The Walled Garden

On 9 January 2007 Steve Jobs introduced the iPhone. It came with a set of applications built by Apple, a mail program, a web browser, maps, music, and nothing else. Outside developers could not install software on it. At Apple’s developer conference that June, Jobs told programmers that they could write web applications that would run in the iPhone’s browser, which he described as a very sweet solution. Developers were not persuaded, and some began breaking the phone’s protections to install their own programs. In October 2007 Apple announced that it would release a software development kit. On 10 July 2008 it opened the App Store with about 500 applications, a quarter of them free. Users downloaded 10 million apps in the first three days.¹

The App Store changed what a phone was. Within a few years it offered millions of apps, from games and banking to messaging and maps, and created a new kind of software business. It also created a new kind of power. Every app that reached an iPhone did so through Apple’s review, under Apple’s rules, and, if it sold digital goods, with Apple taking a share of the price, usually 30 percent. Google’s Play Store did the same for Android, the world’s most used operating system, with somewhat looser rules. Two companies became the gatekeepers between millions of developers and billions of customers.

This essay follows how that happened and how it is being contested. It is a story about software distribution, but its deeper subject is governance: what a platform owner may demand from those who build on it and from those who use it, in exchange for the security and reach it provides. To understand it, we need to start with the world before the gate.

The open computer and its dangers

For most of its history the personal computer was open. Anyone could write a program for Windows or the Mac and distribute it however they liked, on floppy disks, CDs or downloads. The owner of the computer could install whatever they chose. That openness produced an extraordinary variety of software and made the PC a platform for innovation. Microsoft did not take a share of other companies’ software sales; it earned its money from selling Windows.

Openness had costs. Users installed programs that crashed their computers, slowed them down or spied on them. Viruses, worms and other malware spread through email attachments and downloads, causing billions of dollars of damage. Ordinary users found installing and managing software confusing. Security became a constant battle, fought with antivirus software and warnings that most people ignored. The essay on operating systems describes how operating systems came to impose layers of protection between programs; the PC still left the final decision, and the risk, with the user.

Some platforms had already shown another model. Game consoles from Nintendo, Sony and Microsoft allowed only licensed games, and the console makers took a fee on each. Mobile phone operators in the early 2000s controlled which applications could run on their networks’ phones and took large shares of revenue, a system developers found slow and arbitrary. In 2003 Apple’s iTunes Store began selling songs, keeping roughly 30 percent of each sale; the split became the model for the App Store.

The garden opens: the App Store

The App Store combined the console model with the scale of the web. Developers paid a small annual fee to join Apple’s programme and submitted each app, and each update, for review against Apple’s guidelines. Approved apps were distributed through the store, which handled payment, downloads and updates. For paid apps, Apple kept 30 percent and developers 70 percent. Free apps paid nothing.

For users, the store was transformative. Finding, buying and installing software became as simple as tapping a button, and apps were reviewed, sandboxed so that they could not interfere with each other or the system, and could be removed easily. Malware on the iPhone was rare. For developers, the store offered access to millions of customers with credit cards already on file, and the ability to reach them without retail distribution. Small developers became rich almost overnight. The phrase “there’s an app for that,” from Apple’s 2009 advertising, captured a new world.

In 2009 Apple introduced in-app purchases, letting apps sell digital goods and subscriptions inside the app, also at a 30 percent commission. Freemium games, free to download but earning money from purchases inside the game, became the most profitable kind of app. In 2011 Apple required apps that sold digital content, such as e-books, music or video subscriptions, to offer its in-app purchase system and barred them from directing users to cheaper purchases on the developer’s website.² Apps selling physical goods or services, such as Amazon’s shop or Uber’s rides, were exempt. Some companies, such as Netflix and Spotify, eventually stopped selling subscriptions inside their iPhone apps altogether.

Android: open source, controlled store

Google took a different approach. Android, released in 2008, was open source, and phone makers could use it freely, as described in the essay on open source. Users could install apps from outside Google’s store, the Android Market, later Google Play, by changing a setting. Other stores, such as Amazon’s and Samsung’s, existed.

In practice, Google Play carried almost all the traffic. Google licensed its store and its most important apps, such as Gmail, Maps and YouTube, to phone makers on condition that they were installed and placed prominently. It paid manufacturers and carriers to favour Play and discouraged the installation of other stores. It made sideloading possible but surrounded it with warnings. And it required apps on Play to use Google’s billing system for digital goods, with a commission similar to Apple’s. Android’s openness was real in law and limited in practice.

