The FBI’s portrait of the Baton Rouge killer was a careful document. It described a strong man of modest income, impulsive, unable to take rejection, awkward with the women who drew his interest, the kind of man who followed and watched them and might already have a record for trespassing, burglary or peeping. Much of it would prove close to the truth. What the investigation took from it, for months, was something else: the idea that the man was white. By the time that assumption was dropped, in March 2003, five women whose deaths would be tied together by DNA had died, and the detectives of a small town just to the north had been trying since the previous summer to get anyone to look at the man they suspected.
An offender profile is an outline of the person police should be looking for, drawn from the crimes themselves. A good profile narrows a search. A bad one, or a sound one read too literally, can close doors that should have stayed open. The cases that follow, from Louisiana to West Yorkshire to Perm Oblast in Russia, are the ones where the picture and the man did not match, where experts gave juries opinions the juries rejected, and where a profile was handed credit after an arrest it had nothing to do with.
Gina Wilson Green, a nurse, died in Baton Rouge in September 2001. Geralyn Barr DeSoto, 21, who had just registered for graduate classes at Louisiana State University, died in Addis, across the Mississippi River, on January 14, 2002. Charlotte Murray Pace, 22, the youngest graduate of her MBA class at LSU and about to start work in Atlanta, died in Baton Rouge on May 31, 2002. Pam Kinamore, 44, a Baton Rouge wife and mother who ran an antiques shop in Denham Springs, died in July 2002. On July 29, investigators announced that DNA tied the deaths of Green, Pace and Kinamore to one man.
A Multi-Agency Homicide Task Force took over, and the FBI’s Behavioral Analysis Unit prepared a profile. The version the task force released in December 2002, as WAFB printed it, did not name a race. The search around it did. Newsweek reported in 2003 that the authorities had relied for months on an FBI profile that assumed the killer was white, and that it was amended only in March 2003 to cover men of all races. Tips about white men, and about a white pickup truck, shaped much of the work.
The cost showed first in the case of Trineisha Dené Colomb. She was 23, from Lafayette, a reader who dreamed of traveling and who was due to join the Marine Corps in February 2003. Her death came to light near Lafayette in November 2002. Investigators at first did not count her as part of the series, because she was Black and the women before her had been white. On December 23, 2002, they announced that DNA linked her to the same man. "I never dreamed my daughter would be killed by a serial killer," her father told WAFB.
Carrie Lynn Yoder, 26, a doctoral student in biological sciences at LSU who studied coastal ecology, died in March 2003. That month a Florida company, DNAPrint Genomics, said publicly that it could estimate a person’s ancestry from trace DNA, and WAFB reported on March 29 that the task force had opened its search to men of all races. Charles Brenner, a forensic statistician consulted on the case, wrote in 2004 that the change came shortly after an ancestry analysis by DNAPrint, and that he had reached essentially the same conclusion from the ordinary crime-laboratory profile.
The task force did not identify Derrick Todd Lee. Police in Zachary, the town to the north, had suspected him for years in two older cases, and in July 2002 they told Baton Rouge investigators that he was their prime suspect in two killings. Zachary’s detectives were not brought onto the task force. In the spring of 2003, working with the state Attorney General’s Office, they obtained a court order for a sample of his DNA, and on May 5, 2003, investigators took cheek swabs from him. The State Police Crime Lab matched them to the task force’s profile later that month. Joey Watson, Zachary’s police chief, told Newsweek that leads had been lost because of the assumption about race. The Baton Rouge police answered in November 2003 that a white truck linked to Lee by a caller had been checked, and that Lee had not been tested earlier because of task force priorities.
Lee, 34, a laborer from St. Francisville with a long record of peeping, stalking and burglary, was arrested in Atlanta on May 27, 2003. In August 2004 a West Baton Rouge Parish jury convicted him of the second degree murder of Geralyn DeSoto, and he was sentenced to life. In October 2004 an East Baton Rouge Parish jury convicted him of the first degree murder of Charlotte Murray Pace and chose death; the sentence was imposed on December 10, 2004, and the Louisiana Supreme Court affirmed it in 2008. He was never tried for the other deaths linked to him by DNA. He died of heart disease on January 21, 2016, while still on death row.
The irony of the Baton Rouge profile is that its behavioral sketch of a watcher and a prowler described the real man’s record well. The single assumption laid over it, about the color of his skin, kept him outside the search.
