Why this exists

Organisational theory has a recurring weakness: it cannot be run. Claims arrive in prose, illustrated by anecdote and defended by seniority. Debates about structure become narrative contests in which the most senior voice wins and a reorg gets tested on real people because there is nowhere else to test it first.

Fulcrum is the Decision Architecture model made executable. It models an organisation the way the books describe one (teams, domains, dependencies with propagation delays, authority placement, incentive skew), scores its structural health deterministically from 0 to 100 and valuates every legal move from blunder to great. There is no randomness in scoring: two people with the same model get the same number, which converts an argument about structure into an examinable model. That determinism buys reproducibility, not validity: whether the model itself describes real organisations is a separate question, taken head-on on the page below.

The books' worked numbers reproduce exactly at the tagged release they cite, which is a consistency check: it shows the implementation faithfully executes the model the books set out, not that the model matches reality (that question belongs to What grounds the score). The current model goes one step further and prices concentrated authority by scale, the prince band: Machiavelli's Prince governs well up to the Dunbar horizon of about 150 people and the durable large organisation is the republic of his Discourses, so the same concentrated structure is forgiven at 30 people and priced progressively harder at 30,000. Escalation is priced where it actually resolves (a team whose lead sits across the desk is never priced at conglomerate scale) and the resolving centre carries the escalated load on its own queue, so a founder absorbing thirty teams' decisions saturates and prices itself. The republic is priced too: a dependency between two sovereign teams with no shared unit above them has nowhere to take a conflict, so distributed authority without institutions reads as fragmentation at scale, while a local influence gap costs a proportionate slice of the score rather than half of it. Demand travels along dependencies as well as escalation lines: an empowered hub that thirty teams wait on saturates exactly as a deciding centre does, so dependency concentration prices itself the way authority concentration always did. Scale exposes structure: typical organisation archetypes collapse from 83.4 to 12.0 as they grow while well-designed ones degrade gently from 98.7 to 64.9, untouched by the band because they hold no concentration to price. Moves stay asymmetric: on the enterprise archetype the one available approval-layer move scores -2.59 while the best good move is worth +3.15, so a single blunder erases most of what the best repair earns. And structural value lives at the leaves: in a generated organisation of 1,732 teams at the books' tagged release the best whole-organisation move is worth +0.024 while the median best local move is worth +9.7 in its own frame, which is why the app lets you drill into a leaf and see its move space.

It runs entirely on your machine: generate a level, model your own organisation or import one as JSON, then ask the guide for a move-by-move line to a stronger score. No account, no server, no telemetry. The test that would actually falsify the model is external: score real organisations blind to their known outcomes and check whether the ranking holds, a protocol set out under What grounds the score.

Full reasoning at crankthecode.com: Decision Architecture (how organisations fail and recover) and Decision Architecture Patterns (the Gang of Four design patterns applied to organisations) laid the foundations without which the later volumes would not have been viable; The Move Space builds on them to give Fulcrum its move classifications (blunder to great) and Relativistic Decision Architecture completes the theory.