Method & sources
Open model, open assumptions, open to challenge.
Every figure on this page is generated from assumptions.json by a ~150-line Python model, published in full in a public repository. Each input carries a provenance tag: measured from a published source, derived computed from measured inputs, or assumed — a scenario judgment explicitly flagged for challenge. The repository link is intentionally not published directly on this page — see the anonymity note below for why — but is available on request through the Feedback or Contribute links on any page.
Primary sources include Statistics Canada, the Canada Energy Regulator, Budget 2025, the Major Projects Office, the Canada Infrastructure Bank, IEEFA, ACER, the World Nuclear Association, USGS, JERA/METI, the European Commission (CRMA, CBAM), and bank economics teams (RBC, TD, Scotiabank). Four full research briefs with source URLs live in the repository's data/research/ folder.
Found an error? Disagree with a fact or argument? Every input on this site is flagged with its provenance so it can be checked and challenged — that's the point of publishing the model in the open.
The technical apparatus, in one place
Key technical apparatus: a project-by-project sources-and-uses ledger (21 named projects, theme-differentiated financing-source weighting calibrated to real sector precedent — e.g. LNG Canada's sponsor-JV structure vs. BC Hydro-style provincial rate-base financing — with illustrative tenor/DSCR by theme, where DSCR is "debt service coverage ratio," a lender's measure of how comfortably a project's cash flow covers its debt payments); a structural gravity-model grounding for sector diversion-share assumptions (Cipollina & Salvatici 2010 meta-analytic RTA elasticity ~30-45% central estimate net of publication-bias correction, Disdier & Head 2008 distance elasticity ~-0.9 to -1.1, discounted against Canada's own CETA ex-post preference-utilization rate of 49-70%, cross-checked against the 2025-26 realized trade-diversion shock); a trade-by-trade labor/wage-escalation model (BuildForce Canada CLF 2026-2035, national net shortfall 34,300 workers, reconciled against the superseded 2025-2034 edition's 108,300 figure); an evidence-tier convention (SIGNED/PROVEN, STRESSED, CONSTRUCTIVE) applied uniformly across energy-swap, auto-pivot, and all post-v0.4 modules; and a six-stage US-counter-move game tree informing the sequencing doctrine.
How this was built, and the use of AI tools
This plan was directed, authored and decided by a small group of human contributors — the specific numbers, priorities, trade-offs and judgment calls throughout the site are theirs, not a machine's. AI language-model tools were used extensively alongside that human direction: to research and cross-check public sources, draft and redraft prose, build the Python model that turns assumptions.json into every figure on this site, and search for and verify facts against primary sources (government announcements, company disclosures, academic papers, news reporting) before they were written down. Every AI-assisted claim was reviewed against the site's own provenance discipline (measured/derived/assumed, described above) by the human authors before publication, and remains open to challenge by any reader for exactly that reason.
We are disclosing this deliberately and going further than any law currently requires. We checked: the EU AI Act's Article 50 transparency rule (effective August 2026) requires disclosure for AI-generated text on matters of public interest, but explicitly exempts content that has undergone human review with a named party holding editorial responsibility — which describes this site's process. California's AI Transparency Act (SB 942, effective August 2026) and New York's synthetic-performer disclosure law both apply only to AI-generated images, video, audio and synthetic on-screen performers, not text — this is a text-and-data site with no synthetic media. Canada currently has no comprehensive federal AI-disclosure law (the proposed Artificial Intelligence and Data Act died at prorogation in January 2025 and no replacement has been introduced as of this writing). None of that changes our answer: we would rather over-disclose than leave a reader to guess, on a site whose entire premise is that a claim should be checkable, not taken on faith.
Why this site is anonymous. This plan proposes measures — export levies, technology-transfer conditions, a public-diplomacy campaign that names specific US officials' false claims, positions on an active and politically charged trade dispute — that could expose named individual authors and contributors to real personal and professional risk, particularly given how heated cross-border rhetoric has become. Publishing anonymously protects the people who wrote and contributed to this, not the substance of the plan itself: every figure, source and assumption is published in the open exactly as it would be under any author's name, and remains fully checkable and challengeable regardless of who wrote it.
Anonymous does not mean unmonitored or static. The Feedback and Contribute links on every page reach the same small group of human authors directly, and are checked regularly, not occasionally. This site is updated at least weekly as new information, corrections and contributions come in — treat what you're reading as a living document, not a one-time publication, and treat a stale-looking figure as worth a Feedback submission rather than an assumption that no one is watching.
Contact the authors
There are exactly two ways to reach the people behind this site — no other channel or address is authorized to speak for it. Use the Feedback button (top-left corner of any page) to flag an error, disagree with a figure or argument, or raise a question directly. Use Contribute to the plan (top-right corner) to submit an idea, analysis, correction, or supporting document for review, with a tip line specifically for anything relevant to the five accountability-investigation targets described on the Public diplomacy & advocacy page. Both are checked frequently, not occasionally. Whether and how the authors respond to any given submission is their own case-by-case call, based on their own assessment of the request — a reply is not guaranteed, and the absence of one should not be read as agreement or disagreement with what was raised.