Bad Investigations Need a Second Opinion. The Vidocq Society Built One Over Lunch.

Since 1990, a Philadelphia club of investigators, pathologists, and scientists has reviewed cold homicides for free over a monthly lunch. It works because it brings fresh eyes and real expertise to cases agencies gave up on. It is also private, volunteer, and invitation-only. States can build the public version. Here is what that would take.

I will just say it: the world is plagued by shitty investigations.

Not mostly the cinematic kind, with a corrupt detective and a missing gun. The ordinary kind. A scene released too fast. A witness asked what she thinks happened before anyone wrote down what she saw. A theory picked on day two and defended for forty years. A file that lives in one detective’s head and a banker’s box, and walks out the door when he retires. I read these files for a living and for this publication, and the failure pattern is boringly consistent.

What almost never happens is a second opinion. In medicine, a surprising diagnosis gets a consult. In aviation, a crash gets a board. In homicide, a stalled case usually gets a drawer.

Thirty-six years ago, three men in Philadelphia decided to be the somebody. Their answer was charming, eccentric, and genuinely useful. It was also never meant to carry the weight of an entire country’s unsolved murders. That job belongs to the states.

The Club That Meets Over Lunch

In 1990, William Fleisher, a federal agent, Richard Walter, a criminal profiler, and Frank Bender, a forensic artist known for age-progressed busts, met for lunch and talked about cold cases. They named their group after Eugène François Vidocq, the 19th-century French criminal turned detective often credited as the first modern private investigator. The Society’s own history says the first meeting was held at the old Philadelphia Navy Yard. It now meets monthly at the Union League of Philadelphia.

The bylaws cap full membership at 82, one member for each year of Vidocq’s life. Additional special members join alongside them. The roster runs across disciplines: federal, state, and local investigators, forensic pathologists, forensic dentists, scientists, polygraph examiners, profilers, and former prosecutors. A 2010 NPR segment pegged the associate bench at about 50 more.

The format is the part everyone remembers. An agency’s lead investigators fly to Philadelphia, the Society covers the trip, and they present their case to a room of specialists eating a multicourse lunch. Members ask questions, poke holes, and suggest what to do next. Journalist Michael Capuzzo turned that room into a 2010 book, The Murder Room, which made the Society briefly famous.

The record of results is real, if hard to measure. The Society’s first case, in 1991, cleared a man wrongly suspected in a Little Rock, Arkansas, murder. It contributed to the resolution of the 1984 killing of Terri Brooks, closed in 1999, and to the 1995 conviction in the 1984 murder of Drexel University student Deborah Wilson. It worked for years on the Philadelphia boy found in a box in 1957, who was finally identified in 2022 as Joseph Augustus Zarelli.

The Society says it has helped solve hundreds of crimes. It also says, on its own site, that it does not charge and that we do not seek credit for our work. That second commitment is admirable and also the reason nobody outside the room can independently score the first one. In 2002, the Society’s case management director told the Washington Times that roughly one in ten accepted cases had led to an arrest and conviction.

How a Case Gets Into the Room

The Society’s published criteria are short and strict. Only the law enforcement agency with jurisdiction can invite it in, through the Society’s case managers. Families and friends are told not to contact the Society and to ask the investigating department to reach out instead. The case must be a homicide, not a missing person. There must be, at minimum, remains indicating homicide, a known crime scene, and physical evidence. The agency must agree to work with the Society. Because members are volunteers, not every case is accepted.

Then there is the criterion that should stop any accountability reader cold. The victim must not have been engaged in criminal activity that caused the death.

Once a case is in, the agency keeps full control of the investigation. Members’ opinions are not binding. The Society can help with DNA testing, blood pattern analysis, autopsy review, polygraph chart study, and 911 call analysis, and members may travel to the scene if the agency invites them and the prosecutor approves. Files are handled under a security protocol, and members sign an annual confidentiality statement.

None of that makes the Society a bad actor. It is a private nonprofit. It can only work with agencies that trust it, and that trust depends on discretion. But those same constraints mark exactly where a private lunch club ends and public infrastructure has to begin.