Building on someone else’s land

The App Store was not the first platform to discover the power of controlling access. Facebook opened its social network to outside developers in May 2007, letting them build applications that ran inside Facebook and could see users’ friends and spread through their networks. Games such as Zynga’s FarmVille grew to tens of millions of players by spamming friends’ news feeds with notifications. As users complained, Facebook restricted what apps could post and, later, what data they could access. Zynga’s growth stalled, and its dependence on Facebook became a warning to investors.³ In 2014 Facebook announced it would stop letting apps access the data of users’ friends; it was data obtained under the earlier rules that the political consultancy Cambridge Analytica misused, causing a scandal in 2018.

Twitter followed a similar arc. Early on it encouraged outside developers, who built many of the most popular ways to use the service. From 2012 it restricted access to its data to keep users within its own apps, and in 2023, under new ownership, it ended free access to its programming interface, shutting down third-party apps that had existed for over a decade. The writer Nicholas Carr had warned of “digital sharecropping” in 2006, describing users who create value on platforms they do not own. Developers learned the same lesson: building on someone else’s platform is building on rented land.

The economy of the stores

By the late 2010s the app stores were among the largest marketplaces in the world. Consumers spent more than $100 billion a year on apps across Apple’s and Google’s stores by 2020, most of it on games.⁴ Apple’s services business, of which the App Store is a major part, grew to more than $100 billion in annual revenue. Apple argued that the store had created millions of jobs and an ecosystem worth trillions of dollars, that the commission paid for review, security, payment processing, distribution and the development tools developers used free, and that most apps paid nothing because they were free or sold physical goods.

Developers who paid the commission increasingly disagreed. The 30 percent rate had been set when distributing software was costly; by the 2010s, they argued, it was far above the cost of the services provided, and was sustainable only because Apple and Google faced no competition for distribution on their own phones. Subscription businesses, whose customers renewed for years, found the commission especially burdensome. Rules about what apps could do, enforced inconsistently by review teams, produced complaints of arbitrariness. In 2020 Apple rejected an update to Hey, a new email service from the company Basecamp, for not offering in-app purchase, and reversed course after public protest.

The challenge

In 2019 Spotify filed a complaint with the European Commission, arguing that Apple’s rules forced it either to pay 30 percent on subscriptions, more than its profit margin, or to sell subscriptions outside the app without being allowed to tell users how, while Apple’s own competing music service paid no commission.

On 13 August 2020 Epic Games, the maker of Fortnite, one of the world’s most popular games, added a direct payment option inside Fortnite on iPhone and Android, offering lower prices than through the stores. Apple and Google removed Fortnite within hours. Epic immediately filed antitrust lawsuits against both, along with a public campaign, including a parody of Apple’s famous 1984 advertisement, that it had prepared in advance.⁵

In November 2020 Apple announced a Small Business Program, reducing the commission to 15 percent for developers earning up to $1 million a year, a group it said included the large majority of developers, though a small share of revenue.⁶ Google made similar changes.

The cases went in different directions. In September 2021 Judge Yvonne Gonzalez Rogers ruled in Epic’s case against Apple that Epic had not proved Apple was a monopolist in the relevant market, but that Apple’s rules against steering users to other ways to pay violated California’s unfair competition law. She ordered Apple to allow developers to include links and buttons directing users to outside purchasing. In December 2023 a jury in San Francisco found that Google had illegally monopolised app distribution and in-app billing on Android, through agreements with manufacturers, carriers and developers. Judge James Donato’s injunction, entered in 2024, required Google to allow rival stores access to the Play Store’s catalogue and to distribute rival stores through Play for several years.⁷ After a proposed settlement between Google and Epic was withdrawn in July 2026, rival app stores began appearing within Google Play for US users later that month.

Apple’s response to the 2021 injunction became a case of its own. When the order took effect in January 2024, Apple allowed developers to link to outside purchases but charged a commission of 27 percent on purchases made within a week of following such a link, required warning screens and limited how links could appear. Epic asked the court to enforce the injunction. On 30 April 2025 Judge Gonzalez Rogers found Apple in civil contempt, concluding that it had willfully violated the order, barred it from charging any commission on linked purchases, and referred the matter to federal prosecutors for possible criminal contempt proceedings. She found that an Apple executive had given false testimony. In December 2025 the Ninth Circuit Court of Appeals upheld the contempt finding but held that the ban on any commission went too far, allowing Apple to seek a fee based on its actual costs of supporting such links. In 2026 the Supreme Court agreed to hear Apple’s appeal.⁸

