The prosecutor’s knowledge problem begins with a basic institutional fact that the legal system regularly understates and often ignores: criminal evidence is not a single body of information gathered into a single file by a single state actor. It is a dispersed and continuously evolving set of records, impressions, physical objects, digital artifacts, credibility assessments, administrative findings, and witness-related materials that originate in different places, are processed under different bureaucratic rules, and are often stored in systems that do not meaningfully communicate with one another. The volume’s own structure identifies this chapter immediately after “The Myth of the Omniscient Prosecutor,” and that sequencing is analytically correct. Once the myth of prosecutorial omniscience is rejected, the distributed nature of criminal evidence becomes unavoidable. The prosecutor’s constitutional burden exists in a system where the relevant knowledge is institutionally scattered from the outset.
This is not a minor complication in the administration of criminal cases. It is one of the central reasons the Brady framework fails so frequently in practice while retaining formal doctrinal legitimacy. Brady v. Maryland, Giglio v. United States, and their successors rest on the premise that the prosecution can disclose favorable evidence because the prosecution can know, retrieve, evaluate, and transmit that evidence. The Supreme Court’s later articulation that the prosecutor has a duty to learn of favorable evidence known to others acting on the government’s behalf is an attempt to prevent the state from profiting from internal fragmentation, but it does not eliminate the fragmentation itself. As the constitutional materials summarized by Cornell’s Legal Information Institute explain, Brady suppression can occur even when the favorable evidence is known only to police investigators and not to the prosecutor, and the duty therefore extends beyond personal awareness. That rule is normatively sound, but it also exposes the depth of the structural problem: the law assigns unitary responsibility in a setting where evidence is institutionally plural.
Criminal evidence is distributed in at least three senses, each of which carries constitutional significance. First, it is distributed by origin. The facts relevant to a criminal case are generated by many actors who do not occupy the same office, answer to the same supervisor, or operate under the same disclosure culture. Patrol officers produce initial observations, body-worn camera footage, dispatch interactions, and scene notes. Detectives generate interviews, leads, informant communications, search warrant materials, and interpretive narratives that shape the case long before the file is presented for charging. Crime laboratories create analytical reports, bench notes, instrument data, quality-control documentation, and, at times, records of error, contamination, or corrective action. Jail and detention personnel create incident reports, classification records, surveillance data, medical encounter records, and statements by defendants or witnesses while in custody. Probation and parole officers generate supervision records, compliance documentation, violation reports, and observations that may bear on witness credibility or factual context. Administrative oversight bodies may create investigatory findings or disciplinary records relevant to an officer or government witness. Each of these sources may contain information favorable to the defense, material for impeachment, or context that alters the meaning of other evidence. Yet none of them begins life as “the prosecutor’s file.”
Second, criminal evidence is distributed by format. The legal imagination tends to picture evidence as documents, photographs, and testimony. The modern evidentiary field is vastly more complex. It includes physical evidence collected at scenes; recorded interviews; digital extractions from phones, computers, and cloud accounts; automated license plate reader data; body-worn camera archives; dispatch audio; laboratory data packages; metadata; chain-of-custody logs; internal disciplinary records; cooperation agreements; jail calls; text messages; geolocation data; audit trails in evidence management systems; and derivative memoranda created by supervisors, analysts, or prosecutors themselves. The National Institute of Justice and NIST both emphasize that evidence management is not a singular event but an ongoing process in which evidence is tracked, handled, preserved, and documented across stages and by multiple custodians. That description is valuable because it highlights the true administrative reality: evidence is not simply “possessed” by the state; it travels through the state. Each transfer creates another point at which knowledge is added, lost, filtered, delayed, or recoded.