What the Vidocq Model Gets Right

Strip away the lunch and the rosettes and the Society is running three ideas that every state should steal.

The first is independence. The people reviewing the case did not build the original theory and have no career invested in defending it. That matters more than any single forensic advance, because the most common cause of a stalled case is not missing technology. It is a closed mind sitting on an open file.

The second is range. A homicide file contains pathology, scene work, lab results, interviews, timelines, and sometimes a 911 recording. A single detective, however good, is not expert in all of it. Putting a pathologist, a blood pattern analyst, a prosecutor, and an interviewer in front of the same file catches what any one of them would miss.

The third is cost. Agencies pay nothing. That removes the most common excuse for not seeking help, which is that help costs money the department does not have.

Where the Model Breaks

The agency holds the key. A department that is defensive about its original work, short-staffed, or simply uninterested never has to send the file anywhere. The cases most in need of outside review, the ones handled worst, are the ones least likely to be volunteered for it.

Families have no door. The Society is explicit that families should not contact it. That is a reasonable rule for a volunteer group that cannot vet hundreds of requests. It is an unacceptable rule for a public system, because families are often the only people still pushing.

The worthy victim filter is built in. Excluding victims whose criminal activity caused their deaths sounds narrow. In practice, it maps directly onto the cases police already deprioritize: people who used drugs, did sex work, or lived in neighborhoods where every victim gets assumed into the life. A public review system cannot ration fresh eyes by whether the dead deserved them.

The filter is also hard to square with the man on the letterhead. Eugène François Vidocq was not a detective who happened to know criminals. He was one. In 1798 he was sentenced to eight years of hard labor in the galleys at Brest for helping forge a prison release order. He escaped, was recaptured and sent to the galleys at Toulon, and escaped again. In 1809 he turned himself in and offered to inform. Within a few years he was running the Brigade de Sûreté, and almost every agent he hired was a former convict.

Vidocq’s whole career was the argument that a criminal past does not put a person outside the reach of justice, or outside its service. A rule that asks first whether the dead were clean enough to deserve a second look runs against the very legacy the Society chose to honor.

Capacity is volunteer time. Eighty-two people with day jobs and retirements cannot clear a national backlog. By the Murder Accountability Project’s count, roughly 6,280 homicides went unsolved in 2024 alone.

Nothing is measured in public. Confidentiality protects investigations and earns agency trust. It also means no one outside can learn which kinds of mistakes keep recurring, which is the information that would prevent the next cold case.

States Are Already Building Pieces of This

Colorado got there first. In 2007, the legislature created a cold case unit within the Colorado Bureau of Investigation. By statute, it maintains a statewide database of qualifying homicides and assists local agencies when they ask, and it can also assist when a victim’s family member asks. A 16-member task force seats the attorney general, three district attorneys, sheriffs and police chiefs, victim advocates, a forensic pathologist, and four family representatives appointed by legislative leaders.

Colorado also shows the most common failure: the unit had one full-time position until 2022, when the CBI added ten. The state’s 2025 sunset review counted 2,055 cases in its database, 1,975 of them open and 73 solved, and recommended continuing the task force until 2039 and adding expertise in investigative genetic genealogy.

The federal government gave families an actual right. The Homicide Victims’ Families’ Rights Act, signed August 3, 2022, applies to murders investigated by federal agencies, committed on or after January 1, 1970, and more than three years old, where no probative leads remain and no likely perpetrator has been identified. An immediate family member can apply. The agency must review the file within six months, with one extension allowed. The review must assess missed investigative steps, whether witnesses should be reinterviewed, and whether forensic testing was complete. No one who previously investigated the murder may conduct the review. Agencies must report annually to Congress on applications, reinvestigations, and outcomes.

Oklahoma just built the state version. Senate Bill 1636, signed August 11, 2026, and effective November 1, lets families of murder victims and missing people request a review by investigators who did not work the original case. According to reporting on the law, reviews must include new forensic and DNA testing and witness reinterviews, agencies must confirm a request in writing within 30 days, reviews must be completed within six months, families get regular updates and a final meeting, and a family can request review once every five years. Tribal investigators are included.

