparently one in four people in the U.S. with a bank account have shared those credentials with Plaid. Nearly all did so without knowing any better; here, for example, is the interface Betterment offers when you try and add a bank account:

Betterment's add bank account flow

That is not an interface for Chase; it is Plaid,2 effectively training end users to enter their bank credentials in an app and/or on a site that is not their bank! Oh, and because this is very much a hack, Plaid fails between 5 and 10 percent of the time.

Developers, it should be noted, aren’t particularly bothered by this: in fact, they are paying Plaid for every successful log-in. Users, meanwhile, are likely unaware about just how much access and data they are giving away, but at the same time, have a real desire to access new financial services that require a connection to their bank account. The big problem is that the banks aren’t too sure if they want to participate.

The reticence is understandable. There is, for example, the fact that many banks’ technical infrastructure is ancient and built around assumptions that did not include APIs for 3rd-parties. More importantly, though, is the power of inertia: as long as it is hard to move money around, the more likely it is that that money will stay in the bank, collecting minuscule interest; or, if customers need value-added services, the path of lowest resistance will be simply getting them from their bank.

An API-based world could change this dramatically: suddenly consumers could commission robo-advisors to move their cash to whoever is offering the best rates, or to automatically refinance debt. Value-added services from multiple vendors would be equally easy to access, meaning they would have to compete on price or terms. In other words, much like the open Internet, banks fear that profits will be rapidly transformed into consumer benefit.

Jobs for Banks

This is where Visa can potentially make a difference and, by extension, pay for this acquisition. $5.3 billion is very steep — around 50x revenue — and Plaid’s business model is not particularly attractive: the company makes the majority of its money when users connect their bank accounts, which for most applications is a one-time event; contrast that to Visa’s credit card network, which earns a fee on every single transaction.

UPDATE: Plaid makes money every time a user accesses their account for a transaction; for example, every time you pay someone in Venmo. The company says that revenue from these recurring transactions now exceeds revenue earned from bank verification.

What Visa needs to do is figure out what jobs it can do for banks that makes it worthwhile for them to build out the necessary APIs. The most obvious one is security; as a U.S. Treasury Report on Nonbank Financials, Fintech, and Innovation noted:

The practice of using login credentials for screen-scraping poses significant security risks, which have been recognized for nearly two decades. Screen-scraping increases cybersecurity and fraud risks as consumers provide their login credentials to access fintech applications. During outreach meetings with Treasury, there was universal agreement among financial services companies, data aggregators, consumer fintech application providers, consumer advocates, and regulators that the sharing of login credentials constitutes a highly risky practice.

A second job Visa can do is being the devil banks know; that same Treasury report highlighted the fact that the United Kingdom and European Union have initiatives requiring API access to bank accounts, but recommended a private solution. Visa, after this acquisition, is well-placed to leverage Plaid’s widespread use in fintech application and its relationship with banks to come up with a standard that will likely be more favorable to the banks than one imposed by the government.

What is most necessary, though, is selling banks on the idea that they no longer need the equivalent of three people working on bookkeeping machines; hand over those customer service headaches to companies that will specialize in them, over rails Visa will provide.

Visa’s Optionality

This hints at the best case scenario for Visa from this acquisition: a new financial network, with Visa at its center, transforming the consumer financial services industry just as the credit card transformed the consumer retail industry. If that happens, it’s not out of the question that such a network will be so superior to today’s means of moving financial information and data that the company will be able to charge an ongoing toll, instead of simply a set-up fee (and, perhaps, share it with the banks).

The worst case scenario, meanwhile, will see Plaid’s creaky approach deliver barely good enough service to fintech applications in the U.S., with nothing near the reliability or profitability of Visa’s credit card network. Which, from Visa’s perspective, is not a problem either!

Visa will also be able to help Plaid expand internationally, including to more favorable markets like the U.K. and E.U. At first glance, open banking might seem to be a problem for Plaid, but the truth is that screen-scraping is not a long-term solution, and developers will still prefer to use one well-built API that abstracts away thousands of financial institutions instead of re-inventing the integration wheel.

And, most importantly from Visa’s perspective, the credit card business is not going anywhere — if anything, it’s getting stronger. Companies like Stripe are making credit cards more useful in more places, while Apple is making it even easier to use credit cards both online and offline. It is tempting to look at how payments work in countries like China, but that ignores the path dependency of one market using cash until recently, and the other receiving unsolicited Bank Americards 61 years ago. Once a job is done — and credit cards do their jobs very well — it takes a 10x improvement to get users to switch, and, in a three-sided network, that 10x is 10^3.

I wrote a follow-up to this article in this Daily Update.

  1. American Express and Discover, the other major credit card companies, integrate the banking and network components; they still facilitate a network between merchants and consumers []
  2. Well technically Quovo, which Plaid bought last year in a smart move to consolidate its position []

The End of the Beginning

The first American automobile maker, Duryea Motor Wagon Company, was founded in 1895; 34 more auto-makers would be founded in the U.S. in the following five years.1 Then, an explosion: an incredible 233 additional automobile makers were founded in the first decade of the 20th century, and a further 168 between 1910 and 1919. The pace from that point on continued to slow:

New American Car Companies in the 20th Century

On a practical level, that “0” in the 1980’s could be applied to the entire list: by 1920 automobile manufacturing was already dominated by GM, Ford, and Chrysler. AMC, a combination of several smaller brands, was a brief challenger in the 1950s and 1960s, but the “Big Three” mostly had the market to themselves, at least until imports started showing up in the 1970s.

Just because the proliferation of new car companies ground to a halt, though, does not mean that the impact of the car slowed in the slightest: indeed, it was primarily the second half of the century where the true impact of the automobile was felt in everything from the development of suburbs to big box retailers and everything in between. Cars were the foundation of society’s transformation, but not necessarily car companies.

Tech’s Story of Disruption

The story tech most loves to tell about itself is the story of disruption: sure, companies may appear dominant today, but it is only a matter of time until they are usurped by the next wave of startups. And indeed, that is exactly what happened half a century ago: IBM’s mainframe monopoly was suddenly challenged by minicomputers from companies like DEC, Data General, Wang Laboratories, Apollo Computer, and Prime Computers. And then, scarcely a decade later, minicomputers were disrupted by personal computers from companies like MITS, Apple, Commodore, and Tandy.

The most important personal computer, though, came from IBM, with an operating system from Microsoft. The former provided a massive distribution channel that immediately established the IBM PC as the most popular personal computer, particularly in the enterprise; the latter provided the APIs that created a durable two-sided network that made Microsoft the most powerful company in the industry for two decades.

That reality, though, was not permanent: first the Internet shifted the most important application environment from the operating system to the web, and then mobile shifted the most important interaction environment from the desk to the pocket. Suddenly it was Google and Apple that mattered most in the consumer space, while Microsoft refocused on the cloud and a new competitor, Amazon.

Dominance Epochs

Any discussion of dominance in tech touches on three epochs: IBM, Microsoft, and the present day. In this telling, companies like Google and Apple may be dominant now, but so were IBM and Microsoft, and, just as their days of IBM and Microsoft’s dominance passed, so too will today’s companies be eclipsed. Benedict Evans made this argument in a blog post:

The tech industry loves to talk about ‘moats’ around a business – some mechanic of the product or market that forms a fundamental structural barrier to competition, so that just having a better product isn‘t enough to break in. But there are several ways that a moat can stop working. Sometimes the King orders you to fill in the moat and knock down the walls. This is the deus ex machina of state intervention – of anti-trust investigations and trials. But sometimes the river changes course, or the harbour silts up, or someone opens a new pass over the mountains, or the trade routes move, and the castle is still there and still impregnable but slowly stops being important. This is what happened to IBM and Microsoft. The competition isn’t another mainframe company or another PC operating system — it’s something that solves the same underlying user needs in very different ways, or creates new ones that matter more. The web didn’t bridge Microsoft’s moat — it went around, and made it irrelevant. Of course, this isn’t limited to tech — railway and ocean liner companies didn’t make the jump into airlines either. But those companies had a run of a century — IBM and Microsoft each only got 20 years.

