In this post:
- access is not the bottleneck
- how we can work on it now
Space will be terraformed by hordes of robots. See How Space will be Settled and that means…
Access is Not the Bottleneck
Musk is wrong — access is not the limiting factor to developing space, it’s self-reproducing smart machines.
What do we need to have hordes of robots terraforming space? It’s a long list that has been discussed since Von Neumann proposed the Von Neumann probe, but here are a few ingredients:
- a Von Neumann universal constructor, like a ribosome for robots
- end-to-end labor chain collapse from mining to manufacture — implies problem solving and real-time learning
- meshed intelligence for coordination and logistics
I believe our fundamental bottleneck is the lack of this perspective and how that shows up in our approaches to machine intelligence. See The Fallacies of AI for more.
Working on Space on Earth
Why get red in the face trying to brute force space with human settlers when we can work on self-reproducing robots on Earth creating prosperity along the way?
What would be most useful is building end-to-end labor abstractions for analogous industries like deep sea mining. Then retrofit the meta process of developing that process to the space environment.
The best way of going about building these abstractions is focusing on no-data learning and self-learning systems that can be aligned towards a given objective. In order to do this effectively we need to Embrace Alien Machine Intelligence and build the kernel of self-reproducing systems in the digital for the fastest possible cycle time.
See The “Rocketship” Era — Going Big on Startup v2 for my plan to bring this about.