Orchestrated Multi-Model AI System

← Writing · all topics

Method

How the work is done and how to read it. 9 posts, best read in order.

  1. 01

    Status labels as a discipline

    Six labels, applied to every claim. The most useful thing in this project is not a result, it is a habit.

  2. 02

    What a no-go theorem actually buys you

    A closed door is a result. Three of the strongest outputs here are proofs that something cannot be done.

  3. 03

    Preregistration for theory

    Writing down the falsification threshold before running the analysis, so that a null result still counts.

  4. 04

    Auditing your own theory with a second model

    Pointing a different agent at your own framework and telling it to recompute every number from the printed formulas.

  5. 05

    Minimum description length as a theory-selection principle

    Prefer the theory that compresses the data and pays for its own parameters. It is Occam with an invoice.

  6. 06

    Prediction, fit, and numerology are three different things

    The distinction that this programme kept blurring, and the checklist I now use to keep them apart.

  7. 07

    How to read a claim that carries a status badge

    A short guide for readers of this archive, so that a benchmark is never mistaken for a derivation.

  8. 08

    Cold tests: killing your own result first

    Running the pipeline on synthetic nulls and known answers before touching real data, because surprise is not evidence.

  9. 09

    The AI Rogue Panic: What the Machines Actually Do—and What We're Getting Wrong

    Collapsing every AI failure into one phrase obscures more than it explains. The distinctions between reliability, security, privacy, and agentic-systems failures are not academic—they determine which solutions actually work.

← All writing · All topics