VOL. I, NO. 1 • WEDNESDAY, FEBRUARY 11, 2026 • PRICE: ONE UNREDACTED MOMENT OF YOUR ATTENTION
THE REVIEW
“Where the evidence leads—assuming anyone can find it”
The Justice System Has a Plumbing Problem, and You’re Paying for It
The American criminal justice system is broken. Not in the dramatic, made-for-cable-news way you might expect—no rogue judges, no shadowy conspiracies, no constitutional crisis requiring its own miniseries. It is broken the way your kitchen sink is broken: slowly, invisibly, and in a manner that only becomes apparent when sewage backs up into the salad.
This edition of The Review examines a single, unglamorous truth that emerged from a 60-day research window spanning December 2025 through February 2026: the infrastructure that is supposed to convert evidence into justice—the labs, the servers, the databases, the filing systems—has been overwhelmed by the very technologies that were supposed to improve it. We are, to put it plainly, drowning in proof.
You will find six stories inside these pages, each examining a different crack in the pipe. A police department that “solved” 101 percent of its murders through creative accounting. Crime labs so backlogged that Oregon stopped testing DNA for property crimes entirely. A Colorado prosecutor whose murder charge was downgraded because the server couldn’t upload the footage fast enough. An AI that writes police reports and then deletes the evidence that it did so. And a citizen in Ohio who learned that watching body-camera footage of a traffic stop would cost $750.
These are not separate problems. They are one problem wearing six disguises. The connecting thread is throughput—the system’s ability to move evidence from the crime scene to the courtroom—and in 2026, that throughput has collapsed. What follows is our attempt to make that collapse readable, perhaps even occasionally entertaining, and above all, honest. Your skepticism, dear reader, is not only welcome; it is the entire point.
The evidence pipeline: every ⚠ marks a bottleneck documented in this edition
❧ ❧ ❧
Your City “Solved” 101% of Its Murders Last Year. Congratulations?
The numbers on your police department’s report card may be designed to make you feel better, not smarter
The Los Angeles Police Department solved 101 percent of its homicides in 2025, which is either a triumph of detective work or a triumph of arithmetic, depending on how closely you read the fine print.
The number is real. LAPD resolved 78 cases from prior years on top of the 230 new homicides that occurred during the year, and because the FBI’s formula divides total clearances by new offenses, the math produces a figure north of 100 percent. It tells you nothing about how many of 2025’s victims saw their cases resolved. It tells you a great deal about how the department wanted to be perceived.
Quote
“Clearance-rate data going forward must specify how many cases were closed through exceptional means. Transparent data is the prerequisite for policies that ensure crimes are actually being solved.” — Artiense Myrick, Deputy Director, Live Free Illinois
Across the country, the homicide clearance rate has fallen from 93 percent in 1962 to roughly 61 percent in 2024, according to FBI data analyzed by the Murder Accountability Project. Criminologists Philip Cook and Wendy Mancik call this trajectory the “Great Decline.” The sharpest drops are concentrated in gun homicides of Black victims—a disparity the headline number is uniquely ill-suited to reveal.
The problem is structural. Under FBI Uniform Crime Reporting rules, a case can be “exceptionally cleared” when the suspect dies, flees the country, or when a prosecutor simply declines charges. That administrative closure counts identically to a case ending in handcuffs. The Chicago Police Department celebrated a 71.2 percent clearance rate for 2025—its highest in years. The Chicago Tribune documented in January 2026 that more than 60 percent of the 296 total clearances originated in prior years, the oldest from 1987. One detective closed a five-year-old cold case through secondhand identifications and statements from deceased witnesses, with no arrest made. The case was “cleared.”
Meanwhile, three states are pushing back. Illinois now requires monthly disaggregated data starting July 2026. Missouri directs its Department of Public Safety to distinguish arrest-based clearances from exceptional ones. Michigan allocated $115 million annually for clearance-rate improvement—but with financial penalties for agencies that miss targets, creating an incentive to game the statistics rather than improve reality.
The deterrence implications are not academic. The R Street Institute found that nonfatal shooting clearance rates in some neighborhoods have fallen below 10 percent. When nine out of 10 shooters face no arrest—and the neighborhood knows it—the system’s ability to prevent future violence has effectively collapsed for the people who need it most.
