Discover more from Margins by Ranjan Roy and Can Duruk
Who Gets a Slice of Apple’s Pie?
A Discussion of Tech Platform Antitrust
Hi, Jonathon here, writing in Margins because these folks have been kind enough to give me a platform to throw out some opinions. Tech antitrust seems to be in vogue again, so I thought it might be a decent time to write out some of the thoughts I’ve been hashing out over these issues with my friends over the years. These opinions are my own, and in no way represent the views of my employer.
The current discussion over tech platform antitrust appears to revolve around a few central arguments. There’s the traditional Brandeis-adjacent notion of antitrust – the market concentration of large tech firms is too high, and this market dominance is used to crowd out competitors. You can read Stoller for many of the more trenchant arguments on this side. The counterarguments usually come in the form of either arguing that anti-monopolists don’t understand technology, or that many major tech products don’t actually have dominant market positions. For example, Apple’s iPhone represents less than 20% of the global smartphone market and is close to even with Android in the United States. Apple itself likes to point this out – the implication being that the sector is actively competitive and Apple isn’t even in the first place, let alone anti-competitive.
I believe the tension in these arguments comes from a complicated reality – the market dynamics of many software platforms are not easily addressed by regulatory precedent regarding market competition and consumer welfare.
To be a bit reductive for the sake of brevity, the usual Econ 101 consideration for antitrust and monopolies goes something like this: a monopoly (or somewhat similarly, oligopoly) has enormous or total market power, preventing competition from entering the market. They are therefore able to extract maximum value from consumers, capturing more producer surplus (e.g. raising prices), etc. This is generally bad for consumers (the consumer-good standard for antitrust since the 70s has major issues in general when applied to tech, especially for “free” products like Google or Facebook, but that’s an argument for another day). If you have a company manufacturing toys, and they are the only such business in town, they can charge you much more than if they had competitors, and also have no real incentive to make better toys.
If you break up that toy company into, say, 500 smaller companies, they all have incentives to compete, lower prices, and produce better products. Pretty simple theory.
The problem with software (and also some hardware) platforms is that they’re often not single-layered markets. Take general-purpose operating systems like Windows or iOS – there is an additional market for other software goods that run on top of these operating systems. However, in general, the portability of software across ecosystems is limited because each system is uniquely constructed, and so software doing the exact same thing on both systems must be purpose-built for each.
So, if you broke up Microsoft in 2000 when it had 97% desktop OS marketshare into 500 competing desktop OS companies, a developer looking to address that market would now need to write 500 partially-distinct software programs rather than one (again, being very simplistic and overbroad here). Software developers are extraordinarily expensive – hiring large additional quantities per supported platform for development and ongoing support is non-trivial. There’s a reason why so many startups launch only on iPhone. Each sufficiently distinct technical product [that has additional market layers on top of it] effectively spawns its own market. That is to say, there isn’t just an “Apps for phones” market – there is a separate “apps for the phone with OS ___” market for every single operating system. Further, this creates a feedback loop vis-à-vis the desirability of the platform product, since an operating system isn’t as attractive (or indeed is largely useless) without a large and vibrant ecosystem of available software.
For participants higher up on the stack (Note 1), the optimal scenario is to minimize the number of platforms that require support, especially as engineering requires large numbers of expensive and specialized workers. It is cost-prohibitive to support more than a few platforms, which is what leads to a dynamic where only a small number of very successful ecosystems can garner viable developer support, making market entry nigh-impossible.
Some history to support this pattern in the technology world in the last few decades:
x86 vs RISC instruction sets
VHS vs Betamax home media formats
DOS/Windows vs MacOS in desktop operating systems
HD DVD vs BluRay in-home media formats
Nintendo consoles vs Sony consoles vs Microsoft consoles (before Microsoft’s entry, Sega consoles)
iOS vs Android in mobile operating systems
Alexa vs Google Assistant vs Siri (kind of) in digital voice assistants
Many of these competitions ultimately boiled down to one truly dominant platform, as was the case with home media and desktop operating systems.
