by Colin Wilson
The VICAP form lists the types of crime dealt with by the system, as follows:
Solved or unsolved homicides or attempts, especially those that involve an abduction; are apparently random, motiveless, or sexually oriented; or are known or suspected to be part of a series.
Missing person, where the circumstances indicate a strong possibility of foul play and the victim is still missing.
Unidentified dead bodies, where the manner of death is known or suspected to be homicide.
The report form also reminds investigators that ‘Cases where the offender has been arrested or identified should be submitted, so unsolved cases in the VICAP system can be linked to known offenders’. According to SSA McCrary, this reminder is of crucial importance. ‘The strength of VICAP – and inversely, its weakness – lies in getting these reports sent to us. The difficulty lies in convincing local investigators, who feel they have enough paperwork to do and enough forms to fill in as it is, that it is to their benefit to fill in one more. Because if these reports aren’t submitted, and no-one tells us about the crimes, we have no way of knowing these serial killers are out there.
‘Right now California is about to send us several thousand unsolved homicides, cases spread over the last ten years or so which as yet have not been entered in the VICAP system. It’s going to mean an awful headache for someone when all this hits us, but the fact remains – the more cases we get, the better job we can do. As a matter of fact attempts are being made to set up legislation, which would make it mandatory to put all unsolved homicides into the VICAP system.’
These crime analysis reports are entered into the master system via the BSIS computer, which stands in its own centre along the corridor. Because it needs controlled atmosphere and humidity to function at maximum efficiency, casual entry to the centre is barred by cypher and key locks. All the visitor sees through the protective screen is a red light, pulsing in the computer’s steel face like a great bloodshot eye as it ‘talks’ to Washington. Crime reports apart, ‘Old Red-Eye’ is also used for the BSIS Artificial Intelligence System (‘AI’), and other in-house tasks such as the PROFILER system. It also has an additional onward link with the National Crime Information Centre (NCIC) in Washington. The NCIC is linked in its turn, by telecommunications network, with all 17,200 police departments and other law enforcement agencies scattered throughout the United States: and entry to this system enables the NCAVC to request ‘off-line’ checks (for instance, when attempting to monitor the movements of transient serial offenders).
VICAP analyst Kenneth A. Hanfland works from an office close by the BSIS computer centre. As he describes his job this big, cheerful man from Oregon sounds oddly reminiscent of, say, an art expert called in to check on the authenticity of a painting; years of comparing and contrasting Template ‘matches’ with crime reports have given him rare assurance in assessing the modus operandi of specific types of serial killer. After a while the conversation turns to British serial killers: and VICAP analyst Hanfland and SSA McCrary both hear for the first time of Kenneth Erskine, alias The Stockwell Strangler. They learn that Erskine committed seven murders in fifteen weeks in 1986, targeting old age pensioners exclusively, the majority of whom were sexually assaulted. No further description of Erskine is given, other than to add that he was aged twenty-four and British. Hanfland nods, and asks: ‘He’s coloured, right?’ McCrary agrees – and they are right (see here).
The question is, how can any analyst know the offender’s race from those bare details? Senior analyst McCrary explains: ‘In the FBI we don’t go into the race area as such, except in so far as it applies to profiling. Obviously, when the police in any multiracial society are looking for a specific type of offender, it’s important for them to know if he’s black or white. As it happens, most serial killers are white. No-one knows why, but the statistics show it to be so.
‘Most violent crime is intra-racial, i.e. white on white, hispanic on hispanic, and so on. The exception is the black offender, who crosses the racial line more frequently than do other offenders. You told us just now that this strangler in London targeted old folks exclusively, and sexually assaulted most of his victims . . . and it’s a statistical fact that sexual assault of the elderly is the only sex crime that blacks commit more often than whites. Again, no-one knows why.’
VICAP analyst Hanfland’s next-door neighbour in the underground complex is social psychologist Dr Roland Reboussin. Dr Reboussin – he is a doctor of philosophy – is a bespectacled, middle-aged, civilian member of the Behavioural Science Computer Engineering Services Sub-unit: a boffin. His unique role in the operational wing lies in the administration of PROFILER – the first ‘expert system’ to profile serial offenders by computer.
Expert systems are part of that domain within the computer sciences known as ‘artificial intelligence’. Their function is to employ the computer in given areas, in a way which simulates the processes of human intelligence. To achieve this they require a ‘knowledge base’ – which may consist of up to five hundred rules – to guide the automated system towards its interpretative decisions in whichever field the expert system obtains. This knowledge base is made up of ‘artificial intelligence’ – i.e. expertise culled from the minds of human experts employed in a variety of areas: commercial, medical and most recently, criminal investigative analysis.
Dr Reboussin refers with genuine pride to the many successful uses of the