For background on this story, see my 2016 series, “Dead Wrong.”
In the late 1990s, Halifax police were dealing with a disturbing number of unsolved murders, and in 1997, an RCMP–Halifax police task force, “Operation Full Course,” was established to look at cold cases.
Operation Full Course was primarily focused on Andrew Johnson, a man picked up by police in Nanaimo, BC as he disguised himself as a police officer and lured young girls to his car. Johnson was a suspect in the disappearance of two women in Halifax — Andrea King and Kimberly McAndrew.
But then, in 1998, Michael McGray was arrested for the brutal murder of Joan Hicks, a Moncton woman who was murdered alongside her 11-year-old daughter, Nina. McGray was subsequently charged and convicted of six murders (and, later, a seventh after the 2010 murder of his cellmate, Jeremy Phillips).
By 1998 and into 1999, McGray had told his relatives and cellmates that besides the six murders he had been charged with, he had killed as many as 11 other people. So in 1999, Halifax police Sgt. Dave Worrell, an investigator working in Operation Full Course, created a timeline covering unsolved murders along with Michael McGray’s known residences through the years.
The unsolved murders on the timeline include 17-year-old Tina Barron (her body discovered in a ditch in Indian Brook on November 4, 1985); 27-year-old Ann Mason (killed in a Robie Street rooming house on March 28, 1986); 19-year-old Kimberly McAndrew (last seen in the parking lot of the Quinpool Road Canadian Tire on August 12, 1989, and presumed murdered); 31-year-old Jean Myra (her body found at the grain elevators on April 5, 1990); 21-year-old Carla Strickland (her body discovered in Shubie Park on June 5, 1991); 17-year-old Shelley Connors (her body found by the Spryfield Lions Rink on June 1, 1993); 25-year-old Kimber Lucas (her body found behind 5783 North Street on November 23, 1994); 52-year-old Jaime Wyatt (the only man on the list, he went missing from the Twin Elms Motel on Inglis Street on November 12, 1996); 30-year-old Crystal Jack (went missing from her Agricola Street apartment on July 15, 1997, her remains discovered off Highway 101 in 2011); 32-year-old Rachel MacQuarrie (last seen on September 27, 1997, she lived about two blocks from the site of Brenda Way’s murder two years before; her remains discovered off Highway 101 in 2002); and 35-year-old Robin Hartrick (her body discovered on September 10, 1999 behind Harbour View School, about two blocks from the site of Brenda Way’s murder).
Notably absent from the Operation Full Course timeline was Brenda Way, who was killed on November 12, 1995, the morning after Remembrance Day. But police had considered McGray for Way’s murder, at least briefly.
According to a memo written by Halifax police detective Wayne Hurst in 2005, back in 1999 Worrell, the creator of the timeline, saw that McGray was not in custody when Way was killed in 1995, so Worrell asked Constable Steve Maxwell to question McGray about Way’s murder and other unsolved murders committed when McGray was not in custody. Wrote Hurst:
Cst. Maxwell stated he asked McGray if he was responsible for the Brenda Way murder, and McGray replied that he knew of Brenda, as he lived handy, however McGray stated that we got the right guy for that, referring to Glen Assoun.
McGray’s phrase that he “lived handy” to Brenda Way is interesting. In 1995, McGray was living in north end Dartmouth, while Way was staying clear across Dartmouth on Lawrence Street (behind what is now the Braemar Superstore), albeit Way was a transient sort who worked as a sex worker all over Dartmouth, and especially along the Victoria Road and Windmill Road stroll.
Worrell’s timeline isn’t exact — the residences where McGray lived with his girlfriend Tammy McLean are plotted only as to known times when he lived somewhere — but it showed that around Christmas of 1995 McGray was living on Jackson Road:
(In 2014, a person who is not named in court documents signed an affidavit saying McGray and McLean moved into the Jackson Road basement apartment two or three weeks prior to Way’s murder, and that the couple “suddenly moved out of the apartment within a few days of the murder, throwing out all their furniture and leaving their two cats behind.”)
Brenda Way was found murdered behind an apartment building at 109 Albro Lake Road. The apartment building is on the north side of the street. The next street north of Albro Lake Road is Jackson Road. McGray was living in a basement apartment at 48 Jackson Road, which is on the south side of that street. The apartment is about 100 metres from the murder site.
