# T001 — Default-On Sharing: Policy Failure or Product Design The Conway PD corpus documents 1,384 active data-sharing relationships in the Flock platform as of 2025-12-17, 94% of them out-of-state, none with a surfaced memorandum of understanding ([[SharedNetworks 2025-12-17 Snapshot]]). The tension concerns *how this topology came to exist*. Two readings of the same Tier-1 evidence support sharply different downstream FOIA and legislative strategies. The mechanism question is load-bearing for [[Default-On Surveillance Data Sharing as Operational Model]] — the entire essay's analytic claim is structural ("the resting state is 'on'"), and which reading is correct determines whether the remedy is configuration discipline (Statement A) or product-design intervention (Statement B). ## Statement A **The 1,384-org topology reflects a *policy failure* — a governance gap in Conway PD's application of its own per-event sharing standard to platform-level relationship configuration. The toggle is policy-neutral; administrators configured relationships that policy did not authorize.** The governing local rule is CPD Policy 800-32, R3 (1/18/2023), approved as "Chief of Police" by William Tapley ([[CPD Policy 800-32 — License Plate Reader Vehicle Operations]]). Section D, paragraph 4 establishes the standard: > "**Data sharing** with other LE agencies 'if evidence of an offense is indicated.'" This is a per-event standard: it asks, of a specific sharing act, whether an offense is in view. The policy text predates Flock's cloud architecture (drafted around physical cameras storing data on department servers — the policy text says "stored on the department's computer servers") but it remains in force. Conway PD has the policy authority to apply the "evidence of an offense" standard to platform-level relationship configuration; nothing in the policy says that test attaches only to per-event sharing decisions and not to the upstream act of configuring a standing relationship that enables thousands of subsequent queries. The relationships in [[SharedNetworks 2025-12-17 Snapshot]] were configured — by humans, in the platform's admin interface — most likely by [[Lt. Andrew Burningham]] in his role as Flock administrator. He appears in 330 of the production's 389 emails ([[Conway Police Department]]). Configuration is a human act. The policy "evidence of an offense" standard could have been applied to each toggle — would there be evidence of an offense in a specific Florida sheriff's office's hot list that justified pushing it into Conway's alert feed? — and was not. The 1,384 relationships exist because Conway's administrators did not apply the policy test that should have governed them. The configuration discipline failed; the platform's settings are not the source of the failure. Cross-state legal compliance evidence supports this reading. Flock itself, in its August 2025 explainer, confirms that the policy-can-govern-product view is *the vendor's own framing*: "**Importantly**, you are responsible for knowing your agency's laws and policies" ([[Audit-System Policy Emails (Aug 2025 - Apr 2026)]]). The vendor disclaims the responsibility for agency-level sharing decisions; the agency carries it. The 1,384 relationships are what the agency chose, not what the product imposed. ## Statement B **The 1,384-org topology reflects *product design* — Flock's interface defaults to sharing, friction favors sharing regardless of policy intent, and policy designed for per-event acts cannot govern a platform whose unit of action is the standing toggle. Configuration replaces warrant as gate by structural property of the system.** The Aug 2025 Flock explainer from SVP Chris Colwell to Burningham lays out the sharing model in the vendor's own words ([[Audit-System Policy Emails (Aug 2025 - Apr 2026)]]): > "Configurable sharing options: Share with agencies in specific states only · Share with agencies with similar laws (for example, regarding immigration enforcement and data) · Share within your state only or within a certain distance · Share 1:1 (only with specific agencies) · Don't share at all." "Don't share at all" is one option among five. The menu's default — the option a configuration takes when an administrator does not actively narrow it — is *some* form of sharing. Not-sharing requires a specific configuration act; sharing does not. The interface presents sharing as the operational baseline. Cross-state restrictions document Flock's *product-side* capacity to gate sharing structurally: "In some states, sharing is automatically restricted as required by law, and searches with search terms that indicate a purpose prohibited by law have been disabled in our product. For example, in Virginia, out-of-state sharing is disabled and in Illinois, accessing data for certain purposes is not allowed" ([[Audit-System Policy Emails (Aug 2025 - Apr 2026)]]). This is significant: Flock can and does build structural restrictions into the product when it judges them warranted (legal mandates in Virginia and Illinois). Where Flock has not built such restrictions, sharing remains the default — and the agency's per-event policy text cannot reach back into the platform to constitute that structural restriction. Arkansas has no in-product restriction analogous to Virginia's or Illinois's. The Arkansas-deployed product defaults to wide cross-state sharing because the product is designed that way. The 1,384-org topology's structure confirms the design reading. 471 are inbound-only and 427 are outbound-only — asymmetric configurations that no rational application of a per-event "evidence of an offense" test would produce on a relationship-by-relationship basis. The asymmetry is "unexplained in the corpus — it reflects either deliberate Conway-side configuration or default-on settings the configuring administrator did not actively curate" ([[SharedNetworks 2025-12-17 Snapshot]]). Per-event policy analysis cannot be reconstructed for relationships an administrator never actively gated. The 1,384 organizations include `[Federal] Wright Patterson OH Air Force Base` with no recorded activity in the corpus's available window ([[SharedNetworks 2025-12-17 Snapshot]]) — a relationship enabled but apparently unused, exactly what a default-on system produces and exactly what a per-event policy test would never authorize. Critically, the platform makes sharing *continuous* once configured. CPD 800-32's per-event standard ("evidence of an offense is indicated") presupposes a per-event sharing decision. Flock's model collapses that decision into a one-time toggle whose downstream consequence is thousands of queries the configuring officer never sees. A per-event policy cannot govern a per-relationship product. The 1,384 