A world model at the core of Pipi 9

Mike's Notes

Here are some quick notes for Alex explaining how the first 19" server rack, which is currently being built and commissioned, hosts internal models. As the number of racks grows, more models can be hosted.

This is highly experimental. I shall report on results.

Resources

References

  • Reference

Repository

  • Home > Ajabbi Research > Library >
  • Home > Handbook > 
  • Home > pipiWiki > Industries

Last Updated

16/05/2025

A world model at the core of Pipi 9

By: Mike Peters
On a Sandy Beach: 13/05/2025

Mike is the inventor and architect of Pipi and the founder of Ajabbi.

The problem

The world is not a simple place. Any large complex system becomes dynamic, has emergent properties, and has unintended consequences. They are hell to build and update.

Large IT projects;

  • 15% succeed
  • 25% makes no difference
  • 65% fail
The annual global cost of new IT system failure is $USD 3,500,000,000,000 (the size of France's annual GDP) — IT Revolution.

Solution

Just copy how nature does it. Every living biological cell is incredibly complex and self-managing in response to its environment.

Pipi 9

Pipi maintains many models of the world at all levels of reality.

Autonomous agents

Each Pipi instance contains and is driven by autonomous agents.

Swarm intelligence

Each model is curated by an instance of Pipi, which is adaptive in response. Each instance is uniquely internally named in ASCII after one of the old deities from human history, e.g., "Loki," "Thor," or "Niltsi." This will be easier than remembering internal  IP addresses. If the list of names available runs out, then the names of gods from works of fiction could be added, eg, The Lord of the Rings, Game of Thrones, etc. If even more names are needed, there are always ice cream flavours, rose varieties, racehorse names, etc.

Pipi 9 instances function as swarms, similar to a flock of birds. There can be many swarms, and the swarms can also interact.

World model

The world model contains the laws of physics, properties, and other objective constraints of the real world.

Other models

Models can automatically import human-derived ontologies and standards. Only humans apply the rules and parameters. The models are stateful and keep their histories. The connections between models are fluid and dynamic. Some multiple inheritance is also present, which is too difficult to show here in this simplistic outline. 

Self documentation

All assumptions and configurations will eventually be openly viewable on pipiWiki, GitHub, via API, etc. This is a work in progress as I sort out the self-documentation generated by Pipi 9. The problem is mainly how to automatically present a lot of versioned information in multiple languages on web pages, with the required navigation.

Rendering

Once the first server rack (completely isolated from the internet) is operational, documentation rendered in static HTML can be scheduled, copied to a memory stick (analogue security barrier), and then uploaded by FTP to the pipiWiki website. If I remember to do it, it will be monthly at first. Later, I will automate every process and build a team to help with the work.

How it might look

Below, a crude, simplified outline shows examples of the relationships between internal models in Pipi 9. There is a "kind of" hierarchy of models with updates from above, and bubble-up learning from below, all automated.

  • World Model
    • Aviation Model
    • Education Model
    • Health Model
      • UK Health Model
        • UK NHS Model
          • Birmingham Children's Hospital Model
          • Birmingham Women's Hospital Model
        • Circle Health Group Model
          • Albyn Hospital Model
    • Rail Model
    • Sewerage Model
    • Stormwater Model
    • etc

Summary

The whole setup is designed to be fully automated, autonomous, fluid, dynamic, and ever-evolving, but also lawful, unlike LLMS. This is subject to experimentation, change and has much more detail to come, but you get the rough idea.

No comments:

Post a Comment