What this is,
and what it refuses
to do.
The Relative Decline Project measures how fast the major powers are losing or gaining relative position, and prices the conditions under which that contest turns violent. It lives at worldwar3.army, which is a deliberate choice and a demanding one.
The domain
A domain that promises apocalypse and delivers charts reads as clickbait. A domain that promises apocalypse and delivers rigour reads as authority. The distance between those two outcomes is the entire editorial strategy of this site, and two rules close it.
The site prices conflict. It does not predict it. Every claim about fighting is a conditional with named triggers and a published probability band. “Here is what would have to be true” is a statement that survives contact with 2027. “Here is what will happen” is not.
The masthead is the argument, not the domain. This is the Relative Decline Project. worldwar3.army is where it lives. The URL carries the register; the content has to carry the weight.
Editorial standards
These are published here so readers can hold the site to them.
- Every figure carries a date, a source and a confidence grade. No exceptions, including for figures that are obviously true.
- Every chapter carries a steelman of the opposing case, written to persuade rather than to be knocked down. If the counter-case on a page reads weakly, that is a defect in the page.
- Every forecast carries a resolution date and is graded in public, including the misses, with the original text unedited.
- No prophecy. Conditionals and probability bands only.
- Single-source claims are labelled as such and are never presented alongside multiply-confirmed ones without the distinction being visible.
- Ranges stay ranges. Where credible estimates differ by a factor of two, all of them are published with their bases stated.
- Corrections are logged, dated and left visible. Nothing is quietly edited.
- Every weight is published. If you disagree with a score, the method page tells you exactly which number to argue with.
What this site refuses to do
- It does not treat decline as moral failure. A low engine score is a measurement. Europe trading global weight for stability and longevity may be a rational choice rather than a defeat.
- It does not forecast collapse. Every nation in this index will still exist in 2050. Relative decline concerns share, and share is a ratio.
- It does not use civilisational framing. The Europe chapter engages Huntington’s thesis directly and at length, presents the evidence for it, and explains why the site does not adopt it — principally that intra-civilisational conflict massively outnumbers inter-civilisational conflict in every dataset, that civilisations are not actors capable of forming intent, and that a framework which cannot explain 1914 and 1939 is not a framework for explaining Europe. The migration pressure and its political consequences are real and are covered in detail. The frame is what is rejected, not the evidence.
- It does not collapse the engines into a single score. One number would be more shareable and would measure nothing.
Sources
The index is built on official statistics wherever they exist: IMF Article IV consultations and the World Economic Outlook, the Congressional Budget Office, Rosstat, China’s National Bureau of Statistics, Destatis and ifo, Japan’s e-Stat, OECD projections, Frontex, the Uppsala Conflict Data Program, Freedom House, V-Dem, SIPRI, NSIDC and JAXA, the USGS, and national treasuries and defence ministries. Where official data is contested or absent — Russian casualties being the clearest case — independent counts are published alongside their methodologies rather than in place of them.
The underlying research base is a corpus of sourced country and issue files maintained continuously, each carrying the same figure/value/date/source/confidence structure used on these pages.
Corrections log
Corrections will be listed here with the date issued, the original claim, the corrected claim and the evidence that prompted the change. An empty log this early is expected, not reassuring.
Revision schedule
| Version | Scored | Status | Changes |
|---|---|---|---|
| v1.0 | 22 Sep 2026 | Current | Initial publication — 20 nations, 8 flashpoints |
| v1.1 | 31 Dec 2026 | Scheduled | Quarterly re-score; first forecast resolutions |
How to disagree with this site usefully
Argue with an indicator, not with the conclusion. Every score on this site decomposes into published inputs. If you think China’s productive engine should be 85 rather than 71, the robot-density and labour-productivity series are the place to make that case — and if the evidence supports it, the score changes and the change is logged. That is the whole point of publishing the weights.
The single most valuable thing a reader can send is a figure that contradicts one on these pages, with its source attached.
Continue
The method and weights ·
The five engines ·
Back to the index