If this is the case, does that mean that any antitrust measure is futile or counterproductive? That viewpoint seems relatively common in industry (though hilariously Google employees’ top destination for Dem primary donations was Warren, who wants to break up their company). I would argue that resigning society to the whims of dominant technology platforms is an unnecessary capitulation of the public good into private hands. The efficiencies involved in developing hardware or software solutions optimized for the largest audience do mean that a simple overbroad goal like “break up the big tech companies” is likely doomed to fail, but platform power does come with significant capacity for abuse which we would do well to address.
For the sake of specificity, I’m going to focus narrowly on Apple’s controversies surrounding its app store on iOS. The brouhaha may be relatively recent in mainstream media, but in the independent pockets of the software world, Apple’s locked-down approach has long chafed (Note 2).
If we accept the proposition that Apple’s 20% smartphone marketshare actually translates to a 100% monopoly on iOS devices, this means consumers seeking iOS-only phones have no choice but to access software that can only be approved for the device by Apple. This kind of lock-in is not hypothetical. Beyond purchases tied to an Apple account, precious memories in iCloud, and exclusivity and shenanigans around iMessage (Note 3), it can be very difficult to leave Apple’s embrace. This stands in stark contrast to other general-purpose computing platforms, like Windows, which allows arbitrary software installation, and Android, which allows a user to manually “side-load” applications to bypass Google’s app store.
Obviously, this presents significant conflicts of interest, since Apple also develops its own profit-making software and services for iOS devices, and is therefore in a position to pick and choose which of its competitors are allowed to sell their software at all. Here’s a quick non-exhaustive list of the ways in which Apple can exert power by exercising strict control over any app that they will allow on their phones:
Arbitrarily reject applications that compete with their own, like game streaming services which compete with Apple Arcade, their new game subscription
Require that subscriptions be made through the app store even for services that exist outside of iPhones, subjecting them to Apple’s 30% tax on purchases. Apple’s own documentation is here.
This obviously doesn’t affect Apple’s own services and paid software.
Restrict some capabilities and APIs on the phones so that only Apple first-party software can utilize them, giving them an advantage over competitors who cannot use such tools.
Just one example, though a half decade later this was eventually rectified Use app store rules to bar certain types of software. For example, the only way to ship a browser on iPhone or iPad is to use this, which means that all browsers on iPhones are actually running Safari’s web engine and not allowed to use their own.
Ban third-party apps from being set as default for various functions like mapping, messaging, email, video calls, etc. Apple has very gradually reduced these restrictions over time, most notably in the latest major iOS version, iOS 14, but there are still many Apple services that they will not let you replace, like messaging.
Use OS surface area to promote Apple goods and services over competitors like those notifications sent to the settings app to sign up for Arcade and Apple TV+. It’s not just free advertisement for Apple – competitors can’t access this space to advertise at all, and the system-level integrations can give the impression that you’re meant to use Apple software.
There is also a long history of Apple inconsistently applying its own stated rules, often to the detriment of third-party developers and to the consumers who would benefit from more robust competition in the marketplace. So, this entire situation seems pretty straightforwardly bad. Software developers do not want to support writing custom software across dozens of operating systems, so forcing more OS diversity through breakups will not result in a good outcome. However, allowing limited diversity results in adverse incentives and anticompetitive behavior on the part of platform holders.
To resolve this, some have proposed legislation allowing alternate app stores, which would allow users to pick and choose software as they pleased. This seems to resolve the above-stated tension, but unfortunately, here is where things get much more complicated.
Apple’s app store on iOS is not just notable because it represents an unparalleled central control over what forms of software competition are allowed on their platform – it was also a genuine revolution in consumer benefit. Specifically, one key advantage of iOS over other operating systems lies exactly in that users cannot install arbitrary software. By screening everything allowed, Apple can bar malicious applications, as well as promote a higher bar of quality (in theory; a surplus of fart apps suggests the latter hasn’t been an ongoing concern).
One of Windows’ keys failings lies in its end-user friendly approach to third-party software – you can install anything from anywhere. This is exactly what causes, say, your well-meaning Uncle to accidentally install a trojan that will steal his bank account login while browsing a website with, er, Trojans. By contrast, an iPad allows only Apple-approved software to be installed. Because of app store rules, all apps are more limited in the scope of potential harms, subject to Apple’s own vetting, and closes off almost all outside vectors of attack. Your uncle could visit the same website and be reasonably safe since nothing he clicks would allow for software installation. To a security-adjacent tech worker like me, such restrictions aren’t as necessary, but for the vast majority of people, such protections are a genuinely game-changing upgrade in safety and quality of life in computing.