Halifax police investigators demonstrated an astonishing degree of credulity by simply taking McGray’s word that he hadn’t killed Way. As Justice Department lawyer Mark Green pointed out in his 2014 Preliminary Assessment of Glen Assoun’s case, McGray had previously denied some of the killings he was later convicted of, including Nina Hicks, the 11-year-old girl killed alongside her mother in Moncton, and Mark Gibbons, his co-conspirator in a botched 1987 robbery of a cab driver in Saint John. Through the years, McGray has offered to give police details of other murders, but only on the condition that he and unnamed accomplices be given immunity — a deal police have rejected. Arguably, McGray would keep details of the Way murder to himself in hopes of using them as a future bargaining chip.
Moreover, a person with deep knowledge of the Way murder investigation tells me that no attempt was made to further investigate McGray as the possible murderer. His prison visitor logs were not checked; his recorded phone calls from prison were not listened to to see if he had mentioned the Way killing; his prison correspondence was not reviewed; the psychiatric reports on McGray were not read. It was simply: McGray said he didn’t do it, case closed.
In any event, Halifax police considered, and then rejected, McGray as a suspect in the Brenda Way murder in 1999 — just as Glen Assoun was being tried for the murder.
As I wrote in Part 1 of this article, police improperly threatened and cajoled witnesses to provide false testimony against Assoun, and those witnesses’ testimony ultimately led to what we now know was his wrongful conviction.
In December 1999, Assoun was sentenced to life in prison with no chance of parole for 18 and a half years. He steadfastly maintained his innocence and immediately started his attempts at filing an appeal.
Assoun eventually sent a letter to the Association in Defence of the Wrongly Convicted (AIDWYC, now called Innocence Canada), but AIDWYC wouldn’t consider getting involved until all appeals were exhausted, so the organization passed Assoun’s letter on to Newfoundland lawyer Jerome Kennedy.
In 2000, Kennedy had won the release of Ron Dalton, a man who was wrongfully convicted of murdering his wife; after that success Kennedy looked around for other potential wrongful convictions and took up Assoun’s case. Kennedy hired retired RCMP officer Fred Fitzsimmons as an investigator.
Before we go any further in this story, we need a quick primer on something called the Violent Crime Linkage Analysis System (ViCLAS). We have to have some understanding of ViCLAS before we can understand what happened in the Assoun case.
The ViCLAS database was created after police had missed important information in the Paul Bernardo/ Karla Homolka case.
ViCLAS was developed by the RCMP, which explains the system as follows:
The questions [used to build the data for the database] were put together in booklet form and were designed to eliminate as many open-ended questions as possible. This allows for standardized data collection and more efficient search and find capabilities. The booklet, available in French and English, should be completed by the investigator. It takes approximately two hours to complete and can be used as an investigator’s guide. If the investigator can answer every one of the questions, he or she can be assured they have conducted a thorough investigation.
But are the data that populate ViCLAS reliable? This is the question poised by an academic study of ViCLAS, “The Violent Crime Linkage Analysis System : A Test of Interrater Reliability,” which noted:
For the data contained in ViCLAS to be of value for the aforementioned purposes, the assumptions on which it is based must be valid. Perhaps one of the most fundamental assumptions underlying ViCLAS (and all other all linkage systems) is that the database contains reliable data. The primary type of reliability of concern is interrater reliability. A test of interrater reliability involves determining the extent to which two (or more) different investigators enter the same information about a case into a ViCLAS coding booklet (e.g., both investigators agree that the victim was a nurse). In scientific research, a minimum level of 80% agreement is typically deemed acceptable before the inferences and conclusions drawn from coded data are trusted (e.g., Hartmann, 1977). It seems logical that a similarly high level of agreement should be demanded from crime linkage systems because of the consequential nature of the inferences drawn from the data held in these systems.
The study’s authors (who are based out of Memorial University in Newfoundland and include the superintendent of the Royal Newfoundland Constabulary) noted that there is no training required to input data into ViCLAS, and that there has been only one previous study conducted of ViCLAS, which was coauthored by the manager of research and development for the Behavioural Sciences Research Branch of the Royal Canadian Mounted Police — the developers of ViCLAS — so not truly independent (my words, not the academics’).
