WIP: When the State Knows but the State Does Not Know August 2026

When the State Knows but the State Does Not Know

Regulatory Knowledge Fragmentation, Interagency Information Boundaries, and the Production of Incomplete Administrative Knowledge

Working Academic Research Article

Date: August 25, 2026

Researcher positionality and disclosure: The author initiated the Kentucky Open Records Act requests analyzed in this article and has publicly reported on the underlying regulatory questions through F'nAround Media. The analysis therefore treats agency responses as evidence of what an agency stated, possessed, withheld, or reported after a search—not as automatic proof of every underlying factual proposition. Allegation, agency statement, documentary fact, official finding, and theoretical inference are kept analytically distinct.

Abstract

Public organizations can possess relevant information without converting that information into organization-wide knowledge. This article develops the construct of regulatory knowledge fragmentation: a condition in which information material to a regulatory question is possessed, reviewed, or analyzed within one administrative unit but is not transmitted to, integrated by, or operationally available to another unit whose official output depends on that information. The distinction matters because administrative accuracy depends not only on information acquisition but also on information distribution, interpretation, and organizational memory. Building on organizational information-processing and learning theory, the article uses a revelatory case from Kentucky's medical-cannabis licensing system. The Finance and Administration Cabinet Office of Inspector General (OIG) reported in May 2026 that the Office of Medical Cannabis (OMC) had created an effective, transparent, and fair licensing program and specifically described Kentucky's vertical-integration safeguards. Later Kentucky Open Records Act responses state that OMC possessed private-party correspondence and interoffice preliminary notes, analyses, and recommendations concerning applications and potential ownership or vertical-integration compliance, while also stating that OMC possessed no records of communications with OIG concerning Armory Kentucky, LLC's ownership, acquisition, transfer, control, or vertical-integration implications, and no records reflecting investigations or referrals on those subjects. A subsequent Finance and Administration Cabinet response stated that, after a diligent search, it possessed no records discussing whether OIG Investigation 2025-OIG-002 was inaccurate or incomplete on those ownership or vertical-integration questions, or whether the report should be corrected. The article does not infer intent, concealment, illegality, or the substantive truth of the underlying ownership question from those records alone. Instead, it uses the documented information boundary to show how a state can generate formally valid yet informationally bounded outputs. The article proposes a four-stage model of regulatory knowledge fragmentation, identifies observable indicators and rival explanations, and develops propositions for testing the construct across fragmented regulatory systems.

Keywords: regulatory knowledge fragmentation; organizational learning; information processing; public administration; interagency coordination; information silos; regulatory governance; inspector general; open records; medical cannabis; Kentucky

1. Introduction

Government is often discussed as though it were a single knower. Courts ask what an agency knew. Legislatures ask whether the state had notice. Journalists ask why government failed to act on information already in its possession. Yet modern administrative states are not unitary cognitive actors. They are collections of offices, cabinets, divisions, counsel, licensing personnel, investigators, databases, contractors, and oversight bodies whose information is distributed across organizational boundaries.

That architecture creates a deceptively simple problem: information can exist inside government without becoming knowledge available to the governmental unit that needs it. The state may therefore simultaneously possess a fact and produce an official output that does not incorporate that fact. The apparent paradox disappears once possession and integration are treated as separate organizational processes.

This article calls that condition regulatory knowledge fragmentation. Regulatory knowledge fragmentation occurs when information material to a regulatory decision, evaluation, investigation, or public representation is possessed, reviewed, or analyzed within one component of a regulatory system but is not transmitted to, integrated by, or operationally available to another component whose work depends on it.

The construct is narrower than ordinary bureaucratic failure and more specific than the familiar metaphor of an information silo. A silo describes separation. Regulatory knowledge fragmentation focuses on the consequential pathway from separated information to an official output: Unit A possesses or evaluates information; Unit B is charged with evaluating the system or a related issue; the information does not cross the organizational boundary; and Unit B's output is therefore bounded by the information it actually received.

