A technology tree is the natural companion to the topic index, and for five releases this page was a promise with a number attached. The condition it waited on is described below, along with why it was retired rather than met. What follows is the graph as it actually stands — including the parts of it that are thin, the one contradiction in it, and the 148 briefs it cannot place.

Where the graph stands

  • 312 typed depends on edges between briefs
  • 237 of 385 briefs are in the graph — 61.6% — either declaring a prerequisite or being one
  • 157 briefs declare at least one prerequisite (40.8% of the 385 adjudicated — the figure the retired rule measured)
  • 6 layers, longest chain 5 edges
  • 2 briefs asked and recorded none; 0 never asked the question
  • 2241 adjacency links between briefs, for contrast — none of that density is evidence of a prerequisite

The rule this page waited on, and why it was retired

The ship rule was written down on 26 August 2026, before the adjudication that moved its numbers, because a threshold decided after you know the number it governs is a description of whatever happened. It had three clauses: adjudicated coverage at or above 60%, never-asked at or below 5%, and a longest path of at least 4 edges. Clauses 2 and 3 are met — every brief in the corpus has now been asked, and the longest chain runs 5 edges. Clause 1 stands at 40.8% after the largest honest adjudication this project has run, and the threshold was never moved to meet it.

What the adjudication found is a fact about the clause rather than about the corpus. Clause 1 counts a brief only when that brief declares a prerequisite. A brief that is itself a prerequisite — the foundational subject that other briefs are built on — counts for nothing, so the measure systematically penalises exactly the briefs a tree is drawn from. Counted the way a tree is actually read, 237 of 385 briefs sit in the graph, not 157. Separately, 146 briefs were found to disclaim any brief-to-brief prerequisite in their own prose, because their binding constraints are factories, order books, regulators and prices — real dependencies, typed on a separate layer, and correctly not counted here.

The decision, dated. On 15 September 2026 the Institute retired the rule as a gate on publication rather than moving its threshold. The distinction matters and is the reason this paragraph exists: moving 60 to 45 would have been the rule rewritten to fit the number, which is the failure the rule was invented to prevent. Retiring it says something different — that the quantity is not the right precondition for publishing a graph, that the graph is more useful published than withheld, and that the measurement continues. The clause 1 figure is computed on every build and printed above, where it will stay whether it flatters the project or not. The old rule and its miss remain on the Assessment Changes record.

The spine

The longest chain of declared prerequisites in the corpus, foundations first. 6 briefs, 5 edges. A tree with a short spine is a fan; this one is not deep, and saying so is more useful than drawing it larger.

  1. FR-I-24 Space-Based Manufacturing
  2. FR-I-25 Deep Space Infrastructure
  3. FR-II-01 Lunar Industry
  4. FR-II-21 Space Resource Economies
  5. FR-II-23 Moon-Based Manufacturing
  6. FR-II-25 Multi-Planetary Civilization

One edge could not be placed. Neuroplasticity Engineering (FR-III-24) is recorded as depending on Human-Machine Symbiosis (FR-III-23), which is recorded as depending on it. A mutual dependency is an editorial question rather than an extraction bug, and no layering can resolve one. The edge is dropped for the purpose of assigning layers and kept everywhere else, including in the edge count above, so this page does not quietly improve its own graph by deleting what it cannot draw. This pair was examined on 15 September 2026 and both edges were deliberately kept. Both edges were read against the briefs' own dependency prose and both were found to be real claims of different kinds: Human-Machine Symbiosis waits on a RESULT that Neuroplasticity Engineering owns (whether joint human-decoder learning converges is a plasticity question with no interface-side answer), while Neuroplasticity Engineering waits on the APPARATUS that Human-Machine Symbiosis supplies. Two fields that genuinely co-evolve do not have a precedence order, and inventing one to make the graph draw would be a worse error than publishing a graph with a hole in it.

The layers

A brief sits one layer above the deepest prerequisite it declares. Layer 0 is not a claim that a subject has no prerequisites in the world — only that it declares none on this map, which is a statement about the corpus's boundaries as much as about the science.

Layer 0 80 briefs · no declared prerequisite on this map

Layer 1 76 briefs · built on layer 0 or below

Layer 2 44 briefs · built on layer 1 or below

Layer 3 21 briefs · built on layer 2 or below

Layer 4 11 briefs · built on layer 3 or below

Layer 5 5 briefs · built on layer 4 or below

The 148 briefs this tree does not place

Counted apart, because the reasons are different findings. 146 declare no brief-to-brief prerequisite but do carry typed non-brief constraints — a plant, a licence, a price, a supplier, or a result nobody is producing. Those are dependencies, they are frequently the binding ones, and they are published on the Constraint Atlas as their own layer. 2 carry neither, all of which were asked and recorded nothing.

These figures are never added to the ones above, and there is no combined percentage anywhere on this site. Brief-to-brief coverage measures this map. Non-brief coverage measures the world. A brief can carry both, so their sum overcounts, and a blended number would read as progress while referring to nothing. The build asserts the separation and fails if the two are ever presented as one.

What this tree cannot see

  • Adjacency is not dependency. The 2241 adjacency links in the corpus are the densest thing in it, and none of that density was promoted to an edge. A tree built by counting hyperlinks would imply a structure of prerequisites the briefs do not claim.
  • The map cannot see dependencies outside itself. A brief may depend on a subject this project has never scoped, and there is no edge kind for "something that is not on our list". Its absence from this page is not evidence of its absence from the world.
  • An edge is this project's assertion, not a citation. Each one was promoted by reading a brief's own dependency prose, one at a time, and recording the weaker claim where the prose was ambiguous. No published dependency graph exists for most of these subjects to check against. Read an edge as "this brief, having looked, says it is built on that one" — weaker than a measurement, stronger than a guess.
  • Depth is not difficulty. A layer 5 subject is not five times harder than a layer 1 subject; it is five declared prerequisites deep in a corpus of 385 briefs. The tree says what rests on what, and nothing about how far away any of it is.

See also the Topic Index (the map this tree is drawn over), the Constraint Atlas (the non-brief dependency layer), the Frontier Convergence Map, and the Assessment Changes record.