The Newfoundland researchers wrote that “the level of occurrence agreement found in the current study was, for the vast majority of ViCLAS variables, unacceptably low” — agreement ranged from a low of 2.36% for weapon variables to a high of 62.87% for administration variables — and that “our findings raise serious concerns about the validity of inferences drawn using ViCLAS data and, potentially, the validity of inferences drawn from other linkage systems in use around the world.”
Our results … suggest that the data contained in ViCLAS may be unreliable. Continued use of a system that may contain unreliable data is difficult to justify, especially given the many serious consequences attached to linkage decisions. For example, pursuing potential linkages derived from unreliable data can be a waste of valuable police resources and taxpayer dollars and may even result in individuals being improperly considered or even falsely accused of crimes they did not commit. Time spent on ViCLAS-related activities also takes investigators away from other important tasks that may be more pertinent.
The “other linkage systems in use around the world” line intrigued me. That’s because the RCMP has shilled ViCLAS around the planet; explains the RCMP:
Since the inception of ViCLAS, a number of countries have adopted ViCLAS and are using it as their violent crime linkage analysis system. In recent years, countries interested in obtaining ViCLAS must submit a very detailed business case explaining how they plan on using it.
Upon approval by the Officer in Charge of ViCLAS, the country must sign a Licencing Agreement and pay an associated annual fee based on the number of ViCLAS users which starts at $15,000 CDN.
Presently, the following countries use ViCLAS: Belgium, Czech Republic, France, Germany, Ireland, Netherlands, New Zealand, Switzerland and the United Kingdom. Many have translated the ViCLAS questionnaire into their own language.
I wondered whether the sale of ViCLAS internationally might mean that the RCMP has some interest in keeping the details of the operation of ViCLAS secret — recall that RCMP Corporal Roger Robbins wrote that “it is the RCMP’s position that public disclosure of ViCLAS information, even a blank booklet, would be injurious to this sensitive and extremely important investigative technique.” That is, is the secrecy really about keeping investigative techniques protected, or is it about intellectual property rights and profit?: Keep the mystique about ViCLAS, guard it as an un-duplicatable secret, and there’s more money to be made through licensing, no matter the effectiveness (or not) of the system.
I put that theory to the Newfoundland researchers and asked what they thought.
Primary author Brent Snook responded while he was on vacation:
It is hard to say why they want to keep the booklet a secret — the reasons you outlined seem reasonable.
Coauthor Kirk Luther responded at greater length:
The major issue we see is that the only two published studies examining ViCLAS have revealed the data entered into the booklets was unreliable. That is, the study participants entering the information did not all agree on the case facts and how they were entered into the ViCLAS booklets. We see this as a major issue as ultimately you can’t trust the reliability of the data entered into the system — the data used to make real-world linkage decisions. Of course, the claim may be made that the new computerised questionnaire or the ViCLAS Specialists reduce these errors, the RCMP would need to provide relevant data to ensure accuracy and reliability.
A more broad concern is whether crimes can actually be linked. Our previous work has called into question the reliability of offender and geographic profiling, which seems to be one of the aims of ViCLAS — determine if there are any linkages in crimes within/across jurisdictions based on a number of variables (e.g., suspect actions, victim demographics, location).
I can’t speak for why the RCMP don’t want to release a blank booklet. It may be that they have the genuine concern of offenders altering their behaviour to avoid being linked to previous crimes. I am not aware of any literature stating that offenders alter behaviour in such a manner, although it may be out there. However, the major issue is that if you are using unreliable data, it does not really matter what offenders are doing. Further, as your article stated, if you Google around for a bit you can pretty much find the questions used for ViCLAS.
The RCMP created ViCLAS following high-profile, multi-jurisdictional serial violent crimes in the 1980s (e.g., Clifford Olson). My understanding is that it was modelled after the FBI’s ViCAP system. The RCMP has reported that its ViCLAS system is used by law enforcement agencies in the US, Belgium, Netherlands, Czech Republic, New Zealand, Switzerland, Germany, and UK. However, they do not provide data on success rates — e.g., X amount of ViCLAS linkages have resulted in a successful prosecution.