The Great Decline: U.S. homicide clearance rates, 1962–2024. Source: FBI UCR data, Murder Accountability Project
For Further Reading: Perspectives
🟢 PRO — “Solving Crime, Protecting Communities: A Blueprint for Safer Streets” — R Street Institute (June 2025)
🔴 CON — “What’s Driving the Drop in Homicide?” — Council on Criminal Justice (Jan. 2026)
❧ ❧ ❧
The Crime Lab Will Get to Your Rape Kit in 570 Days. Please Hold.
Federal funding cuts and a supply chain you never knew existed are starving the invisible backbone of criminal investigation
Across America, forensic crime labs are drowning in evidence—and the lifeboat is sinking. Oregon halted DNA analysis for all property crimes in January 2025. Washington State’s firearms backlog hit 955 requests with a 379-day turnaround. Colorado’s sexual assault kit processing time reached 570 days. And then the federal government proposed cutting the main grant program that keeps these labs running by 71 percent.
“If you limit those resources even further, there will be backlogs,” said Scott Hummel, president of the American Society of Crime Laboratory Directors, in what may be the understatement of the decade. “Labs are forced to make difficult decisions on how they prioritize their casework.”
The decisions are not abstract. In Oregon, “prioritize” means DNA testing for burglaries and car thefts simply stops until every sexual assault kit is processed. In Virginia, 4,000 biology cases sat backlogged in Q1 2025. In Tennessee, 50 new forensic scientists in four years could not keep pace with demand.
Quote
“It’s important for policymakers to realize that the criminal justice system is demanding more from us, and so we need the resources to keep up with the increased demand.” — James Carroll, Crime Lab Director, Los Angeles County Sheriff’s Department
The Paul Coverdell Forensic Science Improvement Grants Program faces a proposed cut from 10 million under the FY2026 budget—a 71 percent reduction. The Debbie Smith DNA Backlog Grant Program would receive 151 million authorized cap. These programs fund everything from aging gas chromatographs to CODIS database expansion.
Then there is the supply chain nobody talks about. Clinical and forensic labs share the same global supply chains for reagents, pipette tips, and mass spectrometry consumables. A 15 percent Medicare reimbursement cut to clinical labs—scheduled under PAMA provisions—reduces overall purchasing power, leading to supplier consolidation and price increases that forensic labs, with no clinical revenue, cannot absorb. The opioid crisis makes it worse: every new fentanyl analogue that appears on the street requires weeks of validation and thousands in reference materials, crowding out capacity for everything else.
Connecticut’s forensic lab stands as proof that investment works. Once plagued by suspended accreditation and a 12,000-case backlog, the lab now averages a 20-day turnaround. The difference is not technology. It is sustained institutional commitment—a resource considerably more scarce than Taq polymerase.
Forensic lab turnaround: days from evidence to results. Connecticut proves investment works.
For Further Reading: Perspectives
🟢 PRO — “Forensic Crime Labs Are Buckling as New Technology Increases Demand” — Stateline (July 2025)
🔴 CON — “Justice Delayed Is Justice Denied: Shelby County DA Clears Video Evidence Backlog” — WMC Action News 5 (Jan. 2026)
❧ ❧ ❧
The Murder Charge Was Downgraded Because the Server Was Too Slow
Colorado’s courts are running 2026 cases on 2010 infrastructure, and the constitutional right to see the evidence against you now depends on upload speed
In 2017, a vehicular homicide in Jefferson County, Colorado, produced 79 photographs and zero video. In 2025, the same charge generates 362 photographs and up to 90 hours of body-camera and dashcam footage. The infrastructure absorbing this deluge was designed for a world of paper files and DVD burns.
Tom Raynes, executive director of the Colorado District Attorneys’ Council, characterized the state’s digital evidence system as a “2010 Toyota” trying to function in a 2026 environment. The IT system, built in 2015–2016 for $3 million, was designed before the body-camera mandate. It cannot handle the petabytes generated by 14,000 equipped officers.
