Systems

The Foundations We Stopped Questioning

AI gets a new headline every week. The phone in your pocket runs on foundations almost nobody questions. Here is why they are so hard to move, and why that matters.

· Updated · 6 min read

The Foundations We Stopped Questioning

AI is moving very fast right now. There is a new model, a new feature or a new company almost every week, and a lot of it is real progress. I follow it like everyone else.

But lately I keep thinking about the other technology in my day. The stuff I don't think about at all.

The operating system on my phone. The apps I use to talk to people. The map that gets me home. These things get updates all the time. New features arrive, the design gets cleaner, the animations get smoother. What almost never changes is the foundation underneath.

This post is about one simple observation, and I'll try to keep it honest: for most people, choosing a phone means choosing between two operating systems, and the reason is less about the operating systems themselves than about everything that has been built around them.

Two choices, and a lot of history

You can check this yourself. StatCounter publishes mobile operating system share every month, and for years the chart has been almost entirely two lines: Android and iOS. Everything else is a thin sliver at the bottom.

That is not a secret, and regulators have looked at it closely. In 2022 the UK's Competition and Markets Authority finished a market study into mobile ecosystems. Its press release was titled "Apple and Google duopoly limits competition and choice." The European Union's Digital Markets Act, which designates large "gatekeeper" platforms and sets rules for them, comes from similar concerns.

I am not interested in blaming anyone here. Both systems are impressive pieces of engineering, and they got where they are by being good. The more interesting question is structural. Why is it so hard for a third option to exist, even when capable people try?

My answer: a phone OS is not really one product. It is a stack of layers, and each layer leans on the others.

The layers that hold a foundation in place

Apps and how they reach you

On a modern phone, most software arrives through an app store. The store is where people find apps, where updates come from, and where trust is handled. Before an app goes live, it goes through review against published rules. Apple's App Store Review Guidelines are a long public document, and Google Play has its own developer policies.

Review has real benefits. It catches malware and scams, and it gives ordinary users a reason to install things without worrying too much. But it also means distribution is tied to the platform. A new operating system does not just need a kernel and a UI. It needs a way for software to reach people that users will trust, and that takes years to earn.

Payments and fees

The store is also the checkout. For digital goods, in-app payment systems handle billing, refunds and subscriptions. Both major stores take a commission on many of these transactions. Apple's Small Business Program offers a reduced 15% rate for developers under a revenue threshold, and Google Play's published service fees include a 15% tier for the first $1M a developer earns each year, with other rates above that.

The exact numbers matter less than the shape. When the platform is both the shop and the cash register, a developer's business model is partly written by the platform. A new OS would have to offer a payment system that users already trust with their cards, and a reason for developers to support one more.

Developer tooling and SDK lock-in

This is the layer people outside software tend to miss. Building for a platform means learning its languages, its frameworks and its tools. iOS work often means Swift and Xcode. Android work often means Kotlin and Android Studio. On top of that come platform SDKs for push notifications, sign-in, maps, payments, analytics and more.

Cross-platform frameworks help, but they mostly target the same two systems. So when a developer asks "should I support a third OS?", the honest calculation is ugly. New tooling, new APIs, new testing devices, a new review process, and a tiny audience at first. Most teams, reasonably, say no. And without apps, users say no too. This is the classic chicken-and-egg problem that economists describe as network effects: a platform becomes more valuable the more people use it, which makes the leader stronger and the newcomer weaker.

Data, purchases and switching costs

Then there is what we carry with us. Photos, messages, passwords, health records, notes, and years of backups are stored in services tied to one ecosystem. Moving them is possible, and it has become easier, but it is rarely painless.

Purchases are another quiet anchor. Apps and in-app items you bought on one platform usually do not move to the other. If you have paid for a few pro apps, a game, and some subscriptions billed through the store, switching can mean buying some of that again. None of this is hidden. It just adds up to a cost most people would rather not pay.

Habits

The last layer is us. Gestures, where the settings live, how notifications behave, which messaging app the family group is on. After years of use, these stop feeling like design choices and start feeling like how phones simply are.

That might be the strongest layer of all, because it doesn't feel like a layer.

Why this is worth thinking about

Put the layers together and you get a system that reinforces itself. Users stay because the apps are there. Developers stay because the users are there. Payments, data and habits make leaving expensive for both. Nobody has to do anything wrong for this to happen. It is what mature ecosystems do.

The CMA study went into these dynamics in a lot of detail, including app distribution, in-app payments and browsers. The DMA has since pushed for things like alternative app distribution and more interoperability in the EU. Whatever you think of those rules, they show that the questions are serious enough for regulators to spend years on them.

My point is narrower, and I think it holds whether or not you care about regulation. When a foundation becomes this settled, we stop asking whether it is the best design. We only ask what feature is coming next. Updates replace questions.

What questioning could look like

I am not saying phones need to be replaced, or that two operating systems is automatically bad. Most of this works, and works well.

But there are real questions here that rarely get asked out loud:

  • If a new platform had a great idea at the OS level, could it ever reach people, given how much sits around the existing two?
  • How much of what we call "the phone" is the device, and how much is the ecosystem attached to it?
  • Which parts of today's design exist because they are best, and which exist because they came first and everything was built on top?

I don't have neat answers to these yet. I think asking them clearly is the first part of the work.

Conclusion

AI deserves the attention it is getting. But the systems we use every day deserve some of that curiosity too. Phone operating systems are a good example because the pattern is so visible: two strong options, held in place by apps, developers, payments, data and habits, each one reinforcing the rest.

That doesn't make them wrong. It does make them easy to stop thinking about.

So alongside "what's the next feature?", I want to keep asking a slower question: is this foundation still the best we can build, or just the one we got used to? Nothing should be beyond questioning. That is what AzureBlaze is here to explore.

References

← Back to writing