I can’t draw any definitive conclusions from this. But I had the questionable effectiveness of ViCLAS and the RCMP’s desire to guard secrecy of the workings of the database top of mind when I read through the court documents in the Assoun case that were released Friday.
ViCLAS and Michael McGray
Dave Moore was the owner of a Ottawa software company recruited by the RCMP in 1991 because of his computer expertise. In January 2001 he became a member of the RCMP Criminal Analytical Section, where among other projects, he worked with the ViCLAS unit in Halifax.
In a 2014 interview with Justice Department lawyer Mark Green, Moore explained that he had reservations with the ViCLAS system:
As a systems analyst, when you design software, there is always a fail-safes that you make sure are contained within the program. If the fail-safes are not present within the software then you have to be cognizant of those fail-safes not being there, otherwise you generate a false hit.
So a good example of that in the primary software that we had is instead of leaving the information blank if we didn’t know how tall the person was, if you put in a zero, the formula the computation formula for the general averaging to give you a determination would give you a person that’s 24 inches high.
… So because it’s a formula driven software it’s very easy to contaminate the software with false data also generating false information. So you have to be careful in your review of it…. Most if not all of the other members of my class across the country don’t have my ability to decipher the programming.
Around 2002 or 2003, Moore was assigned to look at Michael McGray and to try to determine if ViCLAS could identify any other of McGray’s victims — that is, the 11 other people McGray claimed to have killed besides the six murders he had been convicted of.
“The [ViCLAS] system was stumbling,” Moore told Green. “And the reason it was stumbling is that McGray is an anomaly, in that where the system was designed to track similarities, this was an individual who never does the same thing the same way and his victimology is all over the place. He kills a mother and children. He kills older male gays. He kills anybody and everybody. So whenever that occurs your ability to work within the system which is built on parity is very difficult.”
Moore soon came across the murder of Brenda Way. Wrote Green:
Brenda Way was identified, despite being a cleared murder (as [Assoun] had been convicted of the crime), because Moore manually entered data so that the ViCLAS system would search both cleared and unsolved murders. Moore had a practice of doing this manual entry as he had reservations about the ViCLAS system. This led to Moore conducting further analysis into the Brenda Way murder. Moore also said that due to McGray not being consistent in his actions, he had to “spoon feed” information into the system.
Although McGray was very difficult to assess due to the many inconsistencies in the characteristics of his murders, Moore identified McGray as his top suspect in the Way murder for a long time. Moore was able to place McGray in the immediate area where Brenda Way worked and lived by analysing when McGray received, and where he cashed, his welfare cheques. This information placed McGray in very close proximity to the murder scene. Moore indicated surprise at the possibility of McGray committing a murder so close to where he lived given his previous history and that it must have been a mistake of some sort by McGray. Moore stated there was a great deal of information with which to assess McGray including psychiatric reports as well as letters Moore personally wrote to and received from McGray.
It appears that Moore was doing a more thorough job investigating other potential suspects in the Way murder in the years after Assoun’s conviction than Halifax police had done before Assoun’s conviction.
Incredibly, as Moore continued his investigation, he came to name someone else entirely — that is, not McGray and not Assoun — as the prime suspect in Way’s murder. We’ll get to that story at a later date, but for now, it appears that Mark Green, the Justice Department lawyer, sees McGray as Way’s likely killer.
Glen Assoun’s appeal
By 2004, Fred Fitzsimmons, the retired RCMP officer lawyer Jerome Kennedy had hired to look into Assoun’s case, was making significant progress.
Fitzsimmons came across references in police notes that the names of four potential suspects in the Way murder (including McGray) were run through the ViCLAS system. He gave that information to Kennedy, who began asking for the ViCLAS information from the police and crown prosecutors.
I haven’t yet unravelled the timeline — this is an ongoing investigation — so I don’t know where Kennedy’s request for ViVLAS information fits into the chain of events that I reported on Friday:
In 2004, Moore related his suspicions [about McGray] to his superiors at the RCMP — Ken Bradley and Dick Hutchings — but they told Moore “he was wasting his time” as “the matter had been decided by the Supreme Court and it was not worth pursuing.”