In the fall of 1987 Richmond, Virginia, was gripped by a series of killings its newspapers attributed to one unknown man. Debbie Dudley Davis, 35, died in Richmond in September 1987. Dr. Susan Hellams, 32, a neurosurgery resident, died in Richmond in October. Diane Cho, 15, a high school student, died in Chesterfield County, next to the city, in November. Ray Williams, a Richmond homicide detective, told WTVR in 2014 that the city at night was "lit up like Christmas" because people left their lights burning.
According to WTVR, criminal profilers working with the Richmond police believed the killer was probably a local white man, and the investigation produced no solid leads. Joe Horgas, an Arlington County detective who came into the case later, said the Richmond investigators were focused on a white suspect.
Horgas came to it from the north. Susan Tucker, 44, a writer, died in Arlington in late November 1987, and he saw at once how closely the case resembled the 1984 killing of Carolyn Hamm, a 32-year-old lawyer, a few blocks away. Another man, D.V., had entered a plea in the Hamm case in 1985 and was serving 35 years. Horgas visited him in prison in December 1987 and came away doubting that he belonged there.
On December 29, 1987, two agents of the FBI’s Behavioral Science Unit, Stephen Mardigan and Judson Ray, reviewed Horgas’s files. Their advice was practical. A man like this usually begins where he feels comfortable, near home, and stops only when he dies or is locked up. They suggested that Horgas look for a man arrested for burglary soon after the Hamm killing and released shortly before the Richmond deaths began. The description proved accurate, but the search it prompted could not have worked: Arlington Magazine noted that a release to a halfway house was not technically parole, so the man’s name would not have appeared in the records Horgas and his partner went through by hand. What produced the name was Horgas’s memory of a neighborhood burglar called Timmy from a decade earlier. Timothy Wilson Spencer had been arrested for a burglary in Alexandria four days after Carolyn Hamm’s death and had been released to a Richmond halfway house in September 1987.
Spencer had grown up in Arlington, not in south Richmond, and he had been in the city only a few weeks when the killings began. Horgas arrested him on January 20, 1988. DNA testing by Lifecodes, a laboratory in New York, connected him to the killings, and in July 1988 an Arlington jury convicted him of capital murder in the death of Susan Tucker, a trial Arlington Magazine and WTVR have described as the first capital murder conviction in the United States built on DNA evidence. Later juries convicted him of capital murder in the deaths of Debbie Dudley Davis, Susan Hellams and Diane Cho. The Supreme Court of Virginia affirmed all four death sentences. He was executed on April 27, 1994.
The FBI’s behavioral experts also helped undo the earlier mistake. After the Arlington verdict, police, prosecutors and the FBI’s National Center for the Analysis of Violent Crime reexamined the Hamm killing and concluded that another person was responsible; John Douglas, the FBI profiler, helped with a 35-page report drafted by Mardigan. Governor Gerald Baliles granted D.V. an absolute pardon on January 4, 1989. Spencer was never charged with the Hamm killing.
The summer of 1990 brought three deaths to the Eastside suburbs of Seattle. Mary Ann Pohlreich, 27, died in Bellevue in June. Carol Marie Beethe, 35, died in Bellevue in August. Andrea Levine, 24, known as Randy, died in August; her death came to light in Kingsgate in early September. The Bellevue police treated them at first as separate cases.
Early in the investigation, according to B. J. Welborn’s account in Great American Trials, detectives consulted an outside expert on serial offenders, who concluded that one man had killed all three women and that he was probably a young white man. At trial, the question of whether one man was responsible became the center of the case, and the profilers divided.
Robert Keppel, an investigator for the Washington Attorney General’s Office, testified for the state that the killings were the work of one person, drawing on a statewide database of homicide reports. John Douglas, testifying for the prosecution, said the three cases shared a signature and took place within 67 days in a small area; the differences, he said, were small beside the similarities. For the defense, Russell Vorpagel, a retired FBI agent working as a private investigator, told the jury the crimes were too different to be the work of one man. A second defense expert, Robert Gebo, had first thought that the Pohlreich and Beethe cases were unconnected; under cross-examination he said he now believed one person had committed all three, and the state showed in rebuttal that he had changed his mind after talking with Keppel and Douglas, not because of anything he had learned about the defendant.