Third, criminal evidence is distributed by legal status. Not everything that matters to fairness appears in the conventional evidentiary file. Some information exists as impeachment material rather than substantive proof. Some exists as administrative knowledge rather than trial exhibits. Some is held in personnel or disciplinary systems nominally external to the criminal case. Some is treated as confidential under local policy even though it bears directly on a witness’s credibility. Some exists only as prior inconsistent statements, informal benefits given to cooperators, or internal awareness that a particular employee has a history of dishonesty, coercion, bias, or procedural abuse. The prosecutor’s constitutional duty is not limited to what will be offered as an exhibit. It extends to favorable and impeachment information. The Department of Justice’s own Justice Manual expressly situates federal disclosure obligations within Brady, Giglio, and related authorities, confirming that the duty encompasses both exculpatory and impeachment information. That institutional acknowledgment matters because it shows that the government itself formally recognizes a broader disclosure universe than the narrow trial file. The difficulty is that the bureaucracy still often organizes information as though only trial-ready material truly matters.
The distributed nature of criminal evidence therefore should not be understood as a regrettable side effect of complexity. It is the defining condition of modern prosecution. A criminal case is not built from a single stream of fact production. It is assembled from overlapping streams of state activity that emerge from agencies with different missions, different technological systems, different confidentiality norms, and different incentives. The patrol function is not the laboratory function. The laboratory function is not the jail function. The jail function is not the internal affairs function. The prosecutor inherits the output of these systems without controlling the conditions under which the relevant information was generated, classified, retained, or hidden. The state nonetheless demands that the prosecutor perform as if those systems have already been integrated into a coherent constitutional knowledge structure.
That assumption fails at the most basic level of record architecture. Evidence is rarely maintained in a single universally accessible repository. Police agencies often maintain incident and arrest records in one system, digital evidence in another, body-camera archives in a third, internal affairs materials in yet another, and personnel or disciplinary findings under separate access rules altogether. Laboratories may maintain case files, bench notes, corrective action logs, and analyst competency records in distinct quality-assurance systems. Jail records may sit in custody platforms that prosecutors access only partially, if at all. Oversight findings may exist in inspector general reports, civilian commission materials, or administrative complaint systems never linked to prosecutors’ case management software. The result is not just dispersed storage. It is fragmented knowability. Information may technically exist within government hands while remaining functionally unavailable to the lawyer whom the Constitution treats as responsible for its disclosure.
This is where the distribution of evidence becomes a constitutional problem rather than a mere administrative nuisance. The Brady framework does not ask only whether the state physically possessed the material. It asks whether the suppression of favorable information deprived the defense of a fair process. If the state’s own organizational structure makes relevant information practically unreachable, then the unfairness is not cured by the fact that the information existed somewhere in government custody. A due process regime that tolerates institutional fragmentation without requiring institutional integration transforms constitutional obligation into rhetorical aspiration. The defense is told that the prosecution must disclose; the prosecutor is told to learn what others know; the agencies holding the information often continue operating under systems that do not reliably transmit it; and the court then evaluates the resulting failure as though it were primarily a question of individual diligence. That sequence is one of the recurring engines of systemic noncompliance.
The distribution problem is intensified by time. Criminal evidence does not arrive all at once, and it rarely remains static. New witness statements emerge. Digital extractions are completed weeks or months after arrest. Laboratory testing generates successive rounds of documentation. Internal investigations begin after force incidents or allegations of officer misconduct. Informants receive benefits or inducements over time. Jail calls and custody records accumulate while a case is pending. Supervisory reviews may generate doubts about an investigator, analyst, or custodial witness after the initial charging decision has already been made. The evidentiary field is therefore not merely dispersed but dynamic. A prosecutor who receives a case file at charging has not acquired a complete snapshot of reality. The prosecutor has acquired a provisional and often selective slice of an evidentiary process that continues to unfold across institutions. The duty to disclose persists, but the systems needed to keep the prosecution file constitutionally current are often partial, inconsistent, or absent.
This temporal distribution is especially consequential in plea bargaining, where most cases are resolved before the full evidentiary picture matures. If evidence favorable to the defense is still emerging through digital review, forensic analysis, jail investigation, witness recantation, or administrative scrutiny of government actors, then a defendant may be induced to plead under conditions shaped by incomplete state knowledge. The constitutional injury is not limited to the dramatic trial reversal after hidden evidence is discovered years later. It begins when the state uses its coercive leverage while still epistemically disorganized. In that sense, the distributed nature of criminal evidence is not merely a disclosure problem at trial. It is an upstream bargaining problem embedded in the routine mechanics of case resolution.