The consequences are not theoretical. One Colorado DA’s office received so many sanctions for failing to turn over evidence on time that a judge reduced a defendant’s charge from first-degree to second-degree murder. The evidence existed. The plumbing could not deliver it within the statutory 21-day deadline.
| Evidence Metric | 2017 | 2025 | Change |
|---|---|---|---|
| Photographs per case | 79 | 362 | +358% |
| BWC/Dashcam footage | 0 hrs | Up to 90 hrs | New baseline |
| Annual videos (JeffCo) | ~36,000 (2022) | 67,700+ | +88% |
| Annual video hours | ~24,000 (2022) | 41,000+ | +71% |
District Attorney Alexis King noted that felony attorneys handle about 100 cases each. With four to six hours of body-camera footage per routine case, attorneys are operationally tasked with reviewing 400 hours of video within statutory timelines. This is, as mathematicians say, impossible.
Quote
“Arguably, everybody doesn’t have the same ability to prepare. They might be able to prepare something in two hours that now takes us 40 hours. I’m not ascribing bad motive to anyone in that, but it’s not fair.” — James Karbach, Colorado State Public Defender’s Office
The equity gap cuts deepest on the defense side. Axon’s tiered Evidence.com platform gives prosecutors with premium licenses seamless search and transcription tools. Public defenders receive a link to a massive dump of unorganized files—the haystack—while the prosecution retains the metal detector.
The Colorado eDiscovery Task Force recommended statewide vendor contracts, interoperability requirements, and a massive capital injection. The legislature has yet to fully fund any of it.
One case, two eras: evidence volume per vehicular homicide in Jefferson County, Colorado
For Further Reading: Perspectives
🟢 PRO — “Colorado Courts’ Fragmented Discovery System Needs a Statewide Fix” — Denver Post (Dec. 2025)
🔴 CON — “Massive Volume of Digital Evidence Is Crippling Colorado’s Criminal Justice System” — CBS Colorado (Dec. 2025)
❧ ❧ ❧
The Bullet That Can’t Cross County Lines
Proprietary file formats are making ballistic evidence invisible across jurisdictions, and a single company’s business model is the reason
A bullet casing recovered in one American city can become invisible to investigators in a neighboring jurisdiction—not because the evidence is weak, but because the two departments’ scanners speak different proprietary languages. Welcome to the walled garden of forensic ballistics.
The National Integrated Ballistic Information Network (NIBIN), administered by the ATF, is the country’s only system for searching ballistic images. It has generated over one million investigative leads. It relies entirely on one commercial product—the Integrated Ballistic Identification System (IBIS)—and because no comprehensive federal standards exist for data exchange, jurisdictions using alternative scanners create what NIST researchers describe as “gaps in NIBIN coverage.”
The technical term is “specular friction”: incompatible lighting angles and reflectivity settings between scanners produce images that cannot be reliably compared. When a search fails to match a casing, the investigator cannot know whether there is genuinely no match or whether the match exists on a scanner that speaks a different dialect.
NIST’s feasibility study (Interagency Report 8548) found grounds for optimism. Three-dimensional topography imaging makes cross-platform comparison viable—interoperability tests showed that a 17–21 percent performance drop without a common standard disappeared when the standard was applied. The obstacle is not physics. It is product design and the business model that treats public safety data as a competitive asset.
The same pathology infects body-camera storage. Axon controls an estimated 85 percent of the police body-camera market and dominates digital evidence management through Evidence.com. Denver’s 625 million—extending its control across the entire evidence lifecycle.
Quote
“After 2031, costs could rise even more since Axon controls the market and its technology, giving it leverage to raise prices.” — Paul Joseph, San Jose Police Chief, in a memo to city council
The proposed fix is an “Evidence Portability Standard”—a mandate that vendors allow evidence export to commodity storage, the way number portability forced competition on cell carriers. CJIS compliance is a set of security requirements, advocates argue, not a proprietary technology. The question is whether agencies are permitted to use alternatives.