In October 1991 a King County jury convicted George Waterfield Russell Jr., 32 when he was charged, of the first degree murder of Mary Ann Pohlreich and the aggravated first degree murders of Carol Beethe and Andrea Levine. Judge Patricia Aitken sentenced him to two terms of life without parole and a third term of nearly 29 years. On appeal the Washington Supreme Court held in 1994 that the experts' opinions that one man had committed all three murders were properly admitted. The behavioral testimony that lost was the testimony that called the killings unrelated.
The Yorkshire Ripper inquiry never relied on a Quantico-style profile. It built one of its own out of a hoax. In March 1978 letters began to arrive from a man who said he was the killer of women in West Yorkshire and Manchester. The first, signed "Jack the Ripper," went to the police on March 8; a second went to the Daily Mirror five days later; a third followed. In June 1979 the writer sent Assistant Chief Constable George Oldfield a cassette tape. "I’m Jack," it began, and the voice told Oldfield he was no nearer catching the killer than when he had started.
Oldfield accepted the letters and the tape as genuine. Dialect experts placed the speaker’s accent in the Castletown area of Sunderland, and detectives were told they could eliminate suspects who lacked a Wearside accent. The tape was played to the press, piped down a "Dial-the-Ripper" telephone line and advertised on billboards and in newspapers, in a campaign that cost about a million pounds. Some 40,000 men in the North East were investigated. Voice experts and some detectives warned that the material was a hoax, the Press Association later reported, and women who had survived and heard the man speak said he sounded local to West Yorkshire.
The man responsible was a truck driver from Bradford, and the inquiry had him in front of it again and again. West Yorkshire Police interviewed Peter Sutcliffe nine times. In July 1979, at the fifth interview, two detectives, Andrew Laptew and Graham Greenwood, were suspicious enough to say so in a written report. It was filed away because his voice and handwriting did not match the letters and the tape. Laptew took his concerns to a senior officer, Detective Superintendent Dick Holland, who made him feel, he said later, so small that he "could have crawled under the crack in the door." Sutcliffe said later that while the hunt was looking toward Sunderland, he felt safe.
Oldfield fell ill and was moved off the case, and in November 1980 Detective Chief Superintendent James Hobson took over and reduced the weight given to the tape. Sutcliffe was arrested in Sheffield on January 2, 1981, by two South Yorkshire officers. He pleaded guilty to seven attempted murders on April 29, 1981, and on May 22 an Old Bailey jury convicted him of the murders of 13 women. He received 20 concurrent life sentences.
The government’s account of the review by Lawrence Byford, one of Her Majesty’s Inspectors of Constabulary, went to Parliament on January 19, 1982. It spoke of "major errors of judgment," and it put the tape at the center: "Excessive credence was given to the letters and tape from a man claiming responsibility." Had those errors not been made, the statement said, Sutcliffe would have been identified as a prime suspect sooner.
The hoaxer was identified in 2005, when a cold case review obtained a DNA profile from saliva on one of the envelopes and matched it to John Humble, a laborer from Sunderland. He pleaded guilty at Leeds Crown Court on March 20, 2006, to four counts of doing acts intended to pervert the course of justice, and was sentenced the next day to eight years in prison. He died in 2019. Sutcliffe died in custody on November 13, 2020.
Not every profiling failure is a wrong prediction. In the Russian town of Lysva, in Perm Oblast, a profile was right in part and was then credited with an arrest it played no part in.
By May 1998 a series of killings of girls and women in Lysva had produced several unsolved criminal cases. A psychologist with the Perm Oblast police drew up a written assessment of the likely offender, dated May 26, 1998, working from the case files and from methods the Russian Interior Ministry used for serial violent crimes. It offered three alternative portraits. One described a married man of 36 to 42 with children, and nothing in it proved correct. The other two described a man of 18 to 25 who lived and worked near the area of the crimes and had probably suffered a head injury.
On the evening of June 10, 1998, Vladimir Bratilov, 22, was detained in Lysva after a woman he had robbed that evening identified him to police officers. He lived in that part of town and had been discharged from the army after a head injury. He confessed. On April 29, 1999, the Perm Regional Court convicted him of eight killings, with other charges including robbery, and sentenced him to life imprisonment. Russia’s Prosecutor General’s Office described the case as 11 attacks, eight of them fatal.
