Gordon Brander on Building Protocols for Thought

Table of content

Gordon Brander builds systems that think. After working on five browsers and an operating system at Mozilla and Google, he founded Subconscious—an ambitious project to create a “worldwide decentralized mind” through open protocols.

His core belief: “Open-ended ecosystems are where value is created. The value comes from the system surprising itself.”

The Noosphere Protocol

At the heart of Brander’s work is Noosphere—a decentralized protocol for thought built in Rust. Unlike traditional note-taking apps that lock your thinking inside company servers, Noosphere treats ideas as first-class citizens that can flow freely between tools and people.

The protocol enables:

Brander also created Subtext, a minimal markup language for note-taking. It strips away formatting complexity so you can focus on capturing ideas.

Deep Future: AI Scenario Planning

Brander now builds Deep Future—an AI scenario planning tool that stress-tests strategies against thousands of possible futures. The system uses methods from RAND and military planning adapted for business strategy.

How it works:

  1. AI agents monitor live signals from your environment
  2. The system identifies forces driving change
  3. Multi-dimensional matrices map how forces might interact
  4. Your strategy gets tested across wildly different futures
  5. AI synthesizes adaptive recommendations

This isn’t prediction—it’s preparation. The goal is building strategies that work regardless of which future arrives.

Evolutionary Thinking in Software

Brander’s newsletter Squishy Computer (8,000+ subscribers) explores software through biology. His digital garden contains 165+ notes on systems, evolution, and strategy.

Some of his ideas:

On composability: Small alphabets of simple elements combine into endless possibilities. Like DNA’s four nucleotides encoding all life, software needs simple building blocks that compose without friction.

On emergence: Complex behavior arises from simple rules. You can’t design emergent systems directly—you design the conditions that let emergence happen.

On requisite variety: A system needs internal diversity matching its environment’s complexity. Monolithic tools fail against varied problems.

These aren’t abstract theories. Brander applies them directly to protocol design, asking: how do you build software that evolves beyond what its creators imagined?

Practical Patterns

From his notes on building systems:

Start with a spreadsheet — Before building software, model your problem in a spreadsheet. If it doesn’t work there, code won’t save you.

Work back from UI — Design the interface users need, then figure out how to build it. Not the reverse.

Modularize later — Don’t abstract prematurely. Let patterns emerge from working code before you generalize.

Gall’s Law — “A complex system that works is invariably found to have evolved from a simple system that worked.” Start simple. Complexity comes later.

Multi-Agent Systems

Brander’s current research explores multi-agent systems—AI that works collaboratively rather than as isolated chatbots. His scenario planning work at Deep Future implements this: specialized AI agents monitor signals, analyze forces, and stress-test strategies in parallel.

This connects back to Noosphere’s vision: just as the protocol lets human ideas flow and combine, multi-agent systems let AI capabilities compose into something greater than individual models.