Grid/84 is a playback engine for the history and doctrine of strategic nuclear weapons, and it publishes what it stands on. These are the research dossiers behind the models: the sources read, the figures found, the disputes left open, and the numbers in wide circulation that do not survive checking.
A model of a nuclear exchange is a chain of numbers, and a chain is worth the weakest link in it. Every figure the engine uses carries a source and an evidence tier, but a tier in a tooltip cannot show how the figure was arrived at, which papers disagree about it, or what had to be assumed when nothing was published at all. These documents do that work at length.
They follow one convention throughout: a quoted figure is read from the source, not from a summary of it; a composite is labelled as a composite; and where a widely repeated number cannot be traced to a paper, that is stated rather than quietly dropped.
One of them is a different kind of thing and is marked as such. An explainer is not about the literature; it is about the subject, written for a reader coming to it for the first time, and it is the one to start with.
How a nuclear war kills most of the people it kills. The fires and why they are not ordinary fires; the two layers of atmosphere and why it matters enormously which one the smoke ends up in; the dark, the cold, the failed harvest and the famine; and the years in the order they arrive. Written for a reader with no background in it, and honest about the one link in the chain that is genuinely disputed.
The companion to the years afterEvery figure either run or quoted
What is established about the climatic effects of nuclear war, what rests on a single modelling group, and where the real disagreement lies — which is not where it is usually said to be. Covers the source term, the Reisner dispute, the climate response by injection mass, the Xia famine figures and their four dropped conditions, and ozone.
Toon, Robock, Coupe, Xia, Bardeen, Reisner, Wagman, Tarshish · 1982–2025Underpins the years after
The published record on ballistic and hypersonic missile defence, figure by figure. Which interceptor speeds are official — almost none of them are — how the test records are counted and by whom, what the Gulf War Patriot figures actually were and how they were revised, what the Ukrainian dataset shows, and where the phrase comes from. Eleven systems across four phases of flight.
GAO, CRS, the National Academies, Science & Global Security · 1984–2026Underpins the interception and defence labs
Every prompt-effect formula in the engine traced to its section of Glasstone and Dolan's The Effects of Nuclear Weapons, with the computed values set beside the handbook's own worked examples. Blast, thermal, initial radiation, the fireball, the crater and the fallout pattern — including the places where the book declines to give a number, and two results that fall out of putting its sections beside each other.
Glasstone & Dolan 1977, with the 1962 edition for one tableUnderpins the atlas and every study
From 1964 China moved its war industry into the mountains of the interior, explicitly because of the atomic bomb, at two fifths of national construction investment for sixteen years. It is the largest thing any state has done in response to nuclear targeting — and it is the reason a target list drawn from population centres misreads the country, by between seventy and two hundred and fifty kilometres.
Naughton, Meyskens, Lewis & Xue, Zhang · 1964–1980Where this engine’s own method fails
The evidence for and against a Soviet nuclear strike on China in 1969: what the declassified American record shows, what the Soviet side has and has not released, and which of the crisis's most-repeated claims rest on a single unsourced sentence. With the Chinese nuclear complex as it then stood, and the forces on both sides.
FRUS, CIA, Wilson Center, Soviet memoir · 1969–1970Underpins a study in preparation