Despite this, Moore continued to investigate the Way murder because he thought it likely that Assoun had been wrongly convicted.
Moore went to RCMP inspector Andy Lathem, the head of the major crimes section of the RCMP. Moore told Lathem of his suspicion that McGray, not Assoun, killed Way. Lathem asked Moore to put together a timeline of events. Moore was intending to put together the timeline, but on return from a two-week vacation in March 2004, Moore found that he had been transferred out of the ViCLAS section. He was not given a reason for his transfer.
After he was transferred out of the ViCLAS section, all Moore’s work on the Way murder was erased from the ViCLAS system. Additionally, “hundreds of documents” Moore had kept in boxes, work sheets, and timelines went missing. The missing material includes information about other cases Moore had worked on.
No ViCLAS information was turned over to Kennedy, and Assoun lost his appeal.
Assoun, an innocent man, spent another eight years in prison.
What were the RCMP’s motives for apparently destroying the ViCLAS records and other evidence Moore had collected?
It seems obvious there was a desire to protect the reputation of the Halifax police. Turning over the ViCLAS information may have resulted in the overturning of Assoun’s conviction, and then led to a public understanding of the shoddiness of the original police investigation and how police improperly threatened and cajoled witnesses to provide false testimony against Assoun.
There’s an interesting document in the court files: a memo to the Operation Full Course file from Cst. D.L. Southern, a cold case investigator with the Halifax Integrated Major Crime Unit. The memo is not dated, but was written sometime in late October 2005 or early November 2005.
One person familiar with the case files told me Southern’s memo is a “cover your ass” memo. It read, “ISSUE: Possible disclosure in relation to the Brenda Way Homicide investigation that would be detrimental to this investigation.”
The memo went on to relate how Operation Full Course was begun in order to consider Andrew Johnson as the suspect in the cases of Andrea King, Kimberly McAndrew, and others. Operation Full Course was disbanded in 1999 because Johnson had been named a dangerous offender and so would never be free, but also because of “the operational needs for human resources from both departments” (that is, both the Halifax police and the Halifax detachment of the RCMP).
The memo continued:
During “Operation Full Course,” the homicide of Brenda Way, a prostitute found murdered on the 12th of November, 1995, in the Albro Lake are of Dartmouth, Nova Scotia, came to light. In the investigation, Glen Assoun was identified as a suspect, and in 1998 was charged for the murder of Brenda Way, and convicted of same in 1999. Although the investigation into the Brenda Way homicide came to light during “Operation Full Course,” it was separate from the operation and investigated by members of the Halifax Regional Police.
In 2004, an investigation was commenced by Jerome Kennedy, an attorney from St. John’s Newfoundland, in regards to Glen Assoun’s contention [that] he was wrongfully convicted.
In the course of this investigation Jerome Kennedy hired a private investigator, Fred Fitzsimmons, a retired member of the RCMP. In the course of their investigation, Fred Fitzsimmons became aware of Michael McGray and “Operation Full Course.” Sergeant Wayne Hurst of the Cold Case Section of the Halifax Integrated Major Crime Unit was assigned to liaise with the office of the Public Prosecution Service. The Public Prosecution Service requested information associated to Michael McGray, Glen Assoun, and/or Brenda Way in Operation Full Course. The only information to date provided to the Public Prosecution Service, is a verbal recount of the interview Constable Steve Maxwell had with Michael McGray. In this interview Michael McGray told Constable Maxwell that he knew Brenda Way, he didn’t kill her, and the police had the “right guy” for it.
In the memo, Southern said that Johnson had appealed his dangerous offender status, and if he won, he could be freed entirely from prison, back on the streets. Therefore, Operation Full Course was “resurrected” in October 2005 in order to prepare “an operational plan” should Johnson be released.