None of this is an argument against regulation per se of any specific issue in tech. If a company is abusing dominance today, it is not an argument against intervention to point out that it will lose that dominance in a decade or two — as Keynes says, ‘in the long term we’re all dead’. The same applies to regulation of issues that have little or nothing to do with market dominance, such as privacy (though people sometime fail to understand this distinction). Rather, the problem comes when people claim that somehow these companies are immortal — to say that is to reject all past evidence, and to claim that somehow there will never be another generational change in tech, which seems unwise.

In this understanding of tech dominance, the driver of generational change is a paradigm shift: from mainframes to personal computers, from desktop applications to the web, first on personal computers, and then on the web. Each shift brought a new company to dominance, and when the next shift arrives, so will new companies rise to prominence.

What, though, is the next shift?

Paradigm Shifts

There is an implication in the “generational change is inevitable” argument that paradigm shifts are sui generis. The personal computer was a discrete event, the Internet another, and mobile a third. Now we are simply waiting to see what is next — perhaps augmented reality, or voice assistants.

In fact, I would argue the opposite: the critical paradigm shifts in technology, which drove the generational changes that Evans wrote about, are part of a larger pattern.

Start with the mainframe: the primary interaction model was punched cards; to execute a program you had to insert your cards into a card reader and wait for the computer to read the program into memory, execute it, and give you the results. Computing was done in batches, because the I/O layer was directly linked to the application and data layer.

This explains why personal computers were so revolutionary: instead of one large shared computer for which you had to wait your turn, a user could access their own computer on their own desk whenever they wanted. Still, the personal computer, particularly in a corporate environment, lived alongside not just mainframes but increasingly servers on an intranet. The I/O layer and application and data layers were being pulled apart, but both were destinations: you had to go to your desk and be on the network to compute.

This last point gets at why the cloud and mobile, which are often thought of as two distinct paradigm shifts, are very much connected: the cloud meant applications and data could be accessed from anywhere; mobile made the I/O layer available anywhere. The combination of the two make computing continuous.

The evolution of computing from the mainframe to cloud and mobile

What is notable is that the current environment appears to be the logical endpoint of all of these changes: from batch-processing to continuous computing, from a terminal in a different room to a phone in your pocket, from a tape drive to data centers all over the globe. In this view the personal computer/on-premises server era was simply a stepping stone between two ends of a clearly defined range.

The End of the Beginning

The implication of this view should at this point be obvious, even if it feels a tad bit heretical: there may not be a significant paradigm shift on the horizon, nor the associated generational change that goes with it. And, to the extent there are evolutions, it really does seem like the incumbents have insurmountable advantages: the hyperscalers in the cloud are best placed to handle the torrent of data from the Internet of Things, while new I/O devices like augmented reality, wearables, or voice are natural extensions of the phone.

In other words, today’s cloud and mobile companies — Amazon, Microsoft, Apple, and Google — may very well be the GM, Ford, and Chrysler of the 21st century. The beginning era of technology, where new challengers were started every year, has come to an end; however, that does not mean the impact of technology is somehow diminished: it in fact means the impact is only getting started.

Indeed, this is exactly what we see in consumer startups in particular: few companies are pure “tech” companies seeking to disrupt the dominant cloud and mobile players; rather, they take their presence as an assumption, and seek to transform society in ways that were previously impossible when computing was a destination, not a given. That is exactly what happened with the automobile: its existence stopped being interesting in its own right, while the implications of its existence changed everything.

I wrote a follow-up to this article in this Daily Update.

  1. These numbers are from this Wikipedia article, supplemented with this Wikipedia article; I did not count steam-based automobile makers, motorcycle makers, buggies, or tractor makers []

The 2019 Stratechery Year in Review

2019 was a transition year for me personally, and by extension, Stratechery. The nadir was an Article I regret — The WeWork IPO — where, despite not believing in the company, I wrote the contrarian take because it seemed more interesting (my full mea culpa is here). It was not a healthy approach, but, six years after starting Stratechery, and five years of it being my full-time job, it was perhaps an understandable one.

The turning point was The China Cultural Clash; writing about the crisis surrounding the NBA in China and the implications for the technology industry reminded me of something I had started to forget in my attempt to be even-handed and dispassionate in my analysis: values matter, and there is a freedom that comes from recognizing and articulating those values. Indeed, honesty about values makes analysis better, because underlying assumptions are pushed to the forefront, instead of fading to the background — and, I’d add, it is invigorating! On to 2020!

A drawing of Teams and The Enterprise Growth Framework

This year I wrote 139 Daily Updates (including tomorrow) and 36 Weekly Articles, and, as per tradition, today I summarize the most popular and most important posts of the year.

You can find previous Stratechery Years in Review here: 2018 | 2017 | 2016 | 2015 | 2014 | 2013

A drawing of The Three-Way Market of a Super-Aggregator

Here is the 2019 list.

The Five Most-Viewed Articles

  1. The WeWork IPO — This, to be perfectly frank, is brutal: I not only regret this article for being insufficiently negative (although, for the record, I was clear I would not invest in WeWork), but now also have to face the fact it was my most popular post of the year. Related: Uber Questions, which took the boring skeptical approach to an S-1 I should have repeated.
  2. The Google Squeeze — Google, the real Aggregator, is squeezing OTAs, which acted like Aggregators while depending on Google for demand. It’s easy to say Google is being unfair, but this may be better for consumers.
  3. Shopify and the Power of Platforms — It is all but impossible to beat an Aggregator head-on, as Walmart is trying to do with Amazon. The solution instead is to build a platform like Shopify.
  4. Disney and the Future of TV — TV is moving from a world where distribution dictates business models to one where business models need to fit the jobs consumers want done. That is the best way to understand Disney’s latest announcement.
  5. AWS, MongoDB, and the Economic Realities of Open Source — Amazon’s latest offering highlights the economic challenges facing open source companies — and Amazon should pay attention.
A drawing of The Shopify Ecosystem

Lessons Learned

The good thing about making mistakes is that it is an opportunity to learn; two of these articles are directly connected to WeWork.

  • Neither, and New: Lessons from Uber and Vision Fund — Uber represents something new: a company that is different than incumbents because of technology, yet not itself a tech company — just like the Vision Fund is not a VC.
  • What is a Tech Company? — The question of “What is a tech company” comes down to how much software and its unique characteristics affects the company’s core business.
  • The Value Chain Constraint — Companies succeed or fail not based on technology but rather according to their ability to integrate within their value chains.
A drawing of Amazon's Value Chain

Values and Society

Almost all decisions involve trade-offs; the most difficult are those that require understanding and prioritizing our values.

  • The Internet and the Third Estate — Mark Zuckerberg suggested that social media is a “Fifth Estate”; in fact, social media is a means by which the Third Estate — commoners — can seize political power. Here history matters. Related: Tech and Liberty and the policing of political speech.
  • The China Cultural Clash — The NBA controversy in China highlights a culture clash that both tech companies and the U.S. government need to take to heart. Plus, why Tiktok being Chinese is increasingly a problem. Related: China, Leverage, and Values, about the trade war.
  • Privacy Fundamentalism — The current privacy debate is making things worse by not considering trade-offs, the inherent nature of digital, or the far bigger problems that come with digitizing the offline world.
A drawing of The Synergy Between Tech Companies and Venture Capitalists

Regulation and Antitrust

While regulation was also a theme in 2018, this year I tried to get much more specific about how to think about the challenges presented by the Internet.

  • Where Warren’s Wrong — Senator Warren’s proposal about how to regulate tech is wrong about history, the source of tech giant’s power, and the fundamental nature of technology itself. That doesn’t mean there aren’t real problems — and potential solutions — though.
  • Tech and Antitrust — A review of the potential antitrust cases against Google, Apple, Facebook, and Amazon suggests that only Google is vulnerable.
  • Three Frameworks:
A drawing of The Regulatory Framework for the Internet

The Big Tech Companies

Tech’s continued centralization means that the biggest companies — Microsoft, Apple, Google, Amazon, and Facebook — receive the largest scrutiny.

