Sunday, October 23, 2022

Indian antitrust ruling: "App developers are super dependent on Google" -- device makers choose "between signing a non-negotiable [contract] and commercial failure" -- consumers are "locked-in to [Android]"

On Thursday, the Competition Commission of India (CCI) announced an antitrust ruling against Google, and the following day, the full 293-page order was published. I've read it in full, and even though it's a long read (despite generous spacing), I recommend it to everyone with a professional interest in Google--particularly Android--antitrust issues.

Hierarchical summary of remedies

  • device makers' (OEMs) rights:

    • (i) no more "package deal" for preinstallation of Google apps (such as search, Maps, YouTube, and GMail), but à la carte choice for OEMs and free arrangement on home screen

    • (ii) no more tying of other Google services and products to the must-have Google Play Store

    • (iii) no more denial of Android forks' access to Play Services APIs (Maps, Cloud Messaging, In-App Purchasing, Google+, and others) ("forks" means Android versions deviating from Google's specifications and requirements)

    • (iv) no more exclusivity agreements (with or without incentives) between Google and OEMs favoring its search service

    • (v) no more anti-fragmentation obligations preventing OEMs from (additionally) making devices that run Android forks

    • (vi) no incentives or obligations preventing OEMs from (additionally) making devices that run Android forks

  • Android users' rights:

    • (vii) no restrictions on deinstallation of preinstalled apps

    • (viii) users must get to choose their default search engine on all entry points starting with the initial setup of a device, and those default settings must be modifiable later, requiring as few user interface actions as possible

  • app developers' rights:

    • (ix) the Google Play Store must carry the storefront apps of third-party app stores

    • (x) no restrictions on app developers' ability to distribute apps through sideloading (the CCI is aware of the fact that Google allows it in principle, but also knows that users are at this point simply discouraged from installing sideloaded apps)

In other words, if Google ultimately had to comply with the order, its Android business model--which has always been primarily about monopoly maintenance and the exploitation of the Android app distribution monopoly--would no longer work. As Google told the CCI, "Google would need to charge a license fee for Android and/or GMS [Google Mobile Services] apps in India as it does in Europe." So the cost of Google-licensed Android devices in India might increase, but there would be--which would ultimately bring down prices and/or spur innovation--much more competition (at least in India) in

  • (licensable) smartphone OSs (device makers could make Android-compliant and Google-licensed devices on the one hand and different (though not necessarily different on the hardware side) devices running "forks" on the other hand),

  • general web search services,

  • video hosting platforms (see my commentary further below on that market definition), and

  • Android app distribution (alternative app stores and sideloading would represent viable alternatives to the Google Play Store).

Five relevant markets (each of them geographically limited to India)

  • (a) licensable operating systems (OSs) for smart mobile devices (smartphones, tablets)

    Google offering: Android

  • (b) app stores for Android

    Google offering: Google Play Store

  • (c) general web search services

    Google offering: Google search engine, related services

  • (d) non-OS-specific web browsers

    Google offering: Chrome

  • (e) online video hosting platforms (OVHP)

    Google offering: YouTube

I concur with market definitions a (licensable OSs for smartphones and tablets) and b (Android app distribution), and those parts are consistent with the European Commission's and Epic Games' approach, which I've previously declared myself in agreement with. There can be no doubt about market c (general web search services).

I can see why the CCI believes its market d (non-OS-specific web browsers) makes most sense in this particular India-specific case, where Safari is irrelevant because hardly anyone buys Apple devices there and Chrome faces competition from a cross-platform browser made by Alibaba and named UC Browser, but I wouldn't necessarily endorse that definition in other geographic markets.

