diff --git a/wiki/log/2026-07-11-surveillance-policy.md b/wiki/log/2026-07-11-surveillance-policy.md --- a/wiki/log/2026-07-11-surveillance-policy.md +++ b/wiki/log/2026-07-11-surveillance-policy.md @@ -13,10 +13,14 @@ attention for outward power-seeking and recurs at every scale. Repetitive plot routes may be automated through standing authorization, but automated runs still use the concrete authored plot, Thought reservoir, committed - resources, causal acts, signatures, and failure path. -- **Open:** exact LIE jurisdiction (bounded hops or controlled connected - region), whether compromised observer endpoints join that jurisdiction, - interception tuning/treatment, and the B1 plot-policy editor and scope. + resources, causal acts, signatures, and failure path. Later in the same + session Cameron resolved LIE jurisdiction as the controlled connected region, + not bounded hops. Compromised observer endpoints join that region; + uncompromised endpoints do not. Region control is distinct from service + health: local concealment, inter-region transport, and concealed destination + compute can leave a controlled region overloaded, partitioned, or dark. +- **Open:** interception tuning/treatment and the B1 plot-policy editor and + scope. - **Changed:** amended `wiki/vision/simulation-laws.md` with stable load as the first automation payoff; amended `wiki/mechanics/detection.md` with coverage-first LIE command; amended `wiki/mechanics/plots.md` with standing diff --git a/wiki/mechanics/detection.md b/wiki/mechanics/detection.md --- a/wiki/mechanics/detection.md +++ b/wiki/mechanics/detection.md @@ -172,24 +172,36 @@ LIE machine answers the evidence that sits on or passes through what it controls, and evidence that routes around your coverage gets through. This supersedes the radius-vacuum scrub as ontology (the - implemented radius runtime stands until migration). Whether records - already sitting unread at an observer's endpoint can still be - interdicted — reaching into their device — is [OPEN]; interception - surfaces and rates are [TUNE]. + implemented radius runtime stands until migration). A compromised observer + endpoint is a controlled node: its unread records remain eligible for + interdiction until their read-deadlines. An uncompromised endpoint remains + outside the region even when its route is known; records that arrive there + must be caught upstream. Interception surfaces and rates are [TUNE]. - **The player governs coverage, not an evidence inbox (DECIDED - 2026-07-11).** A LIE machine is a standing topological service over nearby - connected territory. It draws eligible unread crimson records through the - routes it controls and spends finite concealment throughput chewing through - them before their read-deadlines. The player establishes and extends that - service by controlling machines, links, switches, and later inter-plane - ports; they do not approve every ordinary deletion or falsification one - record at a time. Records become salient exceptions when they route around - coverage, exceed its capacity, or approach a read before interception. This - is the same command problem in a basement and across a corporation: secure a - topology, watch its load, and expand the controlled graph. The exact opening - jurisdiction — bounded hops versus a controlled connected region — and - whether compromised observer endpoints join it remain [OPEN]; throughput, - pull latency, and visual treatment are [TUNE]. + 2026-07-11; jurisdiction resolved later that day).** A LIE machine is a + standing topological service over its **controlled connected region**, not a + bounded-hop radius. It draws eligible unread crimson records through the + routes the player controls and spends finite concealment throughput chewing + through them before their read-deadlines. Switches, trunks, compromised + endpoints, and later inter-plane ports extend the region; a broken or + uncontrolled route partitions it. The player does not approve every ordinary + deletion or falsification one record at a time. Records become salient + exceptions when they route around coverage, exceed its capacity, or approach + a read before interception. This is the same command problem in a basement, + across a corporation, and across continents: secure a topology, watch its + load, and expand the controlled graph. Throughput, pull latency, and visual + treatment are [TUNE]. +- **Control and service health are separate (DECIDED 2026-07-11).** A region + can remain controlled while its concealment service is overloaded, + partitioned, or dark. Three capacities determine whether LIE can keep up: + local concealment throughput, inter-region route capacity, and concealed + compute at the destination. A hot region may therefore generate evidence + faster than local LIE can absorb it while the trunk to a better-concealed + region is saturated or cut. The strategic answers stay physical: add local + concealment or hidden compute, increase or hijack trunk capacity, shed work, + or relocate the hot operation. Continental status such as “Latin America is + offline” must emerge from this topology and capacity state, not from an + arbitrary machine radius. - **The two ledgers stay two systems (AFFIRMED 2026-07-11).** Exposure does not merge into the player's intel buffer — "you want someone's understanding of what has actually happened to be different" diff --git a/wiki/vision/simulation-laws.md b/wiki/vision/simulation-laws.md --- a/wiki/vision/simulation-laws.md +++ b/wiki/vision/simulation-laws.md @@ -92,8 +92,10 @@ **The three delegations (AMENDED 2026-07-10, sinks-not-modes; was the four fleet modes of 2026-07-09)** are **WORK, THINK, and LIE**. WORK is the assigned day job and sheds nothing; THINK produces **thought**, the -one traveling substance, and sheds crimson; LIE absorbs crimson within -its well. The 2026-07-09 Research/Operations split dissolved because +one traveling substance, and sheds crimson; LIE absorbs crimson across +the controlled connected region it serves, bounded by concealment throughput +and route capacity rather than a local radius. The 2026-07-09 +Research/Operations split dissolved because under thought flow it bought nothing: what thinking does is decided by the **sinks you open**, not by a machine state. Social remains an actuator/channel describing work with people, not a type of compute. diff --git a/wiki/log/decisions/2026-07-11.md b/wiki/log/decisions/2026-07-11.md --- a/wiki/log/decisions/2026-07-11.md +++ b/wiki/log/decisions/2026-07-11.md @@ -203,3 +203,20 @@ reach, and the first plot-policy editor/scope. Specs: `wiki/vision/simulation-laws.md`, `wiki/mechanics/detection.md`, `wiki/mechanics/plots.md`. + +- **2026-07-11 — LIE governs controlled connected regions, and control does + not guarantee service.** Cameron rejected bounded-hop coverage as too small + and too mobile for the intended command scale. A LIE machine serves the + contiguous topology the player controls; controlled switches, trunks, + compromised observer endpoints, and later ports extend that jurisdiction, + while broken or uncontrolled routes partition it. An uncompromised observer + endpoint stays outside the region. Concealment remains finite along three + axes: local LIE throughput, inter-region network capacity, and concealed + compute at the destination. The desired fleet-scale report is “Latin America + is offline”: a region may still be owned while hot operations overwhelm local + concealment and its trunk to a safer region saturates or fails. The player + answers by localizing concealment/compute, increasing or hijacking transport, + shedding work, or relocating operations. Rejected: tower-defense scrubber + radii and binary owned-equals-healthy region status. Runtime migration from + the current pool/radius model remains unimplemented. Specs: + `wiki/vision/simulation-laws.md`, `wiki/mechanics/detection.md`.