A drawing of Google's Ambient Computing

Media and Technology

The most important development of the year in media was the launch of Disney+; I already linked to Disney and the Future of TV. Also:

  • Netflix Flexes — Netflix is an Aggregator, with a value chain that lets it drive demand, raise prices, and dismiss competition.
  • Spotify’s Podcast Aggregation Play — Spotify is making a major move into podcasts, where it appears to have clear designs to be the sort of Aggregator it cannot be when it comes to music.
  • The BuzzFeed Lesson — The lesson of BuzzFeed is that dominant Aggregators like Facebook have no incentive to act against their self interest and support suppliers. Related: The Cost of Apple News.
A drawing of The Music Value Chain Versus the Podcast Value Chain

The Year in Daily Updates

This year the Daily Update not only continued the trend towards single topics, but often became the place where new ideas and future Weekly Articles were first presented and fleshed out. I’m really proud of this evolution — this was a hard list to cull. Some of my favorites:

A drawing of Facebook and Amazon's Approach To Take On Apple and Google

I also conducted six interviews for The Daily Update:

Netflix' value chain

I can’t say it enough: I am so grateful to Stratechery’s readers and especially subscribers for making all of these posts possible. I wish all of you a Merry Christmas and Happy New Year, and I’m looking forward to a great 2020!

A Framework for Regulating Competition on the Internet

A prompt for writing this piece is a conference I will be participating in tomorrow entitled Antitrust in Times of Upheaval — A Global Conversation; you can livestream the conference here.

I opened 2017’s Defining Aggregators by stating:

Aggregation Theory describes how platforms (i.e. aggregators) come to dominate the industries in which they compete in a systematic and predictable way. Aggregation Theory should serve as a guidebook for aspiring platform companies, a warning for industries predicated on controlling distribution, and a primer for regulators addressing the inevitable antitrust concerns that are the endgame of Aggregation Theory.

This Article is explicitly related to that piece: unlike most Stratechery items, this Article is not based on a specific news event that happened recently, but is rather an attempt to collect a number of ideas and thoughts I have expressed in different Articles, Daily Updates, and Podcasts about Aggregators, platforms, and regulation. I will link to several of those Articles throughout.

Platform And Aggregator Structure

The most important place to start is by pointing out the excerpt above makes what I now believe is a critical mistake: it conflates platforms and Aggregators. In fact, I believe platforms and Aggregators are fundamentally different entities, and understanding how and why they are different is the single most important task facing would-be regulators.

I explored these differences in 2018’s Tech’s Two Philosophies, The Moat Map, and The Bill Gates Line. This is how I illustrated platforms:

A diagram of a platform

The name “platform” is a descriptive one: it is the foundation on which entire ecosystems are built. The most famous example of a platform — one with which regulators are intimately familiar — is Microsoft Windows. Windows provided an operating system for personal computers, a set of APIs for developers, and a user interface for end users, to the benefit of all three groups: developers could write applications that made personal computers useful to end users, thanks to the Windows platform tying everything together.

What is critical to note about Windows, though — and this extends to newer platforms like iOS and Android — is that it was essential for the ecosystem to function. Developers could not write applications for another operating system if they wanted to reach users, and users could not use a different operating system if they wanted to use popular applications.

Aggregators are different. This is how I illustrate them:

A diagram of an Aggregator

“Aggregator” is also descriptive: Aggregators collect a critical mass of users and leverage access to those users to extract value from suppliers. The best example of an Aggregator is Google. Google offered a genuine technological breakthrough with Google Search that made the abundance of the Internet accessible to users; as more and more users began their Internet sessions with Google, suppliers — in this case websites — competed to make their results more attractive to and better suited to Google, the better to acquire end users from Google, which made Google that much better and more attractive to end users.

Notably, unlike platforms, Google is not essential for either end users or 3rd party websites. There is no “Google API” that makes 3rd party websites functional, and there are alternative search engines or simply the URL bar for users to go directly to 3rd party websites. That Google is so influential and profitable is, first and foremost, because end users continue to prefer it.1

Here is a way to visualize the difference:

  • Platforms facilitate a relationship between users and 3rd-party developers:

    A platform value chain is interdependent

  • Aggregators intermediate the relationship between users and 3rd-party developers:

    An Aggregator intermediates supply and demand

To be clear, both roles can be beneficial — platforms make the relationship between users and 3rd-parties possible, and Aggregators helps users find 3rd-parties in the first place — and both roles can also be abused.

Platform and Aggregator Abuse

The potential impacts on competition by Platforms and Aggregators are broadly similar, differing mostly by degree:

Vertical foreclosure: Platforms can make it impossible for 3rd-parties to function on their platform, either through technological means or, in the case of smartphone platforms, by leveraging their gatekeeper role in terms of App Stores.

Aggregators can also ban 3rd-parties — Google can remove a site from search, or Facebook can remove links from the News Feed — but they cannot force that site to simply not exist. Users can still reach those sites via other search engines, links, or by typing in the URL.

Rent-Seeking: Instead of blocking third-parties, platforms can simply extract money; this often works in conjunction with the threat of foreclosure. Apple, for example, does not allow apps that have their own payment processor, or that even link to a website with payment functionality; they are, however, happy to offer their own payment processor, complete with 30% fee to Apple.

Aggregators, meanwhile, make it increasingly difficult to reach end users without paying for advertising; Google and its expansion of vertical search categories like travel is a perfect example. However, Aggregators do not have as strong of a vertical foreclosure stick as do platforms.

Tying/Bundling: Platforms can include additional functionality or applications that have nothing to do with the platform, but rather leverage the platform to gain market share. The most famous example of this is Windows bundling Internet Explorer, the legality of which was never settled in the United States; the Appeals Court remanded the case to the District Court because of concern a per se application would chill innovation (and in this case Microsoft has been proven right: all operating systems come with browsers now).

This is arguably the true Aggregator stick when it comes to rent-seeking: to return to the travel example, online travel agents may not be happy about paying to be a part of Google Search’s travel module and the hit on margins that entails, but it is better than Google launching its own OTA.

Self-Dealing: Platforms can give their own products advantages, often through special APIs that are not available to competitive products. For example, real-time co-authoring of Microsoft Office documents only works in the Office desktop apps if the documents were opened from OneDrive, but not if opened from Dropbox or Box.

For Aggregators, this advantage is more about putting their competitive product in front of users. Google local results, for example, are not even listed in the Google index, yet they are inserted at the top of the search engine results page (SERP).

The question for regulators is when these abuses should lead to action.

Principles of Regulation

That certain companies have advantages or earn sustainable profits is not a sufficient reason for regulators to act; innovation deserves its reward. Moreover, the presence of abuses like those detailed above is also not necessarily a sufficient reason for regulators to act: regulation can have a chilling effect on innovation, it can be ineffective, and it can incur significant opportunity costs on both companies and regulators. Regulators should focus their attention and resources on abuses for which there is no other recourse than regulation.

This is where the distinction between platforms and Aggregators is critical. Platforms are the most powerful economic and innovation engines in technology: they create the possibility for products that never existed previously, and are the foundation for huge amounts of innovation. It is in the interest of society that there be more and larger platforms, not fewer and smaller.

At the same time, the danger of platform abuse is significantly greater, because users and 3rd-party developers have no other alternative. That means that not only are anticompetitive actions unfair to products that already exist, they also foreclose the creation of an untold number of new products. To that end, regulators should simultaneously encourage the formation of new platforms while ensuring those platforms do not abuse their position.

From a practical standpoint, this means that platforms should have significant latitude in mergers and acquisitions, but significant scrutiny in terms of vertical foreclosure, rent-seeking, bundling, and self-dealing.

Apple is the pre-eminent example here: the iPhone specifically and the entire iPhone ecosystem generally has benefitted tremendously not only from Apple’s internally-created innovations but also from acquisitions like P.A. Semi, which led to the creation of Apple’s A-series of chips. However, the combination of Apple’s total control over 3rd-party app installation and rent-seeking on in-app payments has, in my estimation, stunted innovation and opportunity in the app ecosystem.

