What reachability means, what moves the map, and how we keep the model honest.
Reachability is how much would still need to change for an outcome to emerge. An outcome is more reachable when more of the conditions it needs are already present, more turning points are moving in its direction, and fewer major blockers remain.
Reachability is not a probability or a forecast. The six futures are not mutually exclusive: several can become true at once. A highly reachable future is one that needs less to change, not one that is certain to happen.
Core evidence comes from administrative data, product telemetry, transaction and payroll data, job postings, public API or token usage, deployment logs, field experiments, repeated benchmark series, incident registries, audited deployments, corporate filings reporting actual activity, and reproducible empirical studies.
Survey evidence may describe expectations, intentions, sentiment, or perceived barriers. It is context only: it does not set a core indicator state. Lab self-reports are allowed when they directly measure a capability or deployment condition and no independent equivalent exists; they are marked as source-dependent and checked against a search for counterevidence. Five articles about one event count as one piece of evidence.
New evidence is discovered once a day. A person reviews every candidate before it can enter the model, and the model runs only when evidence is admitted. The map changes only when admitted evidence moves it past a frozen materiality threshold and the update passes editorial review. A logic drift audit runs every Monday; if it fails, publication stops until a person investigates.
Model version
2.0-prospective-port (draft)
Map shows
Recorded snapshot of 2026-10-05. Views before 5 Oct 2026 are reconstructed with the 5 Oct frozen model using only evidence available by that date.
Latest weekly audit
FAIL · 2026-10-06
Last model-changing event
2026-10-05
Config fingerprint7f34dee343c8c0e7c5b5550ad7ca665d804cea6857ad14c99b1f6a2e1f4b676f