The regulators

Regulators moved alongside the courts. In March 2024 the European Commission fined Apple €1.8 billion, finding that its rules preventing music streaming apps from informing iPhone users about cheaper subscriptions outside the app were an abuse of its dominant position; Apple appealed. The same month the EU’s Digital Markets Act began to apply to Apple and Google as designated gatekeepers. It required them, among other things, to allow apps to be installed from alternative app stores and from the web, to permit developers to steer users to outside offers, and to stop favouring their own services.⁹

Apple complied with new terms for Europe that allowed alternative stores but introduced new fees, including a per-installation “core technology fee” for apps above a threshold, which developers argued made alternatives uneconomic. In April 2025 the Commission fined Apple €500 million for breaching the DMA’s anti-steering provisions, and Apple again revised its terms and appealed. Britain’s Competition and Markets Authority designated Apple and Google in October 2025 as having strategic market status in mobile platforms, giving it powers to impose conduct requirements, and cited concerns about app review, app store search and commissions of up to 30 percent.¹⁰ Japan passed a law in 2024, fully effective in December 2025, requiring Apple and Google to allow alternative app stores and payment systems. South Korea had legislated on in-app payments as early as 2021.

What app stores are today

Today the app stores remain the main route for software to reach smartphones, and their owners remain enormously powerful. But the walls have been breached in places. In Europe, alternative stores and outside payments are legally required; in the United States, developers can link to outside payment on iPhones, and Google must open Play to rival stores; in Britain, Japan and South Korea, regulators have powers to intervene. Commissions have been cut for small developers and subscriptions after the first year. The core question, whether and how much a platform owner may charge for access to the customers on its devices, is before the US Supreme Court and regulators on several continents.

How app stores changed software and power

Several shifts follow from the rise of app stores.

From installing anything to installing what is approved

On the PC, the owner decided what ran on their computer. On the phone, the platform owner decided what could be offered. Users gained safety and simplicity, and gave up a choice many had never consciously exercised.

From distribution as a cost to distribution as a toll

Software distribution became nearly free technically and expensive commercially. The cost of reaching customers was no longer printing discs or paying retailers but a percentage of revenue paid to the owner of the gate.

From an open market to a governed one

App developers came to live under detailed, changing rules set by a private company, enforced by review teams and appealed, if at all, within the company. The platform became a regulator of an industry.

From platform risk as an exception to platform risk as the norm

Facebook’s and Twitter’s developers learned that platforms change their rules. App developers learned that a rejected update or a policy change could cut off their business overnight. Dependence on a platform became a central risk for software companies.

From competition law to platform law

Antitrust cases, designed for markets of many sellers, struggled with platforms that were also markets. Lawmakers responded with new rules written specifically for gatekeepers, such as the DMA and the UK’s digital markets regime.

What the history suggests

App stores repeat patterns found across this site. Taking choices away, through curation and review, was itself the product, and it opened smartphones to people who would never have managed software on a PC. Distribution beat elegance: the store on the phone won, whatever the merits of alternatives. Incumbents absorbed the challengers’ ideas, building their own versions of popular apps on the platforms they controlled. And who paid shaped what was built: the commission favoured free apps with in-app purchases and advertising over paid software.

But app stores also show something more particular. They are perhaps the clearest example of a private company exercising something like sovereignty over an economy: setting the conditions of entry, the rules of conduct, the rate of tax and the procedures of appeal for millions of businesses. Their owners justify that power by the security and order it provides. Their critics ask what limits should apply to it. That is the question political philosophy has asked of sovereigns for centuries, and it is where the philosophy begins.

Understanding the philosophy

The rest of this essay examines the ideas beneath app stores and platforms, one at a time. Each is explained in plain terms, connected to the tradition of thought that has examined it, traced in the history above, tested against its strongest objection and related to how organisations decide today. The connections to philosophers, lawyers and economists are interpretations offered to make the ideas clearer. The people who built app stores were solving practical problems, and did not, as a rule, set out to answer these thinkers. The essay on online marketplaces discusses platforms as private governments in Elizabeth Anderson’s sense; this essay looks at sovereignty, law and access.

1. Security in exchange for obedience

The idea in plain terms

The App Store offers a bargain: give up the freedom to install anything, and in exchange get a phone that is safe, simple and reliable. Most people accept it gladly. But the bargain has the shape of one of the oldest arguments in political philosophy: that people should submit to a powerful ruler because the alternative is chaos.