Aggregators are different. Yes, they provide value to end users and to third-parties, at least for a time, but the incentives are warped from the beginning: 3rd-parties are not actually incentivized to serve users well, but rather to make the Aggregator happy. The implication from a societal perspective is that the economic impact of an Aggregator is much more self-contained than a platform, which means there is correspondingly less of a concern about limiting Aggregator growth.

For the same reason, though, Aggregators are less of a problem. Third parties can — and should! — go around Aggregators to connect to consumers directly; the presence of an Aggregator is just as likely to spur innovation on the part of a third party in an attempt to attract consumers without having to pay an Aggregator for access. Moreover, there is a Sisyphean aspect to regulating power predicated on consumer choice: look no further than the European Union, where regulators are frustrated that remedies for the Google shopping case aren’t working, even though those same regulators were happy with the remedies in theory; the problem was trying to regulate consumer choice in the first place.

It follows, then, that regulatory priorities should be the opposite of platforms: given that Aggregator power comes from controlling demand, regulators should look at the acquisition of other potential Aggregators with extreme skepticism. At the same time, whatever an Aggregator chooses to do on its own site or app is less important, because users and third parties can always go elsewhere, and if they don’t, that is because they are satisfied.

Here Facebook is a useful example: the company’s competitive position would be considerably shakier — and the consumer ad-supported ecosystem considerably healthier — if it had not acquired Instagram and WhatsApp, two other consumer-facing apps. At the same time, Facebook’s specific policies around what does or does not appear on its apps, or how it organizes its feed, has no reason to be a regulatory concern; I would argue the same thing when it comes to Google’s search results.

There is one additional area where regulators should focus: advertising. Advertising is a core component of Super-Aggregators like Facebook and Google because the incentives are perfectly aligned: Facebook and Google want to serve everyone, which means they want to be free, while advertisers want to have access to everyone, which means coalescing around the largest Aggregators.

First, this results in a type of market failure when it comes to problematic content, as I detailed in A Framework for Regulating Content on the Internet. Second, the sheer scale of the core Aggregator gives a massive scale advantage that can be applied elsewhere; Google in particular is much more of an inescapable platform when it comes to advertising on third party sites; this is where regulatory scrutiny should be focused.


There is one final component to this analysis: if platforms and Aggregators are to be treated differently, regulators need a more flexible way of considering when is the correct time to step in.

Consider the three regulatory issues that I implicitly suggested deserve more attention in this piece: Apple’s App Store policies, Facebook’s acquisitions, and Google’s third-party advertising offerings. None of them fit under a popular conception of a monopoly: Apple sells a minority of smart phones, Facebook acquired Instagram when it had only 30 million users, and the advertising market is both not consumer-facing and has infinite supply.

That doesn’t mean harms don’t exist, though: the apps and services that aren’t created, the advertising-based consumer services that are under-monetized (Snapchat and Twitter) or that aren’t even being funded, and the multitude of websites that can’t realistically even try innovations that entail going around Google.

That, though, is why I am writing this piece. Cases like the European Commission Google Shopping case are excellent examples of how a lack of clear standards lead to sub-optimal outcomes that don’t actually change anything; at the same time, the European Commission’s investigation of Apple’s App Store will surely benefit from increased flexibility in defining relative markets.

Ideally, there would be stricter adherence to better rules, instead of finger-crossing that Brussels gets it right. That, though, likely requires new laws. Indeed, while that seems like a slog, it should, in my estimation, be the focus of those interested in a future where we direct tech’s innovation towards making a larger pie for everyone, as opposed to cutting off slices because it makes us feel better, even if only temporarily.

I wrote a follow-up to this article in this Daily Update.

  1. Google also, to be clear, pays significant amounts of money to ensure it is the default on iOS, and basically built Android to ensure it is the default everywhere else. The European Commission correctly found Google’s contractual moves to secure its position on Android illegal []

Portability and Interoperability

In Facebook CEO Mark Zuckerberg’s March 2019 op-ed in the Washington Post calling for federal regulation of technology, he included something that caught some observers by surprise:

Regulation should guarantee the principle of data portability. If you share data with one service, you should be able to move it to another. This gives people choice and enables developers to innovate and compete.

This is important for the Internet — and for creating services people want. It’s why we built our development platform. True data portability should look more like the way people use our platform to sign into an app than the existing ways you can download an archive of your information. But this requires clear rules about who’s responsible for protecting information when it moves between services.

This isn’t just talk: on Monday Facebook announced a new photo transfer tool. From the Facebook blog:

At Facebook, we believe that if you share data with one service, you should be able to move it to another. That’s the principle of data portability, which gives people control and choice while also encouraging innovation. Today, we’re releasing a tool1 that will enable Facebook users to transfer their Facebook photos and videos directly to other services, starting with Google Photos…

For almost a decade, we’ve enabled people to download their information from Facebook. The photo transfer tool we’re starting to roll out today is based on code developed through our participation in the open-source Data Transfer Project and will first be available to people in Ireland, with worldwide availability planned for the first half of 2020. People can access this new tool in Facebook settings within Your Facebook Information, the same place where you can download your information. We’ve kept privacy and security as top priorities, so all data transferred will be encrypted and people will be asked to enter their password before a transfer is initiated.

This initiative also helps satisfy Facebook’s requirements under Europe’s General Data Protection Regulation. From Article 20:

The data subject shall have the right to receive the personal data concerning him or her, which he or she has provided to a controller, in a structured, commonly used and machine-readable format and have the right to transmit those data to another controller without hindrance from the controller to which the personal data have been provided.

So all is well that ends well, right? Facebook follows the law in Europe and goes above and beyond in the United States, surely leading to new innovation and competition!

As you might suspect, I’m skeptical.

Data That Matters

Start with the obvious objection: why would Facebook, or the other companies that are a part of the Data Transfer Project (including Apple and Google) wish to increase competition? It seems reasonable to assume they would not.

It follows, then, that the data that is being made portable — in this case images and videos — must be a complement to Facebook’s core service. After all, making it easier to give that data away devalues it, and companies always seek to commoditize their complements.

The question that comes next is complement to what? For Facebook, the answer is easy: their social graph. Who you are friends with is the data that is much more valuable, and Facebook is not about to launch a network transfer tool.

There is plenty of evidence that this is the case. Back in the days of Facebook’s Open Graph initiative — which is at the root of controversy surrounding Cambridge Analytica — Facebook was giving away all of the data developers might want, the better to get developers on the Facebook platform. The company drew the line, though, when it came to other social networks.

After this crackdown Facebook “clarified” its position in a blog post:

For the vast majority of developers building social apps and games, keep doing what you’re doing. Our goal is to provide a platform that gives people an easy way to login to your apps, create personalized and social experiences, and easily share what they’re doing in your apps with people on Facebook. This is how our platform has been used by the most popular categories of apps, such as games, music, fitness, news and general lifestyle apps.

For a much smaller number of apps that are using Facebook to either replicate our functionality or bootstrap their growth in a way that creates little value for people on Facebook, such as not providing users an easy way to share back to Facebook, we’ve had policies against this that we are further clarifying today.

This is about as concise a distillation of the “commoditize your complements” approach as you will see, at least as far as data is concerned: if you make Facebook better, you can have it all; if you don’t, or are remotely competitive, you are cut off.

The Privacy Angle

Facebook, for obvious reasons, has come to regret the entire Open Graph 1.0 era, in large part because of attention paid to privacy issues. In fact, the company had started restricting the data it shared with the release of Graph 2.0 in 2014; now 3rd-party developers could only see a user’s friends if those friends also used the same app, much like the Twitter Facebook app of old.

The restrictions in GDPR are even tighter; the last part of Section 20, providing for data portability, states:

The right referred to in paragraph 1 [excerpted above] shall not adversely affect the rights and freedoms of others.

In other words, you can get your personal data out, but no data about your friends, because they didn’t give permission. This does, in a privacy context, make perfect sense. At the same time, it is ground zero for how privacy regulation can often be at odds with encouraging competition: it’s all well and good to get your old photos and videos, but its telling that the most likely first place to put those is in a photo storage app that serves a very different purpose than Facebook; a Facebook competitor would be better served with access to a user’s list of friends.