Where I have my doubts even with a view to India is market e (online video hosting platforms), an arguably gerrymandered definition under which YouTube's only significant challenger would be Vimeo and which may take the TikTok phenomenon into account to an insufficient degree. TikTok has features of a social network and of a video hosting platform (as opposed to a platform that doesn't offer user-generated content), and there may be some indications by now that TikTok's popularity comes at the expense not only of some other social networks but also YouTube. Also, TikTok keeps evolving and widening its impact zone. In India, it was banned two years ago along with dozens of other Chinese apps, though many users still sideload it and others use alternative services. But the YouTube-specific part of the CCI's decision is the one I'm less interested in (and which other regulators than the CCI don't seem to consider a priority either).

Comparison to other jurisdictions' approaches

This is the most comprehensive Google Android antitrust case I've seen so far. In other jurisdictions, this would be enough for two or three cases, if not more.

There are definitely synergy effects. The CCI decision has to explain certain dynamics only once, and can then build on that foundation.

I'm not suggesting that one approach is better than the other. What the CCI has put together is impressive in terms of breadth, depth, cogency, and clarity. The people who work on that case did a great job. But they were standing on the shoulders of giants as the CCI order references what some call "the Cicilline report"--the Majority Staff Report and Recommendations of the Antitrust Subcommittee of the U.S. House of Representatives (PDF). And while it isn't officially referenced, I assume that the CCI officials working on their Google case were well aware of the European Commission's Google Android decision, which the EU General Court largely affirmed last month.

It's possible that antitrust watchdogs in other jurisdictions opted for a piecemeal approach (instead of a sweeping ruling) to err on the side of caution. Also, DG COMP and others may feel they need to underpin their findings with even more detail in order for their decisions to withstand judicial review.

Country-specific factors

When comparing the comprehensive Indian Android case to parallel cases in other jurisdictions, it's important to understand some key characteristics of the Indian market:

  • Android has a market share of about 97%.

  • Apple is vanishingly small in India. It's a price-sensitive market and, as the CCI decision notes, Apple's products are simply too expensive for most customers there. In fact, if Apple's worldwide market share was like the one in India, app developers would deprioritize iOS and Apple would struggle even to survive in a niche.

  • Google's search engine also has a market share around 97% in India. So the CCI had to look at one monopoly reinforcing the other.

  • Many of the devices sold in India are also made in India by local companies like Micromax and Karbonn.

"Best Of" Collection

There's a lot that I like about the CCI ruling. It reflects a profound understanding of the interplay of the different elements of Google's strategy, arguing that "the impact of Google’s practices [...] cannot be seen in isolation, but cumulatively through web of restrictions pursuant to multiple agreements."

In no uncertain terms does the decision express the agency's disagreement with some weak arguments and transparent pretextual justifications. Let me show you some real gems:

  • "The so-called choice for OEMs that Google refers to is between signing a non-negotiable MADA [Mobile Application Distribution Agreement] and commercial failure."

  • "[F]ragmentation can be a source of competition and innovative products, as confirmed by the fact that Google itself created Android by breaking compatibility with Sun Microsystem's Java."

  • "Google utilizes these so-called ‘security measures’ to create a barrier around its basket of apps which lets them flourish at the cost of competitors." (quoting from an app developer's submission)

  • "Surveys can be indicative or misleading on occasions, data cannot."

  • "App developers are super dependent on Google for distribution and reach of their Apps."

Chart: three-sided Android ecosystem

On page 42 of the CCI decision you can find the following chart that explains the three sides of the Android ecosystem (click on the image to enlarge):

For a recent post on Epic Games' aftermarket arguments in the Apple case I drew up a similar chart (click on the image to enlarge):

Obviously, the CCI didn't "steal" anything from me, nor the other way round. Those charts are two independent creations.

Observations on responses from third parties

Amazon explained to the CCI how its own Android fork named Fire OS couldn't compete. Given how powerful and sophisticated Amazon is, that fact bears significant weight.

Microsoft shared its experience with how it is now practically impossible even for them to compete with iOS and Android in the smartphone OS market due to the "app gap." Microsoft also discussed the challenges facing its Bing search service.