The tradition

Thomas Hobbes, writing during the English Civil War, argued in Leviathan (1651) that without a common power to keep them in awe, people live in a state of war of every one against every one, in which life is, in his famous phrase, solitary, poor, nasty, brutish and short. To escape it, people agree to transfer their rights to a sovereign, whose power is nearly absolute, because a divided or limited sovereign cannot guarantee peace. The subjects keep only the right to preserve their own lives.

John Locke replied in his Second Treatise of Government (1689) that people submit to government to protect their lives, liberties and property, and that a government which abuses its power breaks the trust on which it rests. Power should be limited by law, divided, and subject to the consent of the governed. The difference between Hobbes and Locke is the difference between security at any price and security under limits.

Where it shows up

The PC, with its viruses and confusion, resembled Hobbes’s state of nature: anyone could do anything, and users bore the risks. The App Store was a Leviathan: Apple took the power to decide what software could run, and in exchange offered security. Apple’s defence of its rules, in court and in public, has consistently been Hobbesian. It argues that opening the iPhone to outside stores and sideloading would expose users to malware, fraud and privacy abuses, and that the integrity of the system depends on its control. The DMA, the court rulings and the British and Japanese laws are Lockean: they accept the platform’s role but seek to limit its power by law and to protect the liberty of developers and users.

The strongest objection

The objection is that the analogy flatters the critics. Users choose iPhones freely, knowing the rules, and can buy Android phones instead; a citizen cannot so easily choose another sovereign. And the security benefits are real: iPhones suffer far less malware than open systems. This is a strong point. But the choice between two platforms, each with its own gate, is a limited choice, and switching costs, from apps bought to data stored, are high. And the security argument has been stretched to cover rules, such as anti-steering, that have little to do with security.

What it means today

For users and organisations, the question to ask of any platform’s restrictions is which ones serve security and which serve the platform’s revenue. For platform owners, the credibility of security arguments depends on not using them to defend rules that protect income. The Hobbesian bargain is accepted more readily when the sovereign does not appear to be taxing its subjects under the cover of protecting them.

2. The gate and the toll

The idea in plain terms

If one company controls the only road to a town, it can charge whatever it likes for passage. Law has long treated such bottlenecks specially, requiring that access be given on fair terms. App stores are a new kind of road: the only route to the customers on a platform’s devices.

The tradition

In 1912 the US Supreme Court decided United States v. Terminal Railroad Association. A group of railways had acquired control of the bridges and terminals through which all rail traffic into St Louis had to pass. The Court required them to admit competing railways on just and reasonable terms. The case is often cited as the origin of the “essential facilities” doctrine, which holds that a firm controlling a facility competitors cannot practically duplicate may be required to give them access.

Older still is the common-law tradition of common carriers and businesses “affected with a public interest.” In Munn v. Illinois (1877) the Supreme Court upheld state regulation of grain elevator rates, quoting the seventeenth-century English jurist Matthew Hale’s view that when private property is devoted to a use in which the public has an interest, the owner must submit to public regulation. Ferries, inns, railways and telephone networks were held to duties of serving all comers at reasonable rates.

Where it shows up

The app store debate is essentially a debate about whether the stores are essential facilities or common carriers. Developers argue that the only way to reach iPhone users is through the App Store, so Apple controls an essential route and should give access on reasonable terms. Apple argues that the App Store is its own creation, part of an integrated product, and that it may set its own terms. The US courts have not treated the store as an essential facility; Epic’s antitrust claim against Apple failed. But the Ninth Circuit’s ruling that Apple may charge only a fee based on the costs genuinely and reasonably necessary for supporting outside links echoes the common carrier principle of reasonable rates. The DMA, by requiring gatekeepers to allow access, writes something like the essential facilities idea into statute.

The strongest objection

The objection is that the essential facilities doctrine is disfavoured in US law because forcing companies to share what they built discourages investment. Apple invested billions in the iPhone and the App Store; if others can use them at cost, Apple’s incentive to build the next platform falls. And unlike a bridge, the App Store competes with another platform, Android. These points are serious, and courts have been cautious. The counter-argument is that the reward for building the platform comes from selling devices, which are highly profitable; a toll on all software sold for them may exceed what is needed to motivate investment.

What it means today

For businesses that depend on a platform, the history suggests that the terms of access are a matter of law and politics as much as negotiation, and are changing. For companies that run platforms, behaving like a reasonable carrier, charging fees related to services rendered and applying them consistently, is the best defence against being regulated as one.