On December 31, 1998, the Prosecutor General’s Office sent out an information letter on the use of psychology in investigating serial murders. It cited five cases from different regions, Bratilov’s among them, and presented his arrest as the result of the search features in the Perm psychologist’s profile. In 2007 the legal scholar V. N. Kitaeva went back to the Perm Regional Court’s file. Her conclusion, published in the journal Sovremennoe Pravo, was that the arrest came from the woman’s identification and had nothing to do with the profile. The profile’s young local man with a head injury fit Bratilov, but in her reading it did not bring anyone to his door. She reached the same view about all five cases in the letter.
The hunt for the killer of women and children in Rostov Oblast in the 1980s produced one of the earliest Soviet offender profiles. Viktor Burakov, a forensic specialist with the regional police, asked Aleksandr Bukhanovsky, a psychiatrist at the Rostov Medical Institute, for a psychological portrait. In Owen Matthews’s 1999 account in The Moscow Times, Bukhanovsky had inferred a middle-aged, educated man. The police paid the portrait little attention at first.
Andrei Chikatilo, a former teacher, was detained in September 1984 and released after a blood-typing result appeared to clear him. He was arrested in Novocherkassk on November 20, 1990. Bukhanovsky was brought in after the arrest and introduced to him as an academic, and in his account and Burakov’s, it was the psychiatrist who persuaded Chikatilo to confess.
On October 15, 1992, the Rostov Regional Court convicted Chikatilo of 52 murders and sentenced him to death. In 1993 the Supreme Court of Russia removed nine murder episodes for insufficient evidence and upheld the sentence on the 43 that were left. He was executed on February 14, 1994. Whether his portrait helped catch Chikatilo, or merely described him after the fact, rests largely on the accounts of the men who asked for it and wrote it.
For three weeks in October 2002, people were shot at random around Washington, D.C. Ten were killed and three were badly hurt in Maryland, Virginia and the District of Columbia, according to the FBI. The task force, led by the Montgomery County Police Department under Chief Charles Moose, asked for federal help under a law on serial killings, and within days the FBI had about 400 agents on the case.
Television filled the gap with profilers. Writing in The Weekly Standard as the case closed, Eli Lehrer noted that nearly every profiler on the air had assumed the shooter was a white man. Candace DeLong, a former FBI agent, had said the sniper was "less than 25 years old" and worked alone. The men arrested on October 24, 2002, were John Allen Muhammad, 41, a former Army sergeant, and Lee Boyd Malvo, 17, a Jamaican-born teenager who had been living with him; Lehrer pointed out that both were Black.
The Bureau’s own reading also missed. Among the material that Muhammad’s lawyers later said prosecutors had withheld was an FBI profile suggesting a lone shooter. The United States Court of Appeals for the Fourth Circuit held in 2009 that none of the withheld material was both favorable and material to his defense, because the rest of the evidence was overwhelming.
A Virginia jury convicted Muhammad on November 17, 2003, of capital murder in the death of Dean Harold Meyers, and he was sentenced to death. A Maryland jury convicted him on May 30, 2006, of six counts of first-degree murder. Malvo pleaded guilty to the six Maryland murders and was convicted in Virginia of two others. Muhammad never admitted the shootings. He was executed in Virginia on November 10, 2009.
The wrong picture of an offender does not always come from a behavioral unit. In Los Angeles, beginning on December 1, 1974, a series of killings of men in the downtown area and Hollywood went on for about two months. Early police descriptions of the suspect, as Los Angeles magazine later summarized them, were of a white man in his late twenties or early thirties, about six feet tall, roughly 190 pounds, with long blond hair. Vaughn Orrin Greenwood, arrested in Hollywood in early 1975, was a Black man of 32. He was convicted of nine counts of first-degree murder and sentenced to life in prison on January 19, 1977.
Set side by side, the failures look less like bad psychology than like a narrow reading of it. The Baton Rouge profile’s account of a prowler with a record was sound; the race laid over it was not, and the young woman from Lafayette whose death did not fit the assumption was at first left out of the series. The Richmond profilers looked for a local white man; the FBI agents Horgas consulted gave him a description of a burglar’s record that matched the real man’s. In West Yorkshire, a voice on a tape became a test that let the killer through.