Kentucky's new medical-cannabis program provides a useful revelatory case. The state's OIG conducted an independent review of OMC's initial licensing process. The OIG report examined ownership disclosure, parent-company disclosure, management service agreements, and Kentucky's prohibition against applying for multiple license types. It concluded that OMC's process was thorough, fair, and transparent. Subsequent open-records responses, however, document an information boundary: OMC later stated that it possessed no communications with OIG concerning specified post-award ownership, acquisition, transfer, control, and vertical-integration questions involving Armory Kentucky, LLC, while separately acknowledging that it possessed private correspondence and internal preliminary notes, analyses, and recommendations concerning relevant applications and potential compliance questions.

The research question is therefore not whether OIG intentionally issued an inaccurate report, whether OMC intentionally withheld information, or whether Armory Kentucky violated Kentucky law. The present record does not establish those propositions. The theoretically important question is different: what happens when relevant administrative information exists within one governmental unit but is not integrated into another unit's official evaluation?

2. Theoretical Framework

2.1 Organizations as Information-Processing Systems

Galbraith's information-processing view treats organizational design as a response to uncertainty and cognitive limits. As task uncertainty increases, organizations must increase their capacity to process information or redesign structures so that decisions can be made with the information required. The implication for regulation is straightforward: the quality of an administrative output depends not only on whether information exists, but on whether organizational architecture routes it to the decision point where it is needed.

Herbert Simon's bounded-rationality tradition similarly rejects the assumption of omniscient organizational decision making. Administrative actors operate with limited attention, limited information, and structured routines. A regulator can therefore behave rationally relative to the information available within its local decision environment while the larger institution remains informationally incomplete.

2.2 Organizational Learning: Acquisition Is Not Distribution

Huber's model of organizational learning is particularly important because it separates knowledge acquisition, information distribution, information interpretation, and organizational memory. The distinction supplies the conceptual bridge for regulatory knowledge fragmentation. An organization may successfully acquire information yet fail to distribute it to units that need it. Huber specifically identified the problem of how units possessing information and units needing that information find one another quickly and reliably.

Fiol and Lyles likewise distinguish organizational learning from mere organizational change. For public administration, the implication is that receiving a record, conducting a review, or generating an internal analysis does not itself demonstrate that the organization has learned in a manner capable of changing later decisions.

Moynihan and Landuyt's study of public organizations further emphasizes the interaction of structural and cultural conditions in organizational learning. Mission orientation, decision authority, information systems, resources, and learning forums shape whether public organizations convert information into usable organizational knowledge. Rashman, Withers, and Hartley's systematic review similarly describes public-sector learning as dynamic and dependent on source organizations, recipient organizations, relationships, and environmental context.

2.3 Knowledge Transfer and Administrative Boundaries

Knowledge-transfer research demonstrates that moving knowledge across organizational locations is difficult even when transfer would improve performance. Argote and Ingram show that knowledge can be embedded in people, tasks, tools, and their interactions, making transfer uneven across contexts. Public administration adds legal jurisdiction, confidentiality, privilege, cabinet boundaries, investigative independence, and records-management practices to those transfer frictions.

The result is an important distinction between documentary possession and institutional availability. A record can be physically or electronically possessed by one unit without being available to another unit's evaluative process. Regulatory knowledge fragmentation begins at that gap.

3. Defining Regulatory Knowledge Fragmentation

Regulatory knowledge fragmentation is defined here as a condition in which information material to a regulatory question is possessed, reviewed, or analyzed within one administrative component but is not transmitted to, integrated by, or operationally available to another component whose decision, investigation, evaluation, or public representation depends on that information.

The construct contains four analytically distinct stages:

Stage

Condition

Observable evidence

1. Local possession

A unit possesses, receives, reviews, or creates relevant information.

Applications, correspondence, internal notes, analyses, databases, memoranda.

2. Boundary failure

The information is not transmitted across a relevant organizational boundary.

No responsive interunit communications; no referral; no shared record.

3. Bounded output

A second unit issues a decision, report, or representation using the information available to it.

Report scope, findings, decision memorandum, public statement.

4. Persistence

Later evidence does not automatically trigger integration, correction, or reconsideration.