3. Building on rented land

The idea in plain terms

A developer who builds a business on a platform is like a farmer working land owned by someone else. The farmer may prosper for years, but the owner can change the rent, the rules or the use of the land at will. Platform risk, the danger that a platform’s decision will destroy a business built on it, is the defining hazard of the modern software industry.

The tradition

Historians of agriculture have long studied tenancy and sharecropping, systems in which farmers work land owned by others and give the owner a share of the crop. Sharecropping in the American South after the Civil War left many Black and poor white farmers in debt and dependent on landowners who set the terms. Economists studied the incentives: a farmer who keeps only part of the harvest invests less in improving land he may not keep.

Nicholas Carr’s term “digital sharecropping,” coined in 2006, applied the image to users who created content for platforms that kept most of the value. Developers’ relationship with platforms has a similar structure: they invest in building on the platform, the platform takes a share, and the platform can change the terms.

Where it shows up

Zynga built a business on Facebook’s platform and lost much of its reach when Facebook changed the rules. Developers of Twitter clients lost their businesses when Twitter cut off access. App developers have seen updates rejected, features banned and business models outlawed by changes in app store guidelines. Each was rational to build on the platform, because that was where the customers were, and each bore the risk of the platform’s decisions. The network reversal pattern describes what happens when developers lose faith and leave together.

The strongest objection

The objection is that developers choose to build on platforms knowing the terms, and gain enormously from doing so: access to customers, payment, distribution and tools they could never have built. Many of the most successful companies of the last two decades were built on app stores. Calling them sharecroppers insults their success. This is fair; the relationship is far more beneficial to developers than historical sharecropping was to farmers. But the structure of dependence is similar, and the platform’s power to change terms unilaterally is real.

What it means today

For businesses, platform risk should be managed like any other concentration risk: keep direct relationships with customers, maintain channels the platform does not control and avoid models that exist only because of a rule that could change. For platform owners, the willingness of developers to invest depends on predictable rules; a platform that changes them abruptly teaches developers to invest elsewhere.

4. Rules without the rule of law

The idea in plain terms

The App Store has detailed guidelines, but their application can seem inconsistent. Apps are rejected for reasons not clearly stated, rules change without warning, and appeals are decided by the same company that wrote the rules. For a developer whose business depends on approval, this is governance without many of the protections people expect from law.

The tradition

In The Morality of Law (1964) the American legal philosopher Lon Fuller argued that law has an “inner morality” consisting of eight requirements. Rules must be general; publicly promulgated; prospective rather than retroactive; clear; consistent with each other; possible to obey; relatively constant over time; and applied by officials in a way that matches the rules as announced. A system that fails badly on any of these, Fuller argued, is not properly a legal system at all, because it does not allow people to plan their lives by the rules.

The tradition of the rule of law, from Aristotle’s preference for the rule of law over the rule of men to the nineteenth-century jurist A. V. Dicey’s account of equality before the law, emphasises that power should be exercised through known rules, applied impartially, with independent judgement of disputes.

Where it shows up

App review has been criticised on almost every one of Fuller’s requirements. Guidelines have been vague, leaving reviewers wide discretion. Rules have been applied inconsistently, with similar apps treated differently. Changes have taken effect with little notice. The congruence between the published rules and their application has been questioned, as in the Hey episode. And the final decision rests with Apple. Britain’s CMA cited unpredictable app review among its concerns when designating Apple and Google. Regulations such as the EU’s Platform to Business Regulation require clear terms, notice of changes and reasons for decisions, which are Fuller’s principles written into law.

The strongest objection

The objection is that app review is a commercial process, not a legal system, and that a degree of discretion is needed to stop new forms of fraud and abuse that rules cannot anticipate. Apple reviews hundreds of thousands of submissions a week; perfect consistency is impossible. And Apple has improved its processes, publishing guidelines, explaining rejections and creating an appeals board. These points are fair. But the more a commercial process determines whether businesses can exist, the more it is judged by the standards of law.

What it means today

For anyone who runs a system that decides on others’ applications, whether an app store, a procurement process, a grant programme or a promotion committee, Fuller’s eight principles are a practical checklist. For businesses subject to such systems, documenting decisions and pressing for reasons is a way of bringing the rule of law into private governance.

5. Protecting people from themselves

The idea in plain terms

Apple argues that letting users install apps from anywhere would expose them to harm, and that most users are better off if the choice is made for them. That is paternalism: limiting people’s freedom for their own good. Sometimes paternalism is wise. Sometimes it is a pretext.

