Gordon Brander on Living Software

Table of content

Gordon Brander builds software that surprises itself. A founder who previously worked at Google and Mozilla on five different browsers and an operating system, he now explores how to create open-ended software ecosystems that evolve and grow on their own terms.

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

Noosphere: A Protocol for Thought

Brander’s most ambitious project was Noosphere, a decentralized protocol for tools for thought. Built in Rust with Swift and TypeScript bindings, it was the foundation for the Subconscious app—an attempt to create “permissionless multiplayer” knowledge management.

The project drew from p2p networking, PKI, and content-addressed data structures. While Noosphere was archived in September 2024, it represents a serious technical attempt to solve how personal knowledge tools could work across distributed networks without centralized control.

Deep Future: AI Scenario Planning

Currently, Brander is building Deep Future, an AI-driven scenario planning tool. It uses multi-agent systems to stress-test strategies against thousands of possible futures.

The approach borrows from RAND and military scenario methods: AI agents monitor live signals, uncover forces driving change, map possibility spaces, and playtest strategies through simulation. The goal is to help organizations spot leverage points before disruptions hit.

Squishy Computer: A Newsletter on Living Software

His newsletter Squishy Computer has over 7,000 subscribers and explores software through the lenses of evolution and ecology. It’s less about coding tips and more about the systemic properties that make some software ecosystems thrive while others collapse.

From the newsletter’s about page: “I’m interested in software ecosystems that are organic, open-ended, and evolving. The internet is like this, and so is the web. My question is, how can we make more software like this?”

Generative Composition with AI Agents

One of Brander’s most useful frames for working with AI comes from Brian Eno’s generative music. In a January 2026 newsletter post, he draws the parallel:

“It seems to me that AI agents turn all of us into generative composers. Instead of writing the code, we write a generative score for agents to write the code. Instead of playing the music, we’re playing the system.”

He borrows from Stafford Beer: “Instead of trying to specify it in full details, you specify it only somewhat. You then ride on the dynamics of the system in the direction you want to go.”

This isn’t about losing control—it’s about working with variety as creative material. Traditional programming gives us full control because compilers are predictable. AI agents introduce genuine variety. Brander suggests we treat that variety as opportunity rather than bug.

Practical Patterns

His digital garden contains 165+ notes organized by systems thinking tags. Some recurring patterns:

The Dwarf Fortress Approach

Brander has an unusual reference point for the future of creative work: Dwarf Fortress. In the game, you don’t control individual dwarves—you order general activities and let the simulation play out.

He asks: “Maybe the future of creative work looks more like Dwarf Fortress?” The implication is clear: as AI agents gain capability, our role shifts from direct execution to system design. We become conductors, not instrumentalists.