FP1 turns heterogeneous evidence about the AI transition into dated readings that can be audited. Each reading separates three things that most analysis collapses: what was observed, what FP1 infers from it, and which options the decision-holder actually gained or lost. Thresholds and falsifiers are fixed before grading. Revisions stay on the page.
Orientation only. Four dimensions sit past a threshold fixed before tracking opened; one sits on it. A position is a measurement, not a verdict, and not a claim about anyone's options. The full reading and its evidence ledger sit on the Radar.
Method in operation The published readings are editorially produced against a documented method. The computational engine for collection, extraction, and model updating is in development and is not running behind the current Radar.
Reading R-013 took a fast financial signal, three slower physical and macro signals, and one piece of reporting that later proved weak, and asked what they changed for one decision-holder. Nothing below is a forecast.
The chip complex gave back more than a fifth of its value over three weeks while disclosed capital plans, operating results, and booked memory supply did not move with it.
The equity move is not sufficient evidence that physical demand reversed. It is sufficient to narrow financing latitude on long-dated plant. Those are different claims on different clocks.
Reversible queue and permitting optionality is preserved. Irreversible capital stays behind project-specific load, financing, and lead-time gates.
Every decision-bearing sentence FP1 publishes can be traced back to an observation and to the transformation applied in between. The right-hand column names who or what is accountable at each step, because that is the question a reader should be able to answer without asking us.
Most dashboards collapse these into one picture, and the picture then carries more meaning than the evidence supports. FP1 keeps them separate on every surface and shows the reasoning that connects them.
FP1 publishes this distinction because the failure mode of every analytical product is a reader who assumes more machinery underneath than exists. The published work stands on its own; the engine is a separate claim and is not yet made.
An editorially governed measurement, grading, and revision method, applied on a published cadence.
A computational layer for parts of collection, extraction, relationship mapping, and model updating.
Each call scores a published AGI scenario against a falsifier fixed before tracking opened. Nothing is graded after the fact and nothing is re-thresholded mid-window. Sweeps at 29 June and 27 July 2026.
The Biological substrate rests on too few constituents to carry a threshold honestly, and the FNC-1 level is not comparable week to week under the current rebase. Both are reported in the Reading with the limitation named, and neither is registered until it can be defended. An instrument that registers everything it can compute rather than everything it can defend fails on its first grade date. Open the Register →
Four standing instruments, a weekly reading, and the methodology papers that document how each was built and the conditions under which it is retired.
FP1 applies five standing analytical lenses to check evidence quality, incentive structure, historical pattern, plain-language translation, and interaction between opposed systems. They are methods with documented procedures, not autonomous analysts, not sources of authority, and not forecasters. FP1 editors own every judgment that reaches the page, and where the lenses disagree the reading says whether it resolved, stands as documented dissent, or is waiting on evidence that does not exist yet.
What follows is FP1's organising hypothesis about the shape of the transition. It is the frame the instruments were built inside, and it is stated here as a thesis to be argued with. Nothing on the Radar or in the Register is graded against it.
Every phase of human evolution is a new mode of entropy reduction: a better way to model, predict, and align with reality. The driving force of civilization is not power, ideology, or resources alone, but the progressive capacity to reduce entropy through increasingly accurate models of the world.
Human history runs in seven phases, and each is far shorter than the last. Each is a qualitative leap in humanity's capacity to model, predict, and respond to reality. What happens when the next one lasts a single generation?
Foraging bands, stone tools, fire, kinship-tribes. Pure trial and error against an indifferent environment.
Awakening sentience in nature, science, and selves. Drawing from complexity science, evolutionary biology, and Emersonian thought, the book examines how economic systems, governance, and technology can align with principles of sentient mutualism to create a more sustainable and just future.
Interested in partnering, commissioning research, or requesting a brief? Send a note and we'll route it to the right person.
Prefer the periodic letter? Subscribe to State of the Transition for the long-frame read.
First Principles First is a research publication. Our analysis is a research product, not investment advice.