The tradition

John Stuart Mill argued in On Liberty (1859) that the only purpose for which power can rightfully be exercised over any member of a civilised community, against their will, is to prevent harm to others; a person’s own good is not a sufficient warrant. Mill allowed exceptions, for children and for people who lack information: one may stop someone crossing an unsafe bridge if there is no time to warn them, but if they know the danger, they may cross.

In 1972 the philosopher Gerald Dworkin analysed paternalism more precisely, distinguishing hard paternalism, overriding informed choices, from soft paternalism, intervening to ensure choices are informed and voluntary. Thaler and Sunstein’s libertarian paternalism, discussed in the essay on recommendation feeds, proposed that defaults can steer people while leaving them free to choose otherwise.

Where it shows up

Apple’s closed system is hard paternalism: users cannot install software from outside the App Store, however well informed they are. Android’s model is closer to soft paternalism: sideloading is possible, behind warnings. The DMA requires Apple to move towards the soft model in Europe. Apple’s implementation, with multiple warning screens and fees, has been criticised as hard paternalism disguised as soft. The argument that users must be protected from cheaper subscriptions on a developer’s website is harder to defend on Mill’s grounds than the argument that they must be protected from malware.

The strongest objection

The objection is that most users cannot assess the security of software, so informed choice is an illusion, and soft paternalism will lead many to install harmful apps after clicking through warnings they do not read. Mill’s unsafe bridge is the norm, not the exception. This is a strong argument for security protections. It is weaker as an argument for rules about payments and business models, where the harm is to the platform’s revenue, not the user.

What it means today

For organisations that restrict what employees or customers can do for their own good, the test is whether the restriction is aimed at a real harm, whether a less restrictive measure would work, and whether the restricting party gains from it. For platform owners, separating security rules from commercial ones makes both more defensible.

6. Generativity and the tethered appliance

The idea in plain terms

Open systems let anyone build anything, including things no one anticipated. That capacity, which made the PC and the internet so creative, depends on letting strangers run code without permission. Closed systems are safer but can prevent the unexpected. The question is how to keep the creativity while controlling the risk.

The tradition

In The Future of the Internet and How to Stop It (2008), published in the year the App Store opened, the legal scholar Jonathan Zittrain argued that the PC and the internet were generative: they could be used by a broad and varied audience to create unanticipated change. Their openness also made them vulnerable to abuse, which was pushing users towards what he called tethered appliances, devices controlled by their makers, which could be updated or restricted remotely. He predicted that the iPhone represented the future: safe, sterile and controlled. Zittrain worried that if people retreated to appliances to escape insecurity, the generativity that had produced the internet’s innovations would be lost.

Where it shows up

The App Store is the tethered appliance Zittrain described, combined with a mechanism for outside developers: generative within limits set by its owner. It produced an enormous amount of innovation, but within categories Apple permitted. Some kinds of software, such as alternative browser engines, emulators and certain apps that compete with Apple’s own, were long restricted. The web, where anyone can publish without approval, remained generative, and Apple’s restrictions on web technologies on the iPhone became a subject of regulatory concern.

The strongest objection

The objection is that Zittrain’s fear was overstated. The App Store unleashed a wave of innovation, from ride-hailing to mobile banking, that the open PC never did on phones. Curation made millions of people willing to try new software. Generativity flourished within the garden. This is largely true. The question is what did not happen: which apps were never built because they would not be approved, and which businesses never formed because the commission made them uneconomic. That is invisible by nature, and its absence is no evidence that it did not matter.

What it means today

For leaders designing internal platforms, approval processes and tools, Zittrain’s trade-off applies directly: control reduces risk and can stifle the unexpected. The useful question is where to allow open experimentation, under what safeguards, rather than whether to control everything or nothing.

7. Why the fee did not fall

The idea in plain terms

In most markets, competition pushes prices towards cost. The app store commission stayed at 30 percent for more than a decade despite falling costs of distribution. Why? Because developers needed to be on both platforms, while each user typically had only one phone. Each store was the only route to its users, and competition between the platforms did not discipline the fees they charged developers.

The tradition

In 2006 the economist Mark Armstrong analysed competitive bottlenecks in two-sided markets. When one side of a market, here consumers, chooses only one platform (single-homing), and the other side, here developers, must be on all platforms to reach everyone (multi-homing), each platform has a monopoly over access to its single-homing users. Platforms compete for consumers, often by subsidising them, but charge the multi-homing side high prices, because that side has no alternative. The theory predicts that fees to developers will stay high even when platforms compete vigorously for users.