No correction records, no reopening record, no documented supplementation.

The model does not require misconduct. Fragmentation may result from lawful compartmentalization, timing, scope limitations, privilege, data architecture, staffing, jurisdictional design, or ordinary administrative error. Intentional withholding is one possible mechanism but is not part of the definition.

4. Case Setting: Kentucky Medical Cannabis Licensing

Kentucky established its medical-cannabis program under KRS Chapter 218B. OMC is responsible for implementation and administration. The initial licensing structure limited the number and type of cannabis business licenses and required disclosure of specified ownership interests, parent companies, organizational structure, financial support, and management service agreements.

The OIG report explains that Kentucky's initial application rule prohibited an applicant from applying for multiple license types, a restriction the report described as a vertical-integration prohibition intended to increase opportunity in the industry. The report further states that the definition of applicant included the parent company to prevent a parent from holding one category while a subsidiary held another. Applicants were required to disclose ownership interests of at least ten percent, parent companies, organizational structure, and applicable management service agreements. Changes in ownership required OMC approval.

The same report also draws a distinction between ownership and management. It concluded that one manager may manage multiple separately owned cannabis businesses through management service agreements and that such management alone is not vertical integration where the owners remain separate and distinct. That distinction makes accurate ownership information central to the regulatory analysis.

5. The OIG Review and Its Information Scope

OIG Investigation 2025-OIG-002 was initiated on May 6, 2025 after OMC requested an independent review of its licensing processes and procedures. The final report is dated May 7, 2026. OIG states that it reviewed reports, system data, processes, videos, legislative testimony, interviews, and the applications of companies randomly selected for licenses.

OIG's executive summary concluded that OMC created an effective program that was fully transparent and fair to all licensure applicants. Its conclusions state that the eligibility process was thorough, fair, and fully transparent; that OIG reviewed the full applications of every entity or individual issued a license through the lottery and found them complete with no anomalies or irregularities; and that the lottery process itself showed no anomalies or irregularities.

The report expressly addressed vertical integration and management service agreements. It described the multiple-license-type restriction, parent-company treatment, ownership disclosure requirements, and management arrangements. It also stated that management of multiple separately owned businesses was not itself vertical integration.

These statements matter because they establish the type of information that could be material to the report's reasoning: ownership, parent relationships, management agreements, and whether separate legal entities were in fact separately owned.

6. Open-Records Sequence and the Documented Information Boundary

6.1 April 2026: OMC Acknowledges Responsive Ownership and Control Records

On April 30, 2026, OMC responded to an open-records request concerning a Mayfield-associated cannabis facility and entities identified in the request as Botany Bay, nuEra, Alchemy, or affiliates or successors. The request sought ownership and control records, management and operational agreements, transfer documentation, regulatory correspondence, and compliance-review materials. OMC stated that it possessed responsive records and required additional time because of the volume and need for review and permissible redactions.

6.2 August 2026: OMC Reports No OIG Communications on Armory Ownership or Vertical Integration

A later request focused specifically on records from January 9 through July 31, 2026 concerning communications between OMC and OIG about Armory Kentucky, LLC's ownership, acquisition, transfer, control, and potential vertical-integration implications; investigations or referrals on those subjects; and related litigation tracking.

OMC's August 21, 2026 supplemental response stated that it did not possess records responsive to the interagency-communication portion. It specifically stated that it did not possess records containing communications between OMC and OIG concerning Armory Kentucky, LLC regarding the specified ownership and vertical-integration subjects. OMC also stated that it did not possess records reflecting investigations or referrals concerning Armory Kentucky and those subjects.

6.3 The Same Response Acknowledges Internal Review Material

The same August 21 response is analytically important because the absence of interagency communications did not mean an absence of all relevant OMC information. For requests seeking reviews, findings, or communications concerning potential compliance with ownership or vertical-integration requirements, OMC referred to its May 29 response and stated that it possessed private-party correspondence concerning applications or requests under 915 KAR 1:020. OMC further identified interoffice preliminary notes, analyses, and recommendations regarding those applications. It withheld those records under Kentucky Open Records Act exemptions and attorney-client privilege.