“Sergeant Hurst is continuing to liaise with the Public Prosecution Service in regards to Glen Assoun’s contention of a wrongful conviction,” wrote Southern, continuing:
A hearing is scheduled for the 21st of November, 2005, in the Supreme Court of Nova Scotia in regards to Glen Assoun’s contention of a wrongful conviction, with the apparent reasoning being is that Michael McGray is a convicted murderer, and he lived in the same area as Brenda Way during the time she was murdered, and could be responsible…
Even though it has not yet become an issue, there could be a foreseeable problem involving disclosure as Brenda Way is tied into “Operation Full Course.” It was documented in the investigation that Brenda Way knew Andrew Johnson, and he had used her services. Disclosure of this investigation or certain parts of it would most certainly jeopardize any future operations involving Andrew Johnson in regards to an undercover operation or other investigative techniques.
The memo makes no recommendation, but says the Unit should “be aware of” it.
This strikes me as extremely thin reasoning — Brenda Way knew Andrew Johnson, so therefore information about Michael McGray shouldn’t be given to Glen Assoun’s lawyer? Well, the evidence Moore had collected about McGray no longer existed in any event, as it had been destroyed the year before. So now the police had an excuse not to turn over the non-existing evidence to Assoun’s lawyer.
Dave Moore speaks
Dave Moore retired from the RCMP in 2010. On Friday, he sent me the following statement:
Glen Assoun “The Grinding Pursuit to Freedom”
July 12, 2019
First I would like to apologize to Mr. Glen Assoun, there are simply too few checks and balances in our system. Stinchcombe should also have been applied to Viclas and most certainly for the 2006 appeal application.
This case will change that forever.
The Dept of Justice instead of assisting in full disclosure did everything in its power to destroy, hide and conceal the truth.
Nova Scotia is a corrupt province. Rotten to the core.
The cruelty, dishonesty, corruption and continued cover-up make me want to vomit.
The RCMP, HRPD, Dept of Justice were complicit in a massive cover-up.
Those individuals in a position of power to do something failed miserably and should be ashamed of themselves and held accountable. They knew Right from Wrong but simply chose to go on the offensive and attack using every available malicious method short of assassination to attempt to discredit the messenger, me.
Even Associate Chief Justice O’Neil of the NS Family Court Division refused my subpoena to testify at “The Parliamentary Inquiry in Ottawa for the Wrongfully Accused,” citing that he would decide if and when I could testify and that his courtroom was more important.
When a Justice of the Supreme Court wilfully and knowingly breaks laws and is in contempt of the judicial procedures of our country, he puts the administration of justice into disrepute and undermines the legal foundation upon which this country is based.
The Criminal Justice System is badly broken particularly in Nova Scotia!
The oversight body “The Judicial Council of Canada in Ottawa” is a joke and a political white wash for the judicial misconduct of judges. It should be replaced with a proactive investigative committee and judges should be closely monitored. “They are not GOD although many think they are.”
The system is broken but that does not mean that we quit and walk away. Improvements by way of checks and balances come at a slow pace via trial and error. Let us hope the “Assoun Case” will eventually advance us towards an improved system. I am one man who wears the scars of battle over my entire body in an effort to better mankind.
A poem translated by Loren Eiseley in 1969 describes the belief that one person can make a difference.
…A man was walking on the beach one day and noticed a boy who was reaching down, picking up starfish and throwing them into the ocean. As he approached, he called out, “Hello! What are you doing young man?” The boy looked up and said, “I’m throwing starfish into the ocean.” “Why are you doing that”? asked the man. “The tide stranded them. If I don’t throw them in the water before the sun comes up, they’ll die” came the answer. “Surely you realize that there are miles of beach, and thousands of starfish. You’ll never throw them all back, there are too many. You can’t possibly make a difference.” The boy listened politely, then picked up another starfish. As he threw it back into the sea, he said, “It did for that one.”…
Glen Assoun was my starfish.
I’ve requested an interview with Moore and other people related to this story. There is more to come.
The Halifax Examiner is an advertising-free, subscriber-supported news site. Your subscription makes this work possible; please subscribe.
To contribute to the Halifax Examiner’s legal fund, please contact Iris.
Thank you Tim, for all the digging, writing and truth-telling you’ve done on the Assoun case. This huge miscarriage of justice imposed on Glen Assoun is reminiscent of other horrifying mistakes, and tunnel vision exhibited by the crown and the police when men such as Donald Marshall Jr, David Milgaard and Steven Truscott spent years in prison for murders they did not commit.