The Interoperability Contrast

A far more impactful outcome would be if Facebook’s friend data were interoperable. Suppose you created a new app that could, once you authorized yourself, incorporate access to Facebook’s graph in a way that let you connect with friends that also use the app, kind of like the Twitter Facebook app of old.


Update 12/12/19: In fact, Facebook does allow this with the User Friends API, and a spokesperson assured me that Twitter or Snapchat or any other social network is, after a policy change last year, free to implement said API. Of course, that entails using the Facebook Login, and all of the data sharing that follows. It’s also not nearly as compelling as being able to recruit friends to the new app in the first place (which is what I should have focused on). Regardless, I didn’t have this quite right, so consider this a correction.


In this model, the 3rd party developer doesn’t actually get data from Facebook. Facebook, rather, exposes its data in a way that the user can leverage the company’s social graph to bootstrap their experience. This both significantly increases the potential for competition while also leaving the user in charge, not only of their own data but also the data about who their friends are.

The problem with this approach is obvious: Facebook would have to implement it, and it has zero reasons to do so, both because of competitive reasons, and also because regulatory zeal for privacy gives the company cover to not give out any friend data at all. The reason to write this Article, though, is to show why data portability like the sort Facebook announced is such a red herring: it has the trappings of increasing competition, the better to avoid antitrust regulation, but it doesn’t really do anything of the sort, particularly relative to far more impactful interoperability.

Interoperability and the Tech Giants

This idea of comparing and contrasting portability with interoperability is another lens to understand what are the commoditizable complements versus highly differentiated core of the largest tech companies.

Consider Google, another frequent target of regulators. The company has no problem giving you your data, or letting you wipe it out. What the company won’t do is make the Search Engine Results Page (SERP) interoperable. Interoperability would mean 3rd-parties being able to populate some portion of the results, or being able to use the results while providing their own ads; neither is happening anytime soon.

In the case of Amazon, interoperability would mean making the company’s logistics service available as a service to any merchant selling through any storefront; in reality, it is only available to merchants selling on Amazon.com. [This is incorrect. Amazon actually offers exactly this. My apologies for the error.]

For Apple, interoperability could happen at two levels: there could be a way to install apps on your iPhone independent of the App Store, or the App Store could allow apps that incorporate their own payment processors (or simply link to a web page to complete a purchase).

I am not arguing, at least in this piece, that any of these should happen. The only one I feel strongly about is Apple, simply because there is no alternative for developers or customers to the App Store (there are other ways to sell to or reach end users than Amazon or Google, and other ways to find your friends than use Facebook). What is notable, though, is how interoperability for all of these companies cuts to the core of how they extract profit from their respective value chains.

Portability AND Interoperability

To be very clear, I’m pretty excited about Facebook’s announcement. Data portability is absolutely consumer friendly, and I’m glad that Facebook is making it easy to move photos and videos that have been lost to time to applications that are better suited for long-term storage.

At the same time, we shouldn’t kid ourselves that this has any sort of impact on competition. It is interoperability that cuts to the core of these companies’ moats, and to the extent regulators see it worthwhile to act, interoperability should be the priority.

  1. This link only works in Ireland []

Integration and Monopoly

When I started Stratechery in 2013, the conventional wisdom was that modularized ecosystems were best. After all, Microsoft had just spent the last thirty years dominating the tech industry by dominating the operating system layer and benefiting from competition everywhere else in the stack. Sure, the company’s dominance was slipping, but not because the model didn’t work: rather, the inevitable winner of the smartphone era would be Google, whose Android operating system was following the Windows playbook, with the rather attractive addition of being free.

That obviously didn’t happen: not only did Apple establish and maintain a sufficiently large install base to support an iOS ecosystem, the company went on to take the vast majority of profits in the entire smartphone industry; the exact share is uncertain, thanks to inconsistent reporting across the industry, but most estimates put Apple’s smartphone profit share between 70~90% for the last five years.

Apple, famously, is integrated, at least as far as the operating system and the hardware is concerned, and it turned out said integration was not only not a liability, it was actually a tremendous advantage.

The Benefits of Integration

I have written about Apple’s integration multiple times over the years, so rather than repeat myself allow me to highlight three advantages using three Articles.

First, integration provides for a superior user experience. From What Clayton Christensen Got Wrong:

The issue I have with this analysis of vertical integration — and this is exactly what I was taught at business school — is that the only considered costs are financial. But there are other, more difficult to quantify costs. Modularization incurs costs in the design and experience of using products that cannot be overcome, yet cannot be measured. Business buyers — and the analysts who study them — simply ignore them, but consumers don’t. Some consumers inherently know and value quality, look-and-feel, and attention to detail, and are willing to pay a premium that far exceeds the financial costs of being vertically integrated…

Not all consumers value — or can afford — what Apple has to offer. A large majority, in fact. But the idea that Apple is going to start losing consumers because Android is “good enough” and cheaper to boot flies in the face of consumer behavior in every other market…Apple is — and, for at least the last 15 years, has been — focused exactly on the blind spot in the theory of low-end disruption: differentiation based on design which, while it can’t be measured, can certainly be felt by consumers who are both buyers and users.

Second, integration maximizes the likelihood of success for new products — including the iPhone itself. From How Apple Creates Leverage, and the Future of Apple Pay:

The carriers [before the iPhone]…largely offered the same service: voice, SMS, and data, all of which was interoperable. This increased elasticity of substitution gave Apple an opportunity to pursue a divide-and-conquer strategy: they just needed one carrier.

Apple reportedly started iPhone negotiations with Verizon, but it turned out that Verizon was already kicking AT&T’s (then Cingular’s) butt through aggressive investment and technology choices, resulting in increasing subscriber numbers largely at AT&T’s expense. Verizon saw no need to change their strategy, which included strong branding and total control over the experience on phones on their network. AT&T, meanwhile, was on the opposite side of the coin: they were losing, and that in turn had a significant effect on their BATNA — they were a lot more willing to compromise when it came to branding and the user experience, and so the iPhone launched on AT&T to Apple’s specifications.

That is when Apple’s user experience advantage and corresponding customer loyalty took over: for the first time ever customers were willing to endure the hassle and expense of changing phone carriers just so they could have access to a specific device. Over the next several years Verizon began to bleed customers to AT&T even though their service levels were not only better, but actually widening the gap thanks to the iPhone’s impact on AT&T. Four years after launch the iPhone did finally arrive on Verizon with the same lack of carrier branding and control over the user experience; in other words, Verizon eventually accepted the exact same deal they rejected in 2006 because the loyalty of Apple customers gave them no choice…

Third, integration is incredibly profitable because it is, from a money-making perspective, a monopoly: Apple devices are the only ones that run iOS. From Everything as a Service:

Apple has arguably perfected the manufacturing model: most of the company’s corporate employees are employed in California in the design and marketing of iconic devices that are created in Chinese factories built and run to Apple’s exacting standards (including a substantial number of employees on site), and then transported all over the world to consumers eager for best-in-class smartphones, tablets, computers, and smartwatches.

What makes this model so effective — and so profitable — is that Apple has differentiated its otherwise commoditizable hardware with software. Software is a completely new type of good in that it is both infinitely differentiable yet infinitely copyable; this means that any piece of software is both completely unique yet has unlimited supply, leading to a theoretical price of $0. However, by combining the differentiable qualities of software with hardware that requires real assets and commodities to manufacture, Apple is able to charge an incredible premium for its products.

The results speak for themselves: this past “down” quarter saw Apple rake in $50.6 billion in revenue and $10.5 billion in profit. Over the last nine years the iPhone alone has generated $600 billion in revenue and nearly $250 billion in gross profit. It is probably the most valuable — the “best”, at least from a business perspective — manufactured product of all time.

Now, five years on, the conventional wisdom has flipped: integration is clearly the best. Just look at Apple! Indeed, looking at Apple, it is hard to escape that conclusion, but it is worth pointing out that the last week has highlighted a number of potential downsides.

The Keyboard Kerfuffle

Last week Apple did something that was shockingly momentous: it released a new laptop with what appears to be a user-friendly keyboard, both in terms of how it works and how often it works. I say shockingly because it is pretty incredible that such a should-be-mundane feature was momentous.

