Calibrated research on space, defense, and energy.
Every forecast gets a probability, a date, and what would prove it wrong. On the date I grade it in the issue — right or wrong — and score it. The record is public and append-only.
Every issue, free, in English and Spanish. No list sharing, unsubscribe in one click. Privacy.
W1 arrives 15 October 2026
W1 · The Contract, and the Master Variable · 15 October 2026
Most space coverage is a relay.
A company announces something, the announcement gets repeated with adjectives, and nobody goes back. There is no scoreboard, so being wrong costs nothing — and analysis that costs nothing when it’s wrong is worth what you pay for it.
Every call is dated
A probability, a resolution date, and a falsifier — the specific observation that would prove it wrong. Written down before the outcome is known.
Every call gets graded
On its date, in the issue, with a Brier score. Including the ones that go badly, and especially those. A scoreboard nobody can check is not a scoreboard.
Nothing gets edited away
The register is append-only. A call that ages badly stays in the file with its original wording and its original date. Corrections are new entries that point back.
Positions come first
I invest in this sector. Exposure and ninety-day intent are declared at the top of every issue, before any analysis — not in a footer afterwards.
Register file → register/register.jsonl — public mirror goes live with W1 · Disclosure → /disclosure
Open forecasts.
Each one gets graded in the issue on the day it resolves, whatever it says. Probabilities are set by hand before an entry goes public.
The register opens with W1. Every entry carries a falsifier and a resolution date, and is graded in public.
I am Kevin Neuburger. I invest in space, defense, and energy, and I advise companies deciding what those sectors mean for them. Before that I was a strategy consultant at Accenture. No employer, fund, or editor sits upstream of what is published here, and positions are declared at the top of every issue.
Reckoning is the research I would want before putting money at risk, written down before the outcome is known. Every forecast carries a probability, a resolution date, and the observation that would prove it wrong, and is graded in public. The register is append-only; nothing is edited away.
How it is made. Every issue is written collaboratively by me and AI tools, principally Claude (Anthropic). I read every issue before it publishes and I am responsible for what it says. Every figure traces to a primary source listed with the issue, fetched and dated; a figure that could not be verified is marked unverified rather than printed. Errors are corrected in the open, with a dated note, never silently. Anthropic is a customer of companies covered here.
The Spanish edition. Every issue ships in Latin American Spanish. I edit that edition myself. Spanish is my second language, so errors can occur; the numbers, tables, and register entries are identical across editions, and a term whose Spanish usage is unsettled is flagged in a translator's note rather than decided quietly.
Every issue, in two languages.
W1 — The Contract, and the Master Variable
Why launch cost binds whoever buys launch retail, and why most of what claims to be gated by it isn’t. Publishes 2026-10-15.
W1 — en español
Full Latin American Spanish edition with every issue. Same numbers, same tables, same register entries.