Where it shows up

The app stores fit the model precisely. Apple and Android compete intensely for consumers, on devices, cameras and prices. Developers must offer their apps on both, because each user has one phone. Each store therefore has a bottleneck over its users, and both maintained similar commissions for years. Commission cuts came not from competition between the stores but from litigation, regulation and public pressure. The UK CMA’s description of an effective duopoly, with 90 to 100 percent of mobile devices on one of the two platforms, reflects this structure.

The strongest objection

The objection is that the commission is a fair price for the value developers receive, and that it has in fact fallen for many developers, through the small business and subscription programmes. Developers would pay similar shares to retailers, console makers or payment processors. Moreover, the theory of competitive bottlenecks does not show that the fee is too high, only that competition does not constrain it. This is correct as far as it goes. The policy question is whether, without competitive pressure, the fee should be constrained by something else, and courts and lawmakers have increasingly said yes.

What it means today

For businesses deciding where to sell, recognising a competitive bottleneck clarifies negotiating power: where customers single-home, the platform holds the cards. For platform owners, high margins sustained by a bottleneck attract regulatory attention, and reductions offered voluntarily tend to come too late to prevent it.

8. What is a fair share?

The idea in plain terms

Underneath the litigation lies an old question: what is a fair price? Apple says its commission reflects the value of access to its customers and its platform. Developers say it should reflect the cost of the services provided. Courts have begun to say that, at least for some purposes, the fee should be tied to costs. The argument between value and cost is one of the oldest in economic ethics.

The tradition

Aristotle discussed justice in exchange in the Nicomachean Ethics, arguing that a just exchange involves a kind of proportional equality between what is given and what is received. Medieval scholastic philosophers, among them Thomas Aquinas, developed the idea of the just price: selling a thing for more than it is worth is unjust in itself, though they generally treated the common market price, in the absence of fraud or monopoly, as a guide to the just price. They were particularly hostile to monopolists who raised prices by controlling supply.

Modern economics largely replaced the just price with the market price: whatever buyers and sellers freely agree. But it kept the exception for monopoly. Utility regulation, which sets the prices of electricity, water and telephone service, typically allows a monopolist to recover its costs plus a reasonable return on investment, on the principle that where competition cannot set prices, a regulator must.

Where it shows up

The app store commission was a market price only in a limited sense: developers agreed to it because they had no alternative. The argument over it has moved through the stages the tradition would predict. First, public complaints about fairness. Then reductions offered by the platforms. Then courts and regulators asking what the fee pays for. The Ninth Circuit’s ruling that Apple may charge a commission on linked-out purchases “based on the costs that are genuinely and reasonably necessary for its coordination of external links” but no more is, in effect, a cost-based just price for one part of the platform. The question of what Apple may charge for the rest, the value of its customers and its intellectual property, remains contested.

The strongest objection

The objection is that cost-based pricing ignores the value of the platform. Apple built the iPhone and the customer base; the store’s value to developers far exceeds the cost of running it, and Apple is entitled to share in that value, as any business is. Limiting it to costs would expropriate its investment. This is the heart of the dispute, and it has no simple answer. Utility regulation’s answer, a reasonable return on investment, is one possible compromise.

What it means today

For any business that charges others for access to something it controls, the question of fairness is not only ethical but practical: prices far above cost, sustained by control rather than competition, invite the question of what they pay for. For those who pay such prices, the history shows that the answer can change, through law, regulation or public pressure, even for the most powerful companies.

Bringing the threads together

App stores offered a Hobbesian bargain: security in exchange for obedience, which most users gladly accepted. Their owners came to control a gate that developers could not avoid, raising the old legal question of what owners of essential routes may charge. Developers built on rented land, and some lost it when the rules changed. Rules were made and applied without many of the protections Fuller identified as the inner morality of law. Paternalism justified by security was extended to commercial restrictions that security could not justify. The garden was generative within its walls, and what it prevented cannot be counted. The fee stayed high because each platform held a bottleneck over its own users. And the question of what share is fair has returned, after centuries, to the centre of economic argument.

The walled garden

When Apple opened the App Store, it solved a real problem: ordinary people could not safely manage software on their computers. The garden it built made smartphones useful and safe for billions of people and created new industries. Google’s more open version brought the same benefits to the majority of the world’s phones.