The response therefore documents two conditions at once: internal possession or analysis within OMC and an asserted absence of OMC-to-OIG communications on the specified Armory ownership and vertical-integration questions. That juxtaposition is the core empirical observation supporting the fragmentation construct. It does not establish what the withheld materials say, whether they would have altered OIG's conclusions, or why no responsive communications existed.

6.4 August 24, 2026: No Records of Correction or Reconsideration

The research then moved to the Finance and Administration Cabinet, whose OIG issued Investigation 2025-OIG-002. The request sought records discussing whether the report or its public representations were false, inaccurate, incomplete, misleading, or based on incomplete ownership or vertical-integration information; records concerning correction, amendment, supplementation, withdrawal, or clarification; and records discussing whether the report should remain publicly available without correction after questions arose.

On August 24, 2026, the Finance and Administration Cabinet Office of General Counsel stated that, after a diligent search, the Cabinet possessed no documents responsive to the request. As with all negative open-records responses, this establishes what the agency reported after its search. It does not establish that no person ever discussed the subject orally, that no relevant information existed outside the search scope, or that the report was substantively incorrect.

7. Evidence Matrix

Record

What it establishes

What it does not establish

Theoretical relevance

OIG Report 2025-OIG-002 (May 7, 2026)

OIG reviewed initial licensing and expressly analyzed ownership, parent-company, management, and vertical-integration rules.

That OIG knew every later ownership or transfer fact; that every post-award relationship was within scope.

Defines the evaluative output and information categories material to it.

OMC April 30, 2026 ORR response

OMC reported possessing responsive ownership/control, management, transfer, correspondence, and compliance-review records.

The contents or legal significance of every responsive record.

Shows local documentary possession.

OMC August 21, 2026 ORR response

OMC reported no responsive OMC-OIG communications on specified Armory ownership/vertical-integration subjects and no investigation/referral records; it also identified internal correspondence, notes, analyses, and recommendations.

Intentional withholding; substantive violation; what privileged/withheld records conclude.

Documents an information boundary between internal possession/review and interagency transfer.

FAC OGC August 24, 2026 ORR response

After a diligent search, FAC reported no responsive records about correcting or reconsidering OIG 2025-OIG-002 on specified ownership/vertical-integration grounds.

That correction was legally required; that no oral discussion occurred; that the report is false.

Evidence of persistence/no documented recoupling at the searched record level.

8. Analysis: From Information Possession to Institutional Non-Knowledge

The Kentucky record demonstrates why the phrase 'the government knew' is analytically imprecise. At least three distinct states of knowledge are possible: local knowledge, transmitted knowledge, and integrated knowledge. Local knowledge exists when one unit possesses information. Transmitted knowledge exists when the information crosses the relevant boundary. Integrated knowledge exists when the recipient incorporates it into its decision, investigation, or organizational memory.

The open-records sequence is consistent with local knowledge without documented transmission on the specified questions. OMC identified internal records and analyses while reporting no responsive communications to OIG concerning Armory Kentucky's ownership, acquisition, transfer, control, or vertical-integration implications. OIG's published report, meanwhile, had already made system-level statements about ownership disclosure, separate ownership, management agreements, and vertical integration.

This is not proof that OIG's conclusions were wrong. It is proof of something more limited and theoretically useful: administrative review can be complete relative to the record supplied to the reviewer while remaining incomplete relative to information held elsewhere in the state apparatus.

The distinction can be represented as: Documentary possession does not equal information distribution; information distribution does not equal interpretation; interpretation does not equal organizational memory; and organizational memory does not guarantee later correction. A regulatory system can therefore satisfy each unit's local procedural obligations while failing to construct a shared institutional picture.

9. The Knowledge Integration Gap

The article proposes the knowledge integration gap as a measurable component of regulatory knowledge fragmentation. The gap is the difference between information relevant to an institutional output that is demonstrably possessed somewhere within the regulatory network and information demonstrably available to the unit producing that output.

