A Netscape decision in the 90s explains why Google and Meta grow up with each technological revolution

In 1995, engineers of Netscape They faced a problem during a development night: how to allow websites to execute code without being able to steal user data?

Thirty years later, its solution, the ‘Same-Origin Policy‘(Policy of the same origin), has become the invisible architecture that governs all the Internet.

Why is it important. Each website became an isolated universe, unable to communicate with others. That night decision explains why we can barely escape the Apple ecosystem, why our data live trapped in silos and why each technological revolution makes the usual giants more powerful.

The context. Alex Komoroske, former strategy director in Stripe and former director of Google for 13 years, He has identified what he calls the “iron triangle” of modern software. System designers can only combine two of these three elements:

1. Sensitive data.
2. Internet access.
3. And non -reliable code.

The logic is simple: if you allow unknown code to access personal data and have Internet connection, you can steal everything and send it anywhere.

The solution was the total isolation. Each application became a fortress where your Instagram data cannot talk to Uber’s, your Apple photos cannot be processed by Google tools, and each service begins knowing zero about you.

In detail. Komoroske Talk about this phenomenon With the water metaphor going down a mountain. Each obstacle does not stop the flow, redirects it where there is less resistance. Over time, channels are formed that attract more water to become increasingly large rivers.

Planning a trip illustrates this mechanism:

1. Flights in the mail.
2. Hotel in another app.
3. Restaurants in Google Docs.
4. Calendar in a different tool.

The constant friction of copying, pasteing and reformating leads to grant access to a single service that already knows all your context. Without friction, everything works perfect. When you share the trip, you use the tool that already has all the information.

The threat. The AI promises to be different, but is inheriting the same physics. The LLMS They can create almost free software – a developer with AI can build in hours what it took weeks – allowing infinitely personalized tools.

But this “infinite software” distributed through traditional stores does not solve our problems: it amplifies them. More applications mean more silos, more places where your data is trapped. The AI needs context to be useful, but our current security model means that sharing context is a commitment of all or nothing.

Yes, but. The technical pieces to transcend this paradigm already exist. Modern Intel, AMD and ARM chips include “safe enclaves“, encrypted and protected memory regions of anyone, including cloud administrators.

AI brings us a unique opportunity, because it makes the current limitation evident. The technical pieces already exist and it is the first time in thirty years that we can transcend this policy. While nothing changes, the concentration of power will continue to reinforce.

In Xataka | What was ATI: to look at Nvidia to end and forgotten by the technology industry

Outstanding image | Netscape, Xataka, Unspash