None of these men was named by a profile. Lee’s name came from detectives in a neighboring town and a court order. Spencer’s came from a detective’s memory of a neighborhood burglar from a decade before. Bratilov was identified by a woman in Lysva, and Sutcliffe was stopped by two officers in Sheffield. Around Washington, the snipers' own telephone boast about a shooting in Alabama gave investigators the thread they needed. The profiles that did their job did it by narrowing a search. The ones that failed did it by deciding, too early, who could not be the man.
• December 1, 1974: the Los Angeles series later charged against Vaughn Greenwood begins.
• July 1975 to November 1980: the West Yorkshire and Manchester killings and attempted killings for which Peter Sutcliffe was later convicted.
• January 19, 1977: Greenwood is sentenced to life in prison.
• March 1978: letters from a man claiming to be the Yorkshire killer begin to arrive.
• June 1979: the hoax tape reaches George Oldfield.
• July 1979: two detectives report suspicions about Sutcliffe; the report is filed because his voice and handwriting do not match the hoax.
• November 1980: James Hobson takes over the inquiry and reduces the weight given to the tape.
• January 2, 1981: Sutcliffe is arrested in Sheffield.
• May 22, 1981: Sutcliffe is convicted of 13 murders.
• January 19, 1982: the government gives Parliament the conclusions of the Byford review.
• January 1984: Carolyn Hamm dies in Arlington, Virginia; D.V. later enters a plea in the case.
• 1980s: Aleksandr Bukhanovsky prepares a psychological portrait of the Rostov killer.
• September 1984: Andrei Chikatilo is detained and released.
• September to November 1987: Debbie Dudley Davis, Susan Hellams, Diane Cho and Susan Tucker die in Richmond, Chesterfield County and Arlington.
• December 29, 1987: FBI agents Stephen Mardigan and Judson Ray advise Detective Joe Horgas.
• January 20, 1988: Timothy Wilson Spencer is arrested.
• July 1988: Spencer is convicted in Arlington of the capital murder of Susan Tucker.
• January 4, 1989: Governor Gerald Baliles pardons D.V.
• June to August 1990: Mary Ann Pohlreich, Carol Marie Beethe and Andrea Levine die on the Eastside of Seattle, in Bellevue and Kingsgate, Washington.
• November 20, 1990: Chikatilo is arrested in Novocherkassk.
• October 1991: George Waterfield Russell Jr. is convicted of three murders in King County.
• October 15, 1992: Chikatilo is convicted of 52 murders.
• February 14, 1994: Chikatilo is executed.
• April 27, 1994: Spencer is executed.
• October 13, 1994: the Washington Supreme Court upholds the Russell convictions and the expert testimony.
• May 26, 1998: a Perm police psychologist completes a profile of the Lysva offender.
• June 10, 1998: Vladimir Bratilov is detained in Lysva.
• December 31, 1998: the Russian Prosecutor General’s Office credits the profile in an information letter.
• April 29, 1999: Bratilov is sentenced to life imprisonment.
• September 2001 to March 2003: the deaths in and around Baton Rouge and Lafayette, Louisiana, later linked to Derrick Todd Lee.
• July 29, 2002: DNA links three Baton Rouge deaths to one man.
• October 2002: ten people are killed in the Washington, D.C., region; John Allen Muhammad and Lee Boyd Malvo are arrested on October 24.
• December 23, 2002: DNA links Dené Colomb’s death to the Baton Rouge series.
• March 2003: the Baton Rouge search is opened to men of all races.
• May 27, 2003: Lee is arrested in Atlanta.
• November 17, 2003: Muhammad is convicted of capital murder in Virginia.
• August and October 2004: Lee is convicted of the murders of Geralyn DeSoto and Charlotte Murray Pace.
• 2005: DNA identifies John Humble as the Sunderland hoaxer.
• March 20 and 21, 2006: Humble pleads guilty to perverting the course of justice and is sentenced to eight years.
• May 30, 2006: Muhammad is convicted of six murders in Maryland.
• 2007: V. N. Kitaeva publishes her study of the Lysva profile.
• August 7, 2009: the Fourth Circuit rejects Muhammad’s claim about withheld material, including an FBI profile.
• November 10, 2009: Muhammad is executed.
• January 21, 2016: Lee dies on death row.
• November 13, 2020: Sutcliffe dies in custody.