An open thesis about the natural environment

Building a business is a fascinating strategy game to me. There are so many things to tend to, competing arguments, and limits on time, money and attention. I enjoy learning how to make it work. But that does not answer what I want the work to be for.

I have loved nature my entire life. I want the most interesting technology of my lifetime pointed at conserving and restoring the finite resources we depend on.

Ryan Orbuch's Natural General Intelligence makes a distinction I find useful: automating knowledge work does not replace the physical systems that sustain us. Environmental stewardship needs observations of the Earth and evidence about what our actions change.

Connecting research to a place

My question is narrower. How do we find where a useful technology belongs, learn enough about the site to judge whether it fits, and help the people involved decide what to do?

This is the work I want TinyMacro to do. I want to connect applied research to actual opportunities to conserve resources. The benefit has to be something someone can check. A company looking more environmentally responsible is not enough.

I think agents could make deeper exploratory research practical before anyone commits to a project. That is a hypothesis to test. Research still has to account for local conditions, engineering constraints and the people responsible for the decision.

An index built around evidence

A pile of documents does not tell someone whether a technology belongs in their building. I want the information organized around real things: technologies, sites, resources, suppliers and the relationships between them. That is what I mean by an ontology.

Palantir explains the value of this well in its AIP architecture overview. Its ontology models the nouns and verbs of a company's operations, the objects it works with and the actions people take on them, in a form both people and agents can read. Data, logic and actions live in one place, so an agent works from the same picture a person does, and every action can be traced back to the data behind it.

I want that for the natural environment. The nouns are buildings, sites, technologies and resources. The verbs are the decisions people make about them.

A proposed decision would connect a technology to a site and retain its sources, assumptions and unanswered questions. Someone reviewing a calculated number should be able to see its inputs and method. Conflicting evidence should remain visible.

The ontology could also be the index. Finding a technology would lead to where it has been used, what those applications required and what evidence supports the comparison. Useful research would have somewhere to accumulate beyond a conversation history.