The Default
Is the Moat.

I started with a small, irritating question: Why does choosing a computer operating system still feel like choosing between Windows, macOS, and Linux?

There are other operating systems, of course. But when you need your applications, your hardware, and your work to function without friction, the theoretical list of options gets much shorter.

I found myself thinking: If Linux can be free and capable, why hasn't it displaced Windows? Why doesn't a better alternative simply win?

Then the question became more interesting. Maybe being better isn't enough. Maybe some markets are designed, gradually and without any single designer, to keep their winners winning.

The hidden cost of choosing differently

People don't buy an operating system in isolation. They buy access to the applications they need, support from the people around them, familiar habits, and confidence that tomorrow's work will still work.

That means the cost of switching is not just the purchase price. It's everything a person risks losing on the way out.

Why Windows became the default

Windows didn't win only because of its features. Microsoft's licensing and distribution through PC manufacturers gave it extraordinary reach. Its installed base then encouraged software developers to build for Windows first, which made Windows more useful to the next buyer.

The US Department of Justice called this the applications barrier to entry. The more software existed for Windows, the harder it became for another operating system to offer a practical substitute.

And this wasn't merely an innocent advantage of scale. US courts also found that Microsoft had used exclusionary practices to protect its operating-system monopoly. The historical distinction matters: network effects can arise naturally, while firms can still exploit them unfairly.

My frustration wasn't really about Windows. It was about the strange feeling of having a choice on paper and very little choice in practice.

A model I'm using to think about this

The Default Loop

Here's how I now describe the mechanism. This is my working synthesis of network effects, distribution, switching costs, and behavioral inertia, not a new economic law.

The loop isn't inevitable, and not every market works this way. But when it does, the leading product can keep winning without winning every head-to-head comparison.

At some point, a product stops being just a product. It becomes an assumption.

This is also a marketing problem

Marketing is often described as persuading someone to choose you. In established markets, a harder task comes first: getting someone to reconsider the choice they no longer actively make.

Think of a business that keeps its CRM not because anyone loves it, but because the automations, reports, training, integrations, and team habits are already built around it. A replacement can look better in a demo and still be the worse decision after migration costs.

Psychology reinforces the structure. Familiarity makes a choice feel safer. The possibility of losing what already works can loom larger than a promise of gaining something new. And the apparent popularity of an option becomes social proof, even when popularity tells us little about fit.

So distribution isn't only about reaching attention. It can shape what people consider normal, credible, and even possible.

This changes how I think about differentiation. Being unique is not enough if adopting your uniqueness requires the customer to rebuild their world.

Could AI break the loop, or build a new one?

An operating system became valuable partly because applications were built for it. What happens if an AI agent can operate tools on behalf of a person, across different software environments?

If a useful task can be completed without learning five interfaces, then those interfaces may lose some of their hold over the user. Open integration protocols such as MCP are one piece of that possibility: they give AI systems shared ways to connect with tools.

But this is a hypothesis, not a solved problem. Agents still need permissions, reliable integrations, trustworthy results, and safe handling of mistakes. A Windows-only application does not magically become portable because an AI can click around it.

There's also a darker possibility: we replace operating-system lock-in with agent lock-in. The assistant that remembers our preferences, coordinates our work, and controls our tool access could become a default that's even harder to leave.

The useful question: What would make leaving easy?

I don't think the right way to challenge a powerful incumbent is always to build a better copy of its product. Sometimes the better move is to remove the dependency that protects it.

If I were examining a locked-in market, I'd start with four questions:

1. What is the default? Name the choice customers make without reconsidering.
2. What makes it sticky? Separate compatibility, data, training, distribution, contracts, and habit.
3. Which dependency can be bridged? Make one important workflow work without forcing a total migration.
4. Can the bridge earn trust? Test with a real user and a real task, not just a persuasive demo.

That's the part that interests me as someone who designs marketing and automation systems. Not another grand platform that claims to replace everything. One well-chosen connection that gives a person or business a choice they didn't effectively have yesterday.

I don't know whether AI will free users from old defaults or produce stronger new ones. But I know which question I want to keep asking:

What are we calling a necessity today that is really just a dependency nobody has made easy to escape?