And yet, here we are: because the entire MacBook line does not yet have the new keyboard, Apple still hosts this support article that suggests cleaning your laptop’s keyboard with compressed air.

How to clean your MacBook with compressed air

This is, needless to say, not normal, nor are keys failing on multiple computers for multiple generations of computers.

It’s that last bit — the multiple generations bit — that is of interest here. Apple first released its notorious butterfly keyboard in April 2015, and has only now replaced it in one model in November 2019. Over that time period the company has sold $99 billion worth of Macs, the majority of which have been laptops. This is truly the power of integration!

Or, to put it another way, the power — and downside — of monopoly. No, Apple does not have a monopoly in computers — how amazing would that be! — but the company does have a monopoly on macOS. It sells the only hardware that runs macOS, which is why millions of customers kept buying computers that, particularly in the last couple of years, were widely reported to be at risk of significant problems.

To be clear, Apple didn’t commit some sort of crime here. At the same time, it is hard to imagine the butterfly keyboard persisting for four-and-a-half years and counting if the company faced any sort of competition. Integration can produce a superior user experience, but once an integrated product faces no more competition it can result in something that is downright user-hostile.

NFC and Innovation

The second story comes from Germany. From The Verge:

Apple could be forced to allow rival payment services on iOS to compete with its own Apple Pay service in Germany after the country’s parliament voted in favor of the measures on Thursday, Zeit Online reports. The legislation came in the form of an amendment attached to an anti-money laundering bill, and it will need to pass through the country’s upper house before it can become law next year.

If these measures become law, then, in Germany, Apple could be forced to allow other companies to access its phone’s NFC chips, which it has historically tightly controlled access to. Zeit Online notes that the change could result in individual banks offering NFC payments through their own apps, rather than going through Apple’s service. Apple would reportedly be allowed to charge a fee for access to the NFC chip, but it wouldn’t get the reported 0.15 percent fee that it currently gets from each Apple Pay transaction.

There is absolutely a competition component to Apple Pay and Apple’s restriction on NFC. Because of its control of iPhones generally and their built-in NFC chips specifically, Apple is able to give Apple Pay a significant advantage relative to competing payment apps (which need to use clumsy QR codes); that means that Apple can leverage its position in smartphones into a strong position in payments.

What is worth highlighting though, particularly in the context of this article, is how integration can hamper innovation.

To back up, NFC stands for “Near-Field Communication”, which is a protocol for two electronic devices to communicate when they are within 1.5 inches (4 centimeters) of each other. There are three use cases for NFC chips on smartphones:

  • Smart card emulation, where NFC devices act like payment cards; Apple Pay is an example of this, as are a host of other use cases, like transit tickets or smart keys.
  • Reader/Writer, where one active NFC devices reads or writes to a passive NFC device, such as an NFC sticker that receives power from the magnetic field generated by the active device.
  • Peer-to-peer, where two NFC devices exchange information on an ad-hoc basis.

In short, NFC makes it possible for two devices to communicate without any prior setup, making possible a range of use cases far greater than, say, Bluetooth…and yet the only NFC technology most of you have probably used is for payments. Why?

I think that Apple deserves a lot of the blame. While Android devices have had NFC chips since 2010, Apple only added them to iPhones with 2014’s iPhone, and it was limited to Apple Pay. Two years later Apple made it possible to read some NFC tags and use Apple Pay, and only two months ago made it possible to write NFC tags. No other payment solution can use Apple Pay, only five transit systems can use Apple Pay with an “Express Transit” mode, and only 11 more with time-of-purchase authentication.

The problem is that the NFC chip on iPhones is not open: it is integrated with iOS, and Apple is holding the reins tight. Given the company’s 0.15% skim of Apple Pay transactions, and previous attempts to charge 3rd parties for integrating into its ecosystem or building accessories, it’s fair to wonder how much financial considerations weigh in NFC’s sluggish roll-out. What seems unquestionable, though, is that innovation in the entire sector has been retarded because of Apple’s total control of NFC chips in iPhones.

App Store Control

The final story is from this weekend; from the Washington Post:

Apple removed all vaping-related apps from its App Store on Friday, siding with experts who call vaping “a public health crisis” and “a youth epidemic.” Some of the 181 vaping apps removed by Apple permit the user to control the temperature or other settings on vaping devices. Others offer users access to social networks or games. The App Store has never permitted the sale of vaping cartridges through apps.

“We’re constantly evaluating apps, and consulting the latest evidence to determine risks to users’ health and well-being,” Apple spokesman Fred Sainz said in a statement. Apple cited evidence from the Centers for Disease Control and Prevention and other groups that have linked vaping and e-cigarette usage to deaths and lung injuries.

It’s certainly tempting to cheer a decision like this, particularly given the health crisis that emerged around vaping this year, and the more widespread concerns about vaping being an on-ramp to tobacco use. Then again, given that the crisis seems centered on counterfeit cartridges, being able to connect to your smartphone could be a real benefit. Note this paragraph written by a medical marijuana user:

But there are also more sophisticated devices that have USB and even Bluetooth interfaces to enable the patient to control heat settings, display lights, and update the firmware. The Bluetooth devices are accompanied by apps on the iOS and Android mobile platforms which can allow the patient to measure and monitor their usage, and, as is the case with PAX to identify the medication loaded into the device, and to understand its contents, such as the overall cannabinoid profile, the terpene mix, and other components. It also allows a user to validate the authenticity of the medication as well as testing and batch results.

Those apps — and by extension, device functionality — are no longer available to iPhone users1 — you can’t get this level of functionality in a browser — not because regulators ruled them illegal, or because Congress passed a law, but because a group of technology executives said so. And, what they said held sway because the App Store is integrated with the iPhone: Apple has a monopoly on what apps can or cannot be installed.

To be fair, you may not find a vape app ban problematic, no matter how concerning the principles at stake; how about the company banning an app that shows where protestors and police are clashing in Hong Kong, or an app that tracks drone strikes? In both cases you can argue that Apple is simply abiding by the standards of the countries in which it operates, but the core reason why there is even a question about app removal is because of Apple’s control.

Apple’s approach to the App Store also raises both competition and innovation questions. With regards to the former Apple is leveraging its control of the App Store approval process into a tax on digital goods and/or an advantage for its own competing services; when it comes to the latter Apple’s restrictions on developer business models (which has improved a bit since that article, but is still lacking, particularly in terms of standalone trials and upgrades) has made it difficult for rich productivity apps in particular to emerge on iOS, and often that 30% is the line between being viable or not, particularly for licensed material.

Make no mistake, Apple’s close control of the App Store has had tremendous benefits, not simply for Apple and its bottom line, but also for developers, particularly by convincing customers scarred from the Windows malware debacle that it was safe to download and pay for apps. But that control — borne from innovation — has had significant costs as well.

Integration Versus Monopoly

This article is not a legal argument: in particular, I have used the term “monopoly” very loosely. What makes Apple so brilliant from a business perspective is that it has managed to, via hardware and software integration, earn monopoly profits in a way that would not normally be classified as a monopoly.2 Still, it seems to me that while “integration” results in good outcomes, “monopoly” doesn’t: note the contrast between the advantages of integration I began with and the bad outcomes of late:

  • The superior user experience of Apple’s integrated products somehow ended up with Apple delivering the user-hostile butterfly keyboard for four years and counting.
  • Apple’s ability to leverage its user base to bring new products and features to market also meant that Apple could retard the development of NFC applications.
  • Apple’s ability to drive superior profits from software-differentiated hardware is increasingly augmented by the attempt to extract rents on digital goods and/or give the company’s own services a competitive advantage.

The issue in all cases is a familiar one in technological markets, which often start in an ultra-competitive state, but quickly devolve into monopolies or duopolies as things like network effects and economies of scale takeover. We have seen similar progressions in operating systems, in search, in social networks, in digital advertising, in e-commerce — Apple pressing its advantages is hardly an exception!

The reason I find the Apple example particularly illustrative, though, is that it helps draw a line between the sort of healthy integration that is broadly beneficial, and monopoly rent-seeking that mostly goes to the dominant companies’ respective bottom lines.