A simple empirical representation is KIG = R_p - R_i, where R_p is the set of relevant records or information possessed within the defined regulatory network and R_i is the subset shown to have been integrated into the focal decision or review. The expression is conceptual rather than a claim that relevance can always be reduced to a single numeric value. In empirical work, the sets can be coded by topic, date, source, materiality, and transmission status.

A high integration gap does not necessarily imply negligence. It identifies a condition requiring explanation. The causal mechanism may be scope, timing, privilege, jurisdiction, technical incompatibility, staffing, legal independence, or strategic behavior. The value of the construct is that it separates the observable information architecture from contested claims about motive.

10. Rival Explanations and Boundary Conditions

First, timing may explain the apparent gap. Some ownership or transfer information may have arisen after OIG completed substantive fieldwork or after the period the report was designed to examine. The OIG investigation began in May 2025 and focused on the initial licensing process; later events cannot automatically be treated as omitted evidence.

Second, scope may explain non-transmission. OIG expressly focused on applicants, applications, and licensees selected through the initial lottery. Post-award operational or ownership changes may fall outside that scope even if related to concepts discussed in the report.

Third, confidentiality and privilege can restrict transfer or public disclosure. OMC invoked preliminary-record exemptions and attorney-client privilege for some internal materials. A lawful restriction on disclosure is not equivalent to organizational failure.

Fourth, entity naming can create search error. OMC stated that it possessed no responsive records concerning 'nuEra' in one portion of the August response, while the regulatory record may use legal entity names, management-company names, DBAs, parent names, or individuals. Researchers must therefore avoid treating brand-name search results as complete organizational maps.

Fifth, a negative open-records response establishes the agency's representation after its search, not metaphysical nonexistence. Records can be outside retention periods, held by another unit, oral, privileged, misindexed, or outside the wording of the request.

These boundary conditions are essential. Regulatory knowledge fragmentation holds water only when the researcher can demonstrate relevant local possession and a consequential failure of transmission or integration within a properly defined time and scope.

11. Propositions for Future Research

Proposition 1. The probability of regulatory knowledge fragmentation increases with the number of organizational boundaries separating information acquisition from decision authority.

Proposition 2. Fragmentation is more likely when regulatory systems distribute licensing, enforcement, legal review, and independent oversight across separate organizational units.

Proposition 3. The probability that local information becomes institutionally consequential increases when formal referral, shared-database, or mandatory cross-unit notification mechanisms exist.

Proposition 4. Independent oversight can improve legitimacy while still producing bounded knowledge when the reviewing body depends on information selected or transmitted by the entity under review.

Proposition 5. Negative-record responses across multiple agencies can be used as network evidence to map missing information pathways, provided researchers distinguish absence of records from absence of events.

Proposition 6. The longer an official output remains uncorrected after materially inconsistent information becomes available elsewhere in the regulatory network, the more analytically important organizational memory and correction mechanisms become.

Proposition 7. Regulatory systems with explicit ownership, control, affiliate, and change-of-control rules will be especially sensitive to knowledge fragmentation because legal compliance depends on information distributed across entity and agency boundaries.

12. Methodological Contribution: Open Records as Information-Flow Tracing

The case also illustrates a methodological use of public-records law beyond document collection. Sequential requests can be designed to trace information flow. One request establishes possession. A second tests transmission. A third tests investigation or referral. A fourth tests correction or organizational memory. The unit of analysis becomes the pathway between records rather than the isolated record.

This approach should be used cautiously. Requests are researcher interventions and may themselves alter institutional behavior. Search terms can create false negatives. Exemptions can hide content necessary to determine materiality. The method is strongest when requests are date-bounded, entity-specific, and constructed so that positive and negative responses both produce interpretable evidence.

For replication, future studies should code each record by originating unit, receiving unit, date, subject, legal authority, transmission status, privilege status, decision relevance, and whether the information appears in the focal output. Such coding would permit network analysis of administrative knowledge flows across agencies.

13. Implications for Public Administration and Regulatory Design

The principal policy implication is that transparency within units is not enough. A regulator can maintain extensive records and still fail to produce integrated institutional knowledge. Governments should therefore evaluate information architecture as part of regulatory design.

