published
All Open Source is already Open Spec. The next step is to ship only the spec.
Instead of shipping source code, the Maintainer simply publishes a specification. The Adopter can hand this to their agent, which builds an implementation fitted to the Adopter's system.
Maintainer's POV
Adaptability: Everybody who uses your spec can adapt it to their unique system and their unique use cases, you just have to have the cool idea and build it for yourself and your users.
Security and Maintenance: You don't have to take responsibility for people's security and you don't have to (make your agents) do maintenance on any code for Adopters.
Lower Bar to Publish: You can get your ideas out there without needing to make the code perfect.
Adopter's POV
Adaptability: You are no longer constrained by the maintainers capacity to accept changes, or the community's ideas of what's best for your system. You can adapt the idea to fit what you and your users need.
Security and Maintenance: The code can be totally private to your servers and you can put as much firewall around it as you would like. You don't need to wait for maintainers to ship patches. But yes, the burden of maintenance does shift to the Adopter.
Infrastructure
It runs on the same infra as Open Source...its just git:
- A central repo for the open spec
- A community that is contributing advancements that Maintainers can adopt
- Community can share forks of the spec and Adopters can pick the one that meets their needs
- Maybe some mechanism that scans all forked specs and proposes improvements to the main spec.
Don't get me wrong, Open Source is still immensely valuable! Agent output is non-deterministic, and token production has associated cost. This is merely an augmentation to the paradigm. There's value to having well specced out software that can be quickly replicated by anybody.