Specifically, companies that create and or compete in new markets should be allowed to win, and reap the benefits of their innovation. I have no issue with Apple’s smartphone profits, or Apple Pay’s 0.15% or the App Store’s 30%.

What should be restricted, though, is leveraging a win in one area into dominance in another: that means Apple winning in smartphones should not mean it gets to own digital payments, and inventing the App Store does not mean it gets 30% of all digital goods (or be allowed to diminish the user experience of its competitors). Apple Pay and App Store payment processing should win because they are better — which can include being the default! — not because they are a point of integration that has curdled into monopoly-type behavior that results in worse outcomes for everyone.

I wrote a follow-up to this article in this Daily Update.

  1. To be clear, existing apps will continue to work for now, including on new iPhones, but are not available to users who have not previously downloaded them; it’s also not clear if those apps can ever be updated to support new devices []
  2. That noted, it seems likely the European Commission is going to classify Apple as a monopoly provider of iOS devices []

The Google Squeeze

In 3Q 2014 Google had $16.5 billion in revenue and $2.8 billion in profit. I proceeded to write an article entitled Peak Google. Fast forward to last quarter, and Google had $36 billion in revenue and $6.7 billion in profit, increases of 118% and 139% respectively. It is difficult to imagine being more wrong!

For the record, my thesis was not that Google’s revenue and profit growth were over; rather, like Microsoft in the 2000s, Google would continue to grow but that its relevance had peaked, in large part because brand marketing would become much more important on the web.

Frankly, this explanation makes things worse in a way: it is certainly true that some types of advertising have, as I predicted, worked much better on platforms like Facebook or Instagram (and it’s also true that Facebook gave up on competing for advertising on 3rd-party sites); it is also worth noting that much of that advertising is less traditional brand advertising, meant to increase brand affinity for future conversions, than it is demand-generating direct advertising (as opposed to Google’s demand-capturing direct advertising in Search). What truly misses the mark, though, is the suggestion that Google’s relevance has in any way decreased.

Five Years of Growth

I have owned up to getting the Peak Google article wrong in the Daily Update, particularly this post in 2017, but in the interest of accountability — and, naturally, this article — a quick review is in order.

First, I should have been clear from the get-go that the analysis did not apply to YouTube. Not only is YouTube a natural fit for brand advertising, which is traditionally video-based, it was also barely monetized at that point; clearly significant growth was coming from that property alone.1

Second, despite the fact I spent most of the early years of Stratechery writing about mobile and the extent to which its impact was underrated, particularly by those in the United States who had already adopted personal computers, I underrated mobile’s impact! First, mobile dramatically increased the number of users Google served in both developed and developing countries. Second, mobile dramatically increased usage from existing users, as the Internet was now in people’s pockets or purses, not only their desks or backpacks. Google’s market was in the process of getting much larger.

The biggest mistake, though, was in underestimating just how far Google could go in terms of showing users more ads, even once you accounted for more users using Google more often.

MOAR Ads

The first and most obvious way that Google showed users more ads was by literally inserting more ads into mobile search results. This was a development I tracked closely, wondering just how far the company would go as it added first a third ad, and then a fourth. I wrote in that Daily Update:

Admittedly, this observation has been largely critical: is Google’s revenue growth due to actual increased engagement or simply due to stuffing the screen with ever more ads? In fact, the appropriate answer is “Who cares?” I suspect my disproving attitude stemmed from Peak Google: the company continued to defy my general narrative, so surely the cause must be something untoward like a modern paid inclusion model. And, well, that’s kind-of-sort-of what it is…

That certainly may be vaguely distasteful — it was much more edifying for Google to argue, as they did a decade ago, that folks simply using the Internet meant they made more money — but that doesn’t mean it isn’t effective, and frankly, I haven’t given it enough credit.

Just as important, though, is the way in which Google responded to the threat posed by vertical search alternatives. Back when I wrote Peak Google, there was a lot of talk that mobile was a problem for Google because the new app paradigm would make it more likely that end users would go around Google. They would use Yelp for local search, Amazon for shopping, or Expedia for travel; sure, URLs and bookmark managers may have been too confusing for most users, but the “App Revolution” meant a vertical search engine was only a tap away.

In fact, though, this threat ended up being overstated for several reasons.

First, it turned out that users didn’t want to juggle multiple apps any more than they wanted to juggle multiple URLs. Search in the built-in browser was still the easiest and most obvious place to start.

Second, Google was willing to pay whatever was necessary to ensure it was the default search engine for those built-in browsers, including billions of dollars on an ongoing basis to Apple.

Third, Google started to transform mobile results in particular to be much more useful: instead of forcing users to click a link for an answer — something mobile users dislike about as much as downloading new apps — Google would give it to them; they didn’t even need to “feel lucky”. Most importantly, though, when it came to vertical search categories, Google would offer an entirely new kind of results page.

Expedia and TripAdvisor

This reason to discuss this topic now is because of disappointing earnings results from Expedia and TripAdvisor. From CNBC:

Shares of Expedia and TripAdvisor both reached new year-to-date lows during midday trading on Thursday, tumbling as much as 25%. The stock plunge comes after both the travel service stocks reported third-quarter earnings misses after the bell Wednesday. Both companies pointed to weakened visibility in Google search results as a long-term revenue headwind.

Expedia CEO Mark Okerstrom said on Expedia’s earnings call:

What we saw was a continued shift of essentially the free links further down the page, by other modules that were inserted and ultimately a shift of traffic from the SEO channel over to some of the other products whether it’s flight metasearch or hotel metasearch over time. Now of course as related to the hotel product, the lodging product, we are able to pick up some of that volume and that ultimately resulted in spending more on sales and marketing than we had otherwise would have. We are happy with the returns that we saw on it, but ultimately, not as good returns as we would see from the SEO channel.

TripAdvisor CEO Steve Kaufer said on TripAdvisor’s earnings call:

We did see some incremental SEO headwinds over the course of the quarter. It’s always hard to know exactly what Google is doing. We think of it as how far down the page are we is our organic result. And I think you’re seeing this across the industry as Google has gotten more aggressive. We’ve been predicting this of course for the past many years. We talked about it on our last call. We know that this SEO piece is an ongoing trend and we’re not predicting that it’s going to turn around.

Expedia and TripAdvisor used to be the same company; they play in closely related spaces. Expedia is an “OTA” — Online Travel Agent — where you can book hotels, plane tickets, etc.; TripAdvisor is focused on reviews but monetizes as a meta-search engine, i.e. by referring users to OTAs (although TripAdvisor, with its “instant booking” product, is not far from being an OTA itself).

OTAs and Aggregation Theory

OTAs have always been a special case when it comes to Aggregation Theory; like Aggregators, they serve customers on a zero marginal cost basis, and they have power over supply (hotels, primarily) by virtue of delivering them demand. The hangup for me is how they acquire that demand: first and foremost from Google.

Expedia’s Google play is straightforward: deliver highly-ranked answers to common queries like “Tickets to Tokyo” or “Hotels in Sydney”, and also become very good at buying search ads. TripAdvisor, meanwhile, leverages its reviews to rank highly on a whole host of terms related to traveling, and then offers booking functionality alongside those reviews.

What is notable in both cases, though, is that it is Google that ultimately owns the customer relationship, which is why I have always hesitated to call OTAs Aggregators:

Google owns demand

This arrangement between OTAs and Google has long been beneficial to both sides. Google drives traffic to the OTAs, which can monetize that traffic via commissions extracted from suppliers.2 Google, meanwhile, not only receives relevant results it could serve to customers, but also makes billions of dollars from OTAs buying search ads.

Google and the OTAs value chain

What has changed, starting with Google’s search results, which then spilled over to these companies’ financial results, is the hotel module:

Google hotel search results

First, note just how many screens you now have to scroll to reach organic results — at least 3 on an iPhone 11 Pro which is 812 points tall.3 Again, this isn’t necessarily new — Google has been adding ads for a while — but what makes the hotel module compelling is that, while it is easy to ignore the ads, the module is genuinely useful! You have a map of the city with prices of various hotels, an opportunity to specify your dates, and several options to click through on.