Potential controls include mandatory cross-unit referrals for ownership or control changes material to prior oversight findings; shared entity-resolution systems that connect legal names, DBAs, parents, affiliates, managers, and principals; documented post-report correction protocols; and defined triggers requiring an inspector general or reviewing body to assess whether later information materially affects a published conclusion.

Independent oversight also requires attention to information dependence. An oversight body can be structurally independent yet informationally dependent on the agency it reviews. Independence of judgment does not eliminate dependence on the completeness of the information pipeline.

Finally, the construct changes how accountability questions are framed. Instead of asking only 'Who knew?', investigators can ask: Which unit possessed the information? Which unit needed it? Was there a formal transmission mechanism? Did transmission occur? Was the information integrated? If not, why not? Was the output later reconsidered?

14. Conclusion

Administrative states do not know things in the same way individuals do. Their knowledge is distributed across organizational components. That makes information integration a governance function rather than an administrative detail.

The Kentucky medical-cannabis record provides a bounded but unusually observable illustration. OIG publicly evaluated a licensing system in which ownership, parent-company relationships, management agreements, and vertical-integration restrictions were central. OMC later reported that it possessed internal correspondence and preliminary analyses concerning relevant applications and compliance questions while possessing no responsive communications with OIG concerning specified Armory Kentucky ownership and vertical-integration subjects. Finance later reported no responsive records concerning correction or reconsideration of the OIG report on those grounds.

Those records do not prove corruption, concealment, a licensing violation, or an inaccurate OIG conclusion. They establish a documented information architecture from which a generalizable theoretical question emerges.

Regulatory knowledge fragmentation explains how the state can possess information without the state, as an integrated decision system, knowing it. The contribution is therefore not the accusation that one office failed. It is the proposition that regulatory accuracy is a function of information transmission across organizational boundaries, and that the gap between possession and integration can be observed, measured, compared, and redesigned.

References

Argote, L., & Ingram, P. (2000). Knowledge transfer: A basis for competitive advantage in firms. Organizational Behavior and Human Decision Processes, 82(1), 150-169. https://doi.org/10.1006/obhd.2000.2893

Fiol, C. M., & Lyles, M. A. (1985). Organizational learning. Academy of Management Review, 10(4), 803-813. https://doi.org/10.5465/amr.1985.4279103

Galbraith, J. R. (1974). Organization design: An information processing view. Interfaces, 4(3), 28-36. https://doi.org/10.1287/inte.4.3.28

Huber, G. P. (1991). Organizational learning: The contributing processes and the literatures. Organization Science, 2(1), 88-115. https://doi.org/10.1287/orsc.2.1.88

Moynihan, D. P., & Landuyt, N. (2009). How do public organizations learn? Bridging cultural and structural perspectives. Public Administration Review, 69(6), 1097-1105. https://doi.org/10.1111/j.1540-6210.2009.02067.x

Rashman, L., Withers, E., & Hartley, J. (2009). Organizational learning and knowledge in public service organizations: A systematic review of the literature. International Journal of Management Reviews, 11(4), 463-494. https://doi.org/10.1111/j.1468-2370.2009.00257.x

Kentucky Revised Statutes, Chapter 218B.

915 KAR 1:010. Initial and renewal application for cannabis business licenses.

915 KAR 1:020. Cannabis business licenses.

Primary Record Appendix

Kentucky Finance and Administration Cabinet, Office of Inspector General. (2026, May 7). Report of Investigation, File ID No. 2025-OIG-002, Office of Medical Cannabis.

Kentucky Office of Medical Cannabis. (2026, April 30). Initial response to open records request received April 23, 2026.

Kentucky Office of Medical Cannabis. (2026, August 7). Initial response to open records request received July 31, 2026.

Kentucky Office of Medical Cannabis. (2026, August 21). Supplemental response to open records request received July 31, 2026.

Kentucky Finance and Administration Cabinet, Office of General Counsel. (2026, August 24). Open Records Request Response O-27-107.