The walled garden: why ballistic evidence can’t cross county lines. Source: NIST IR 8548, EFF
For Further Reading: Perspectives
🟢 PRO — “Beware the Bundle: Companies Are Banking on Becoming Your Police Department’s Favorite Vendor” — Electronic Frontier Foundation (May 2025)
🔴 CON — “San Jose Approves Expanded Taser, Body Cam Deal with Axon Despite Monopoly Concerns” — San José Spotlight (Oct. 2025)
❧ ❧ ❧
The Robot Wrote the Police Report, Then Destroyed the Evidence
AI-generated police reports are proliferating across American law enforcement—and they’re designed to be unauditable
In Heber City, Utah, a police department testing Axon’s Draft One AI report-writing tool got back a narrative that claimed an officer had transformed into a frog. The software had picked up audio from The Princess and the Frog playing on a television in the background. “That’s when we learned the importance of correcting these AI-generated reports,” Sergeant Rick Keel told Fox 13, in what may be the most generous assessment of an AI hallucination ever committed to the public record.
The frog incident is funny. The underlying problem is not.
Draft One, Axon’s fastest-growing product since its 2024 launch, uses a variant of OpenAI’s GPT-4 to process body-camera audio and create police report drafts. The pitch is seductive: police staffing shortages are real, and officers spend up to 40 percent of their time on paperwork. But the Electronic Frontier Foundation’s two-part investigation found that when an officer finishes editing a Draft One report, the initial AI-generated draft is deleted. No tracked changes, no version history, no audit trail.
Quote
“Police should not be using AI to write police reports. There are just too many questions left unanswered.” — Matthew Guariglia, EFF Senior Policy Analyst
This is not a technical limitation. An Axon product manager acknowledged the system does not retain drafts because doing so would “create more disclosure headaches for our customers.” The company built the tool to delete its own work product specifically to avoid the legal obligation to disclose it.
Legal scholar Andrew Guthrie Ferguson calls the result “generative suspicion”: the AI, trained on thousands of existing reports, learns the genre—it “knows” that a robbery report should contain “intent to deprive permanently” and that a resisting arrest report mentions “tensing of the arms.” Even when audio is ambiguous, the algorithm inserts these legal terms of art to complete the pattern. Ferguson calls this the “Mad Libs effect.”
The courtroom consequences are real. Officers caught contradicting their reports can now blame the AI. Cross-examining an officer about words they did not write becomes futile. And some departments have turned off Axon’s built-in safeguards—including the intentionally inserted errors (like the frog) designed to force officer review.
California’s SB 524, effective January 2026, now requires AI disclosure in police reports and mandates retention of initial drafts. Utah’s SB 180 requires similar disclosure. The Anchorage Police Department tried Draft One for three months and dropped it—the expected time savings did not materialize.
The vanishing draft: how AI police reports erase their own history
For Further Reading: Perspectives
🟢 PRO — “AI Police Reports: Year in Review” — Electronic Frontier Foundation (Dec. 2025)
🔴 CON — “Concerns About AI-Written Police Reports Spur States to Regulate” — Policing Insight (Oct. 2025)
❧ ❧ ❧
Want to See That Body-Camera Footage? That’ll Be $750.
The promise of police transparency has collided with the crushing cost of actually delivering it—and some states have decided you should pay
For every hour of body-camera footage captured, agencies average three to five hours of labor to blur bystander faces, obscure license plates, and mute sensitive audio. A 10-minute video contains roughly 18,000 individual frames, each potentially containing personally identifiable information that someone must classify and obscure. Multiply by the NYPD’s 29,500 equipped officers—and the scale of the bottleneck becomes clear.
The cost of making footage safe to release has split state legislatures into two camps. On one side, Representative Rashida Tlaib’s Stop Body Camera Paywalls Act would ban fees for agencies receiving federal grants. Washington’s HB 2644 would mandate cost-free public access. On the other: Ohio’s HB 315, signed in January 2026, authorizes police agencies to charge up to 750 per request. Local municipalities immediately began implementing the fees.
Quote
“Public bodies should be in the business of making it easier—not harder—for the public and the press to access important government records like body-worn camera footage.” — Gunita Singh, Reporters Committee for Freedom of the Press
The Ohio law’s defenders point to real costs: officers spending “hours and hours” blurring faces instead of patrolling, and serial records requesters who flood agencies with volume demands. The critics point out that $750 prices out journalists, nonprofits, civil rights organizations, and criminal defense attorneys—the very people for whom body cameras were supposed to be an accountability tool.