Here is the rub, though, at least from an OTA perspective:

Google Partners

In case you’re not sure what “Google Partners” means, here is a screen you get when you click on one of those hotels and look at the prices:

Google hotel module listings are ads

Everything in the hotel module is an ad, or perhaps more accurately, paid-inclusion. This particular example does not have Expedia.com (but it does have Hotels.com, which Expedia owns) or TripAdvisor, but they are in others; it is this module that Okerstrom was referring to when he said:

Now of course as related to the hotel product, the lodging product, we are able to pick up some of that volume and that ultimately resulted in spending more on sales and marketing than we had otherwise would have. We are happy with the returns that we saw on it, but ultimately, not as good returns as we would see from the SEO channel.

I would think not; the SEO channel is free, the hotel module isn’t.

Aggregating OTAs

At this point the conclusion seems easy, no? Google being evil, yet again. In fact, while I understand the frustration of Expedia and TripAdvisor, I think it is a bit more complicated.

Start with the theoretical perspective: the stable structure of an Aggregator-dominated market is that the Aggregator controls demand and suppliers come onto the Aggregator on the Aggregator’s terms. In other words, there are three players in the value chain: suppliers—Aggregator—demand. Notably, though, that has not been the case in travel, where Google has controlled demand but OTAs have controlled supply.

One way to achieve equilibrium would be for Google to become the one OTA to rule them all. Indeed, this would be difficult to compete with (and was a fear when Google acquired ITA in 2010). The truth, though, is that OTAs have put in significant effort to bring suppliers on board, and they deal with all of the pesky payment and customer support issues that Google loves to eschew. Instead Google has realized it can get OTAs to effectively pay Google to take care of the messy parts for them.

The Hotel Module in the Value Chain

With the hotel module, Google captures demand more efficiently, which not only makes Google search more attractive to end users, but also transforms OTAs into suppliers, paying to provide the service that Google doesn’t want to. It is a textbook example of what Tren Griffin calls Wholesale Transfer Pricing:

Wholesale transfer pricing = the bargaining power of company A that supplies a unique product XYZ to Company B which may enable company A to take the profits of company B by increasing the wholesale price of XYZ.

In this case the unique product is demand — users. And this is where I am tempted to defend Google: at the end of the day, the company has the dominant position in its value chain largely by providing a better product. Search was better to start, but Google didn’t rest on its laurels: it made search better on mobile in particular with these sorts of modules, and while users could download another app or go to a different URL, they simply don’t want to.

At the same time, I get the frustration of the OTAs specifically and all of Google’s suppliers generally: if not even four ads will deter users, and if Google is going to pay whatever it takes to be the default search engine, then isn’t it unfair for the company to collect rent in this way?

Competing With Google

Here it is worth at least considering the biggest OTA of all, Booking Holdings. The company reported its earnings a day after Expedia; from Morningstar:

Booking Holdings Inc. reported third-quarter results that beat expectations. Profit at the online travel site was $1.95 billion, or $45.54 a share, up from $1.77 billion, or $37.02 a share, a year earlier. Adjusted earnings were $45.36, up 20% from a year earlier. Analysts polled by FactSet were expecting $44.50 a share. Revenue was $5 billion, up from $4.8 billion a year earlier. Analysts were expecting $4.85 billion.

Booking Holdings CEO Glenn Fogel argued on the company’s earnings call that the company was relatively insulated from Google’s actions:

Regarding SEO, we saw some headwinds in the SEO channel that did create some modest pressure, but it’s a small channel for us.

Fogel added later on:

In the end, what’s most important for us to get customers to come to us directly. We’ve talked about this a lot in the past. It’s one of the things that I think is very important. For us to have our own future is to create a service that is so wonderful, so good that people just naturally will come back to us directly. And we will not be as dependent on other sources of traffic.

This seems like unequivocally a good thing, no? Booking knows it can’t depend on the Google channel, that its future is best secured by innovating and building a customer experience that convinces users to go to Booking directly. That is competition working to the benefit of customers!

I had a similar thought while reading this profile of Yelp CEO Jeremy Stoppelman; the ‘hotel module’ was long ago preceded by the ‘local module’ on Google search, much to Yelp’s consternation. What always gave me pause about Yelp’s complaints, though, is that, as I noted earlier, the company was at one time held up as the canonical threat to Google on mobile; why didn’t the company earn more direct customers, and instead spend so much time and energy kvetching about Google’s search results? That is why I found this bit of that profile compelling:

It also surfaces reviews algorithmically via recommendation software. It segregates reviews that the software flags for being solicited or biased, or because it doesn’t know enough about a user. Which means Yelp hides almost 30% of the reviews posted to its site, according to the company. This review filter is, to put it mildly, enormously unpopular among businesses…

“I’m sure we could have been making a lot more money if we allowed ourselves to be compromised and just said: Anything goes on Yelp. You want 5 stars? Tell your friends to go write a bunch of reviews for you and they’ll be on Yelp and then you can advertise. And wouldn’t it be wonderful?” said Stoppelman.

Instead, Yelp went another route. It is vigilant about reviews, and has passed on some easy ways to make money from users’ data. It doesn’t let businesses target users who happen to be walking by with an ad, for example. Despite persistent rumors, it’s hard to imagine Yelp fitting in as an acquisition target for Big Tech — in just two interviews with BuzzFeed News, the outspoken Stoppelman took shots at Facebook, Amazon, and Google…

“When I look out at other companies,” Stoppelman said, “I see other priorities, namely growing revenue as much as possible. So why didn’t Facebook crack down on certain types of content, or why did they allow sensational stories or stories that are not true to blast across the network and get amplified so much? Had they had the foresight to say, ‘Hey, this is bad for the world’ or ‘This is bad for our long-term brand, we should shut it down,’ it probably wouldn’t have turned into an eventually traumatic political issue.

“But at the end of the day, collecting attention is the way that they make money, and they dial up the algorithm — the same as YouTube, same for Google. You know, it’s like Google and Facebook did the same thing: Use the algorithm to optimize for maximum attention. And if you optimize for maximum attention, you’re leaning into human nature of rubbernecking at train crashes, and all the worst stuff that humanity can provide. And that’s where you end up. And I’m sure it was like rocket fuel for their business, but now we’re paying the price.”

This is by far the most compelling pitch I have heard Yelp give for itself: “The big companies are full of spam and misinformation, while we take the time to get reviews right.” It is hard not to wonder just how much more popular Yelp’s product might be if this message were spread as stridently as its anti-Google arguments.

And, of course, there is Amazon: more product searches start on Amazon than Google, not because Amazon spent its energy complaining about Google favoring its own shopping results, but because Amazon went out and delivered a better experience for users.

Monopoly Concerns

I remain very concerned about monopoly, particularly, when it comes to consumer tech, digital advertising; this Wall Street Journal story is an excellent overview of how Google makes it extremely difficult to compete (for competitive ad-tech companies) and extremely difficult to go elsewhere (for its customers).

What gives me pause about search, on the other hand, is that there are not constraints on user movement. It really is trivial to use Yelp, or Amazon, or Booking, both on the web and on a smartphone. Is customer inertia something that requires regulation, or is it a possible spur to making products that are that much more compelling?

One answer, perhaps, lies in Google’s behavior itself: unlike traditional monopolies, it is hard to argue that Google’s product isn’t getting better. Sure, OTAs need to pay to play on the hotel module, but the hotel module is a genuine improvement over 10 blue links. The same can be said of the other areas where Google gives answers instead of options. I absolutely get the argument that this might be an unfair extension of Google’s search dominance, but the possibility of stifling innovation, both directly and also its incentives, are worth consideration.

I wrote a follow-up to this article in this Daily Update.

  1. It remains a big problem that we don’t know exactly what YouTube’s financials are; if Google won’t tell us the SEC should make them []
  2. I’m using “Commissions” here broadly; there are multiple monetization models for OTAs, including selling rooms directly, charging hotels fees after-the-fact, etc. []
  3. A “point” is the functional equivalent of a pixel in the user interface; an iPhone 11 Pro has a 3x retinal display which means that three physical pixels represent one “point” []