The practical doctrine
A thesis without a doctrine is just aesthetics. This is the operating stance we take on our own work, cover in our publications, and use to evaluate everyone else's.
Use centralized systems, but require exit
Use centralized systems where scale creates real value, but require competition, transparency, portability, and exit. The right posture toward the frontier is not rejection and not devotion. It is leverage with a clear off-ramp.
Push intelligence to the edge
Push useful intelligence toward the edge through open models, local compute, interoperable tools, and user-controlled memory. Every task that can run at the edge and does not should be treated as an unresolved dependency.
Keep humans as principals
Keep humans as principals and AI systems as delegated agents with bounded permissions and visible audit trails. Agents that act on your behalf must be legible to you and revocable by you.
Turn productivity into ownership
Turn AI productivity into broad capital ownership, not only cheaper consumption. The people building the machines should end up owning the machines, and the people using them should end up owning the outputs.
Defend the primitives
Defend encryption, self-custody, permissionless innovation, open protocols, and the right to run intelligence locally. Every one of those is a load-bearing pillar of sovereignty at the individual level.
Teach the skills
Teach people to direct systems, judge outputs, allocate capital, and build assets. The scarce skill of the next decade is not prompting. It is judgment under abundance.
Measure what matters
Measure progress by the number of people who gain real agency from the singularity. Everything else is a proxy.
"Use centralized systems, but require exit. Push intelligence to the edge. Keep humans as principals."
Where this shows up