Some Indian app developers explained their problems very well. Those are real app developers, not the fake developer lobbying fronts that Apple (ACT | The App Association) and Google (Application Developers Alliance) use.

Android OEMs mostly tried to keep clear of coming across as de facto complainants. Some provided minimalistic answers while others explained their problems in ways that were distinguishable from a complaint, but have helped the CCI understand the issues.

Revenue-sharing agreements

In Western jurisdictions it has recently been difficult for antitrust enforcers to make cases against revenue-sharing and other types of exclusivity agreements. The CCI, however, is undeterred by that fact. The following sentence is food for thought also in other jurisdictions:

"In digital markets strong network effects makes the application of as efficient competitor test difficult."

The CCI explains quite convincingly that "OEMs would not have any incentive to pre-install competing search services" because "in one instance, the competing search service has to offer 90% revenue share to the OEM to secure a default position on the secondary browser, as against 10% offered by Google (generally). This is un-sustainable for the competitors."

I don't want to take a firm position on the applicability of the as efficient competitor test to digital markets here. But if the numbers in that example given by the CCI are right and a competing search engine would have to offer a 90% revenue share in order to compete with a Google offer of a 10% revenue share (because Google's search engine is also a much stronger economic engine), it's obvious that under the traditional AEC test Google could maintain its monopoly for all perpetuity by doing deals with all the major access points.

Google's huge payments to Apple are mentioned by the CCI as evidence of how much value Google derives from being the default search engine, but given Apple's tiny market share in India (compared to which its investments in manufacturing in that country are really huge), I can see why the focus is now more on what Google does alone than what it does in collusion with Apple.

Differentiated perspective on fragmentation

When I saw the CCI press release, I was worried that the negative take on anti-fragmentation rules went too far. The actual decision is nuanced and differentiated. Some anti-fragmentation rules are acceptable, but fragmentation must not serve as a pretext for monopolization.

As a small app developer I just wish to note that for the most part, we have the same interests as OEMs in loosening Google's death grip on the ecosystem, but I would hope that regulators and courts would also take into consideration that the little guys will struggle with a degree of fragmentation that may still be profitable for a major device maker and the very largest app makers.

Incomplete understanding of switching costs

There is really just one sentence in the CCI decision that in my eyes doesn't reflect as favorably on the CCI's understanding of the issus as the rest:

"Thereafter, the end consumer is locked-in to the OS and faces substantial switching costs, primarily in terms of cost of new smart device."

Switching costs are a big issue in all those mobile platform cases, such as the EU Commission's Google Android case and Epic Games' lawsuits against Apple and Google. The CCI is right about lock-in and substantial switching costs, but it's not a good idea to focus too much on just the cost of a new phone or tablet. Users are locked into those operating systems, not so much into devices. Devices are replaced from time to time; many people's upgrade cycle is two years (which is when mobile carriers typically offer a discount on a new phone if one renewes for another 24 months). It's about apps (for lack of a multi-platform app store like the ones that Epic Games and Microsoft--as know from a recent filing with the UK's Competition & Markets Authority-- would like to operate on mobile devices) and media content. It's also about familiarity with user interfaces. It's about having to re-enter lots of passwords. I migrated from Android to iOS and back, so I know that purchasing a device is really just a limited part of the picture.

In another context the CCI order at least recognizes the part about having to repurchase apps:

"[T]he users of smart mobile devices in India face considerable switching cost to shift to iOS between Android and iOS (and the need to download and purchase existing apps for the new smart mobile OS)."

Given that Apple's products aren't much of an option for most customers in India, it's understandable that the CCI didn't have to go into a lot more detail concerning switching costs. That's different in other jurisdictions. Let there be no doubt that the CCI order is a great piece of work. Someone will always find something to disagree with or identify room for improvement. That changes nothing about what a world-class job those CCI officials did. I hope they can defend at least the most important parts of the ruling in the further process.