The state often responds to this difficulty by invoking the concept of the “prosecution team,” but that phrase obscures more than it clarifies. It suggests a stable and coordinated body of actors who share information toward a common constitutional purpose. In reality, the entities whose knowledge may matter to fairness often do not behave like members of a single team in any operational sense. Police agencies may cooperate with prosecutors in building the accusatory case while resisting the disclosure of officer misconduct. Laboratories may provide test results but not easily accessible quality-control concerns. Jail and probation personnel may generate witness-relevant records without any systematic process ensuring those records are screened for Brady significance. Administrative divisions may know damaging facts about state witnesses while treating those facts as internal employment matters rather than case-related constitutional material. The “team” concept solves a doctrinal attribution problem by linguistic fiat, but it does not solve the underlying problem of institutional fragmentation.
Nor does mere access permission solve it. A prosecutor may theoretically be permitted to request information from another agency, but a request-driven model is inadequate when the prosecutor does not know what to ask for, which repository to search, what naming conventions the other agency uses, or whether the other agency will classify the material as discoverable at all. This is one of the most important points in understanding distributed evidence. The problem is not only that information is spread across agencies. It is that the prosecutor’s knowledge of the evidentiary universe is mediated by the very institutions whose records may create constitutional and civil liability. When disclosure depends on another agency’s self-identification of its own embarrassing or credibility-damaging material, the state is effectively trusting defensive bureaucracies to operationalize transparency against their own interests. That is not a serious compliance design.
The distribution of criminal evidence also undermines simplistic distinctions between substantive evidence and impeachment evidence. In practice, the two are often interdependent. A forensic report may look inculpatory until bench notes, calibration issues, analyst error history, or contamination concerns are disclosed. An officer’s narrative may appear straightforward until prior dishonesty findings, coercive interview practices, or repeated inconsistencies emerge from administrative records. A cooperator’s statement may seem powerful until benefits, inducements, dismissed charges, immigration assistance, or protective accommodations are surfaced. Evidence is not just a pile of items; it is a web of meaning. Distributed information affects the interpretation of other distributed information. When one portion remains hidden in another institutional silo, the apparent significance of the evidence that is disclosed may itself be distorted.
For that reason, the state’s recurring effort to compartmentalize “case evidence” from “personnel issues” is often analytically false. A witness’s disciplinary history may directly affect the reliability of testimony. A laboratory analyst’s error record may directly affect the interpretation of a report. A jail officer’s prior misconduct may affect the credibility of alleged admissions or alleged behavioral incidents in custody. A detective’s pattern of affidavit problems may affect the integrity of searches, confessions, and witness handling. The bureaucratic category under which a record is stored does not determine its constitutional relevance. Yet bureaucracies routinely behave as though storage classification controls legal significance. This is one of the central ways distribution becomes concealment without needing to announce itself as concealment. Information does not have to be burned, altered, or expressly hidden. It need only be maintained in a system that treats constitutional relevance as someone else’s problem.
That dynamic is particularly acute when criminal evidence intersects with confidential administrative systems. Internal affairs units, professional standards bureaus, inspector general offices, civilian oversight entities, and human resources divisions may all possess information that bears on witness credibility or institutional practice. But those systems often arise from different legal cultures. They are built to manage discipline, labor relations, public relations, or risk containment, not to facilitate defense disclosure. They may employ different burdens of proof, different retention rules, different taxonomies for misconduct, and different access restrictions. A finding that an officer was “not truthful” in one administrative context may be highly significant for Giglio purposes, yet remain invisible to line prosecutors because it sits within a disciplinary architecture built for management, not constitutional fairness. In that sense, the distributed nature of criminal evidence is inseparable from the distributed nature of government purposes. The same government contains multiple sub-systems, and those sub-systems do not all understand themselves as participants in a common due process obligation.