I don’t want to undersell the value of gatekeeping app stores here – as someone who works on products adjacent to security, Apple’s app store, in particular, is so incredibly good that frankly, most people would be most secure using iPads as their primary personal computers. To be able to arbitrarily select any app in the store and download it, with no fear of malware corrupting the computer or stealing sensitive data, is pretty much the utopian scenario for most people using computers.
Unfortunately, platform holders gatekeeping the software a user is allowed to install is the only way to maintain that level of security, ease of mind, and simplicity of use. Android adds several speed bumps compared to Windows to make self-harm more difficult, but even so, the platform suffers from a far more problematic malware and security situation than iOS. There is a reason why Android has trended towards tighter Google control and more emphasis on the Google Play Store over time. Google’s PR for Android 12 promised to loosen app store rules while “maintaining security” but such vague promises don’t figure into this discussion at the moment.
So where does that leave us? Caught between the benefit to users of a locked-down platform and the vast potential of the platform-holder to abuse their powers, what might we do to maximize both market competition and consumer good?
Some Solutions Offered by an Internet Rando:
If operating systems are anything to go by, each of these technology cases is likely to need unsatisfyingly specific technocratic regulations rather than grand sweeping rules. If we stay focused on Apple’s case, here are some potential options:
Separate the body which writes app store terms and conditions into an independent entity, separate from Apple business chains of command. Give this body appeal and overrule power on individual approval decisions, in a public decision-making process. I’m not sure how I feel about Facebook’s independent council as far as speech goes, but for the app store, it might work. Alternatively, delegate the setting of these standards to the state.
Require all default applications to be open to user modification (i.e. do not force users to use Apple services over third parties).
Ban promotion of products and services in OS surfaces or require that these be open to third parties.
Require all device and platform APIs used for applications (outside of the system core like perhaps the settings app) to be available to third parties – no secret 1st party APIs that perform better or provide special functionality.
It would be satisfying to give a catch-all answer for antitrust in technology. I’m personally inclined towards universal rules with simple terms; the complex regulatory state in the United States more often than not lends itself to kludgeocracy. Unfortunately, the tradeoffs involved in many tech businesses do not lend themselves to such easy answers. I’ve focused primarily on software platforms here. While some of the broad contours of the case I’ve made about operating systems can also be extended to other platforms like browsers, switch domains (to search, for example) and you could conceivably end up with very different-looking problems and solutions.
When it comes to platform-holders restricting freedom of installation on operating systems, restrictions such as Apple’s do create genuine and important benefits for consumers. However, Apple’s incentives as a player in the markets of its own creation are adverse for competition. Right now governance and regulation of that market rests in the hands of private profit-making interest; perhaps it is time for the state to assume that responsibility for the public interest instead.
If I had to extract one higher-level takeaway from all of this: the philosophical core of American antitrust law, dating back to the Sherman Act, does not seek to ban the attainment of enormous market power a priori. It does, however, seek to prevent abuses of that accumulated market power, in order to create and defend a competitive marketplace. It was a good notion then; it is still.
Note 1: For the non-technical, a collection of technology and subsystems necessary to create a solution, which can be visualized as a stack of such systems. For example, hardware components rest towards the bottom of the stack, with firmware on top, then the operating system, then the applications atop the operating system, then applications built for those applications, etc.
Note 2: This is where I need to stress extra hard that I’m absolutely not speaking for my employer here, which is a multibillion-dollar company full of great people who interviewed me with a life-size kangaroo cutout wearing like 10 flower garlands. This refers more to prominent indie devs in the Apple ecosystem like Steve Troughton-Smith who’ve sounded the alarm about Apple’s control before
Note 3: This “bug” persisted for around 3 years – which is to say, Apple knowingly allowed a user experience that would break texting your iPhone-owning contacts if you ever switched away to another mobile OS. Three years is glacial in the software world – enough for 3 major iOS releases.