AI-powered redaction tools from vendors like CaseGuard and Veritone can detect faces automatically. But they cannot determine whose face it is—distinguishing a protected witness from a public official, a minor from an adult. Because a single missed frame of a protected individual can result in a lawsuit, human review remains legally mandatory. AI changes the title of the person reviewing the footage. It does not eliminate the person.
The California Supreme Court ruled in 2020 that governments must bear redaction costs. Ohio charges the requester. Six state legislatures introduced bills in 2025 to limit public access to body-camera footage through fee structures. The trend is unmistakable: transparency is being converted from a civic right into a market transaction.
Body-camera transparency: the legislative split between free access and paywall states
For Further Reading: Perspectives
🟢 PRO — “Tlaib Introduces Bill to Prevent Restricting Access to Body Camera Footage with a Paywall” — Office of Rep. Rashida Tlaib (2025)
🔴 CON — “Madison Enters the Debate Over Redaction Fees for Police Body-Camera Footage” — Tone Madison (Feb. 2025)
❧ ❧ ❧
EDITORIAL
The Plumber’s Manifesto
There is a theory in public administration, rarely stated so bluntly, that the quality of a democracy can be measured by the boringness of its infrastructure debates. A healthy society argues about water pressure, fiber-optic buildout, and highway maintenance. A sick one argues about whether the plumbing exists at all.
By that measure, American criminal justice is gravely ill. Not because the people inside it are villains—the detectives, prosecutors, lab technicians, and public defenders who populate these pages are, overwhelmingly, professionals doing their best with inadequate tools. The system is sick because the infrastructure connecting them has been neglected to the point of operational insolvency, and the workarounds being deployed to manage that insolvency—statistical manipulation, algorithmic shortcuts, fiscal triage, proprietary lock-in, transparency paywalls—are making the disease worse while making its symptoms invisible.
The pattern is the same in every story in this edition. A public system generates data faster than it can process it. The bottleneck is invisible because the metrics are designed to measure outcomes, not throughput. Workarounds fill the gap—exceptional clearances, solvability calculators, AI-drafted reports, redaction fees—and each workaround degrades the system’s integrity without producing visible failure. Reform proposals target the symptom. The architecture rots.
The prescription is not any single policy but a design principle: treat the infrastructure of criminal investigation as a public utility. Fund forensic labs the way states fund water treatment. Mandate open data standards for ballistic evidence the way CODIS standardized DNA. Build state-operated digital evidence commons that serve prosecution and defense equally. Require evidence portability so that no single vendor can hold public safety data hostage. And insist on AI audit trails so that the chain of authorship is preserved alongside the chain of custody.
None of these reforms require new technology. All of them require something considerably harder to procure: the political will to invest in systems that voters never see and that produce no ribbon-cutting opportunities. The plumbing is invisible. It is also essential. And right now, it is broken in ways that the people who depend on it most—victims of violent crime, defendants exercising constitutional rights, witnesses deciding whether to cooperate—are the last to know.
The justice system does not need another app. It needs a plumber.
For Further Reading: Perspectives
🟢 PRO — “2026 Public Safety Agenda for the Federal Government” — R Street Institute (Jan. 2026)
🔴 CON — “Democrats Wilson, Warren, Hill, Abner for Felony Courts” — Houston Chronicle Editorial Board (Feb. 2026)
Production Note: This edition of The Review was produced through a collaboration between a human editor and an AI assistant (Claude, by Anthropic). All factual claims are drawn from the sourced research report “Drowning in Proof” (December 2025–February 2026), which synthesized federal and state legislation, inspector general work plans, task force reports, investigative journalism, academic research, NIST technical standards, and vendor documentation. Quotes are attributed to their original published sources. Your skepticism remains appropriate and encouraged.
Coming Next: The Courthouse That Runs on Fax Machines—examining how aging court technology intersects with the digital evidence crisis. Also: an interview with the Connecticut forensic lab director who turned a 12,000-case backlog into a 20-day turnaround.
Distribution
The Review is not available to the general public. Distribution is limited to subscribers who have been personally vouched for by existing readers. There are no invitations. If you know, you know.
© 2026 The Review. All rights reserved.
Editor: Daniel Markham | Submissions: submissions@thereview.pub
Content generated Feb. 10, 2026