From Sci-Fi to Specs: Reverse-Engineering the Future
How I use science fiction as a design tool — starting from a vivid picture of the future and working backwards to identify the technical bricks that need to exist.
The method
Most engineers start with what exists and try to improve it incrementally. I do the opposite. I start with a clear image of what the future should look like — informed by science fiction, neuroscience, and raw intuition — and work backwards to identify the technical components that need to exist.
This isn’t daydreaming. It’s a structured decomposition process.
How it works in practice
Take BrainNet as an example. The vision is clear: humans and machines exchanging mental representations directly, without the lossy bottleneck of language.
Working backwards from that:
- You need a shared representational format → World models and global workspace architectures
- You need hardware that reads and writes to biological brains → Brain-computer interfaces
- You need coordination protocols for heterogeneous agents → Multi-agent systems
- You need the whole system to scale without central control → Swarm intelligence and decentralized consensus
Each of these is a real, active research field. The vision gives you the why and the priority ordering. The research gives you the how.
Why science fiction matters
Science fiction isn’t prediction — it’s specification. When Arthur C. Clarke described geostationary satellites, he wasn’t predicting the future. He was writing a spec that engineers eventually built.
The best sci-fi is technically grounded enough to be actionable. Arrival’s depiction of non-linear time perception. Interstellar’s representation of higher-dimensional space. Ghost in the Shell’s cyberbrain network. These aren’t fantasies — they’re design documents for technologies that don’t exist yet.
The gap between vision and execution
The hard part isn’t having the vision. It’s maintaining the discipline to work on the boring, unglamorous bricks that the vision requires. Reimplementing a paper in PyTorch isn’t exciting. Debugging a drone’s IMU calibration isn’t exciting. But these are the atoms that the future is made of.
The vision keeps you oriented. The work keeps you honest.