This helps explain why reforms limited to prosecutorial training so often fail. Training can remind prosecutors to ask for favorable evidence, but it cannot by itself create interoperable databases, mandatory transmission rules, common definitions of impeaching conduct, durable retention practices, or searchable links between witness identities and agency records. The distribution problem is infrastructural. It requires systems that assume evidence relevant to fairness will arise outside the prosecutor’s immediate sightline and must therefore be automatically captured, indexed, and made retrievable. Without such systems, the prosecutor remains dependent on memory, informal relationships, ad hoc requests, and the variable goodwill of agencies whose incentives are not aligned with radical transparency.
The point is not that all distribution is avoidable. Some degree of distribution is inevitable in a large and differentiated justice apparatus. Patrol, forensic science, detention, supervision, and prosecution are distinct functions. The point is instead that the legal system has treated inevitable distribution as though it were already administratively solved. It has confused the normative rule of imputed knowledge with the practical existence of institutional knowledge. Those are not the same thing. Imputation is a doctrine of responsibility. Institutional knowledge is a condition of organizational design. One can assign responsibility to the prosecutor for what others know without having built any reliable mechanism through which that knowledge is gathered and transmitted. That is precisely what the present system so often does.
The consequences extend beyond ordinary Brady analysis into municipal liability and structural accountability. When a county or municipality repeatedly operates criminal cases without integrated systems for tracking officer credibility, laboratory error, custodial witness misconduct, or cross-agency favorable information, the resulting disclosure failures begin to look less like isolated mistakes and more like the foreseeable consequences of policy choice, custom, or deliberate indifference. Distribution alone is not unconstitutional. But unmanaged distribution in the face of known disclosure duties can become a structural constitutional failure. Once the state knows that evidence is institutionally dispersed and that line prosecutors cannot reliably retrieve it, continued reliance on fragmented systems becomes a choice about acceptable constitutional risk.
That is why this chapter is central to the whole volume. The prosecutor’s knowledge problem is often spoken of as if it were an abstraction about information overload. It is more specific and more serious than that. The state has organized the production and custody of criminal evidence through multiple agencies, multiple databases, multiple professional cultures, and multiple secrecy regimes, while continuing to insist that a single officeholder can stand as the constitutional guarantor of disclosure. The distribution of evidence is not a peripheral feature of the system. It is the fact that makes the rest of the volume necessary. The later chapters on imputed knowledge, institutional incentives to withhold information, bureaucratic complexity, illusory compliance, structural constitutional failure, institutional liability, and civil conspiracy all proceed from this condition. If evidence were centralized, transparent, and automatically reportable, many of the later chapters would lose their force. They do not lose their force because the evidence is not centralized, not transparent, and not automatically reportable.
The phrase “distributed nature of criminal evidence” therefore should be read with full seriousness. It describes more than the location of files. It describes the architecture of state knowledge. It names the reality that the facts relevant to guilt, innocence, credibility, coercion, reliability, and fairness are produced by a bureaucracy that does not naturally converge into a single, constitutionally searchable memory. It names the reality that the prosecution inherits evidentiary fragments, filtered through institutions with their own interests and blind spots. It names the reality that due process is asked to function atop a system whose information flows were largely designed for operations, control, and risk management rather than disclosure to the accused.
Once seen clearly, the implication is unavoidable. A fair criminal process cannot be secured merely by commanding prosecutors to disclose favorable evidence while leaving the evidence itself institutionally dispersed, technologically fragmented, and administratively filtered. The legal duty must be matched by a structural response. Systems must be built around the recognition that favorable information may arise anywhere within the state’s evidentiary ecology and that constitutional fairness requires more than passive receipt of what investigators choose to transmit. Until then, the prosecutor’s duty will remain real but operationally compromised, and the state will continue to speak as though its evidence were unified when, in practice, its knowledge is distributed across a bureaucracy that cannot reliably make itself known.