The history also shows what happens when a garden becomes the only place to grow. Its owner becomes a sovereign over an economy, and the questions that apply to sovereigns begin to apply to it: on what terms may it rule, what may it charge, and how may its subjects appeal? The working principles that follow are not complicated. If you build on a platform, treat its rules as a risk and keep a way to reach your customers directly. If you run one, separate the rules that protect users from the rules that protect revenue, apply them as consistently as law, and price access as if you might one day have to justify the price. And if you are a user, remember that the safety you enjoy is real and that it is not the only thing you are paying for. Every garden has a wall. The question is who holds the key, and on what conditions they will open the gate.

Selected bibliography

• Aristotle. Nicomachean Ethics, Book V. Fourth century BCE.
• Aquinas, Thomas. Summa Theologiae, II-II, q. 77. Thirteenth century.
• Hobbes, Thomas. Leviathan. 1651.
• Locke, John. Second Treatise of Government. 1689.
• Mill, John Stuart. On Liberty. 1859.
• Munn v. Illinois, 94 U.S. 113 (1877).
• United States v. Terminal Railroad Association of St. Louis, 224 U.S. 383 (1912).
• Fuller, Lon L. The Morality of Law. Yale University Press, 1964.
• Dworkin, Gerald. “Paternalism.” The Monist 56, no. 1 (1972): 64–84.
• Armstrong, Mark. “Competition in Two-Sided Markets.” RAND Journal of Economics 37, no. 3 (2006): 668–691.
• Carr, Nicholas. “Sharecropping the Long Tail.” Rough Type, December 2006.
• Zittrain, Jonathan. The Future of the Internet and How to Stop It. Yale University Press, 2008.
• Epic Games, Inc. v. Apple Inc., No. 4:20-cv-05640 (N.D. Cal.), Rule 52 Order, 10 September 2021; Order Granting Motion to Enforce Injunction, 30 April 2025.
• Epic Games, Inc. v. Apple Inc., No. 25-2935 (9th Cir. 11 December 2025).
• In re Google Play Store Antitrust Litigation, No. 3:21-md-02981 (N.D. Cal.), jury verdict, December 2023.
• European Commission. Decision in Case AT.40437, Apple – App Store (Music Streaming), 4 March 2024.
• Regulation (EU) 2022/1925 (Digital Markets Act).
• Competition and Markets Authority. Strategic market status decisions on Apple’s and Google’s mobile platforms, October 2025.

Research limitations

The account of Steve Jobs’s remarks about web apps relies on reports of his 2007 conference presentation. App Store launch figures are Apple’s own. Figures for consumer spending on apps come from market research estimates. Apple’s claims about the value of its ecosystem come from studies it commissioned and are not independently verified here. The description of Google’s agreements with manufacturers draws on the findings in the Epic v. Google case as reported. The account of the Epic v. Apple contempt proceedings relies on reports of the court’s orders and the appeal; the Supreme Court had not decided the appeal at the time of writing, and the ruling may change the picture. The status of Apple’s EU terms and of its appeals against EU fines was changing during 2025 and 2026. Japan’s smartphone law and Britain’s conduct requirements were in early implementation. The philosophical connections to Hobbes, Locke, Matthew Hale, Fuller, Dicey, Aristotle, Aquinas, Mill, Dworkin, Thaler and Sunstein, Carr, Zittrain and Armstrong are the author’s interpretation; their works are paraphrased apart from short, attributed phrases. The decisions in the decisions box are judged by the author, separating what could be known at the time (process) from what happened (call).

Footnotes

1. “Today in Apple history: App Store launches,” Cult of Mac. ↩

2. European Commission decision on Apple App Store (music streaming), March 2024, as summarised by Slaughter and May. ↩

3. Wikipedia, “Facebook Platform”; Wikipedia, “Zynga”. ↩

4. “Global app spending passed $100 billion in 2020,” The Wire. ↩

5. Wikipedia, “Epic Games v. Apple”. ↩

6. Apple, “Apple announces App Store Small Business Program,” 18 November 2020. ↩

7. “Google and Epic abandon injunction modification request,” MacRumors, July 2026; “Google and Epic propose Android settlement,” PPC Land. ↩

8. Epic Games v. Apple, No. 25-2935 (9th Cir. 2025), Justia; “Epic Games’ Ninth Circuit win affirming civil contempt finding,” Cravath; “Supreme Court declines to pause order holding Apple in contempt,” CNBC; “Supreme Court will hear Apple’s appeal,” The Next Web. ↩

9. “EU imposes first fines for DMA breaches,” Law Society Gazette, April 2025; “Apple’s new EU App Store terms still breach the DMA, says developer coalition,” Mobilegamer.biz. ↩

10. “UK regulator confirms Apple and Google hold strategic market status,” TechRepublic, October 2025. ↩