relationships exist because the product makes sharing the path of least resistance, and Conway's policy — designed for a per-event architecture the product has obsoleted — has no operative handle on the configuration moment. ## Why it matters The two readings drive different follow-on FOIA strategies and different legislative-briefing framings: - **If Statement A is correct (policy failure):** the remedy is documentation. Follow-on FOIA targets Conway's network-admin training records, internal sharing-configuration approvals, any documents showing how the agency reconciles CPD 800-32 with platform-level configuration. The legislative ask is for a policy-update mandate — agencies must apply per-event evidentiary standards to platform-level configuration, not only to per-event acts. The vendor is policy-neutral; the agency is the regulated party. - **If Statement B is correct (product design):** the remedy is structural. Follow-on FOIA targets Flock's UI default-state documentation, the configuration screens themselves (whether the agency was opted into national sharing on day-1 by default), any vendor onboarding materials that recommend the default-on posture. The legislative ask is for product-design regulation — sharing relationships must require positive per-relationship per-purpose justification at the product level, mirroring Virginia's structural restriction. The vendor's design is the regulated object. Whether the [[Default-On Surveillance Data Sharing as Operational Model]] synthesis's analytic claim ("the resting state is 'on'") attributes the resting state to configuration discipline or to product design determines whether the policy reform path runs through Conway's chief or through Flock's product team. ## Resolution status **`resolved-via-D001`** in favor of **Statement B** on the central mechanism question. The three-phase dialectic [[D001 Synthesis]] (with antecedent phases [[D001 Thesis]] and [[D001 Antithesis]]) settled the mechanism question on the strength of facts both phases conceded: the per-event grammar of CPD 800-32 §D.4, the absence of per-relationship evidentiary documentation anywhere in the corpus, the asymmetric 471 inbound-only / 427 outbound-only configuration pattern, the federal Wright-Patterson relationship with no recorded activity in the available window, Flock's demonstrated capacity to gate sharing at the platform level when law mandates it (Virginia, Illinois) and its choice not to where law does not (Arkansas), and the menu's architectural baseline of sharing as the default mode. Statement B's account — the topology as a structural property of a default-on platform with independently configuring tenants — explains every one of those uncontested facts. Statement A's account requires reading §D.4 against its own grammar in a way the thesis itself conceded was doctrinally vulnerable, and assumes a per-relationship Conway-side affirmative configuration the corpus contains no evidence of. **Statement A survives as a sharpened sub-question.** The dialectic did not extinguish the agency-culpability claim — it narrowed it. Conway PD's failure to *apply §D.4* to platform-level configuration fails on the structural objection that §D.4's per-event evidentiary trigger is logically prior to the events the standard regulates: the standard asks, of a specific sharing act, whether evidence of an offense is in view, which presupposes a triggering event the configuration moment has not yet produced. But Conway PD's failure to *develop a successor internal directive adequate to platform-era configuration* is a real agency omission, separable from §D.4's reach. The chief has, and has long had, unilateral authority to issue such a directive. The absence of one is an act, not an inaction. That narrower culpability claim is filed as [[T002 - Successor-Policy Omission]] for its own dialectical treatment. **What remains empirically bracketed** (the verdict on mechanism is not bracketed; these are sub-questions the synthesis sharpened): Conway PD administrator session logs in the Flock tenant — would establish what fraction of the 1,384 relationships involved an affirmative Conway-side toggle versus a counterparty-initiated configuration. Flock's internal product documentation of default-state sharing posture by jurisdiction and contract tier — would settle whether Arkansas-deployed tenants ship with a default-on baseline subsequently overridden, or a default-off baseline subsequently relaxed. Internal Conway PD correspondence between [[Lt. Andrew Burningham]] and the chain of command on platform configuration practice — material primarily to T002, but bears on the authorship-distribution question here as well. Supplemental FOIA could move each. **What remains definitionally bracketed:** whether the platform-configuration moment is itself "an act of sharing" within §D.4's intended scope is a question of statutory/policy construction the corpus cannot settle. The thesis read it broadly; the antithesis read it narrowly. The text supports the narrow reading on its grammar and the broad reading on a purposive interpretation. That question would be settled by a court or by an authoritative agency interpretation, not by additional FOIA. ## Discovery Surfaced 2026-05-23 during the Dialectical Wiki Pipeline application survey, as the highest-leverage mechanism question in the wiki's mature synthesis layer. The parent synthesis [[Default-On Surveillance Data Sharing as Operational Model]] explicitly raises the tension in its `## Open questions` section: "Who configured each of the 1,384 relationships, and when — a single administrator over time, or a default state the platform supplied? The corpus names the likely configuring administrator but does not document the configuration events." T001 makes that latent contested question a first-class object the wiki can reason about. ## Notes - This tension is the pilot tension for the Dialectical Wiki Pipeline's introduction to D:\arkansas-surveillance (2026-05-23). It will be pushed through dialectic D001 — Default-On Sharing Mechanism (see `wiki/dialectics/D001 - Default-On Sharing Mechanism/`). - The two statements share the same Tier-1 evidentiary base (CPD 800-32 policy text, Flock Aug 2025 explainer, SharedNetworks 2025-12-17 export) — they differ in *what mechanism the evidence implies*, not in *what the evidence says*. This is a framing tension on a shared corpus, not an evidentiary tension on contested facts. - Neither reading is alleged to be the basis for a finding of unlawful conduct. Inter-agency law-enforcement data sharing through commercial platforms is at present largely unregulated rather than prohibited. The tension is structural — about how the resting state of the system came to be — not legal.