Other states looked and walked away. Kansas Senate Bill 242, a state Cold Case Homicide Victims’ Families’ Rights Act, was introduced in the 2023 to 2024 session and died. Minnesota Senate File 1067 was introduced in February 2025 and never left committee.

The Blueprint: What a State Review Team Would Take

Put the Vidocq Society’s strengths together with the best of the federal act, Colorado, and Oklahoma, and the design of a public review team gets specific. This is the general model any state could adopt.

The Michigan Build

Michigan is not starting from zero, which is exactly why it should move.

The Michigan State Police already run district cold case teams. The Third District team reopened a 2003 Saginaw homicide in 2021, worked the evidence with Western Michigan University’s cold case program and the MSP laboratory, used a cold case grant to send reanalyzed evidence to a private lab, and referred the case to the Attorney General in September 2024. Charges followed that December, and a judge ordered the case to trial in August 2025. In January 2025, MSP launched a Seventh District team in Gaylord, led by a detective sergeant with two detective troopers to follow, with a mandate to organize the district’s cold cases and rate each one’s solvability.

The university partnerships are the closest thing Michigan has to a public Vidocq bench. Western Michigan University’s program began in 2020 and partnered with MSP in 2021. Students, hired as cadets or student assistants, digitize files and support detectives, with up to 15 positions each semester. Its work contributed to the 2022 sentencing in the 1987 murder of Roxanne Wood and to the 2023 identification of a primary suspect in the 1988 murder of Cathy Swartz, and local reporting in April 2026 credited the program with helping solve seven cold cases. Michigan State University announced a second cold case solved with MSP in May 2025.

What Michigan does not have is a law that ties any of it together. A family in this state has no statutory right to request review of a local homicide, no deadline for an answer, and no guarantee the reviewer is anyone other than the agency that let the case go cold. With a long-run clearance rate of 54 percent and nearly 21,000 unsolved homicides by the Murder Accountability Project’s count, the gap is not abstract.

Raising the Floor on First Investigations

Cold case review is a hospital for cases that were mishandled or unlucky. A serious state would also work on prevention.

Most of what kills a case happens early. In the Clutch Justice 30-Day Evidence Preservation Protocol, a Field Kit guide written for ordinary people in legal crises, the starting premise is that evidence is perishable and most of it is lost in the first 72 hours, not through bad faith but because nobody recognized it as evidence in time. The protocol tells individuals to write a factual narrative before interpreting anything, to capture records before they are overwritten, and to list witnesses before talking to them, because asking a witness what they think happened contaminates what they saw.

If we expect a person who just got fired to follow that discipline, we can expect it of a homicide unit. Translated into institutional standards, that means a written scene and evidence log opened on day one; independent documentation of each witness’s account before anyone shares theories with them; preservation requests for video, phone, and digital records issued in the first days; a standard file index so a case can be handed to someone else without losing its logic; and a mandatory written supervisory review at 30 days by someone outside the case’s chain, asking what has not been done yet. Each of those is cheap. Each of them is the kind of thing a cold case reviewer later discovers was never done.

The feedback loop closes the system. When a review panel finds that scenes keep getting released early, or that a lab keeps sitting on items, that finding should go to the people who license and train officers, every year, by name of the failure and not the officer. That is how aviation got safer. It is how homicide investigation could too.

Why I Keep Coming Back to This

I have spent this month inside 1977 records from the Sigrid Stevenson case in New Jersey, and on any given week I am reading Michigan files that look a lot like it. Different decades, different agencies, same shape. The gaps that matter most are rarely technological. They are a step nobody took, a record nobody kept, and a theory nobody outside the original team was ever allowed to test.

The Vidocq Society showed what happens when you let experts who owe nothing to the original theory look at the file. They did it on their own time, at their own expense, for agencies that chose to ask. States can do it on purpose, for every family that asks.

Bad investigations will keep happening. They do not have to stay buried.

The Vidocq Society set a lunch table for the dead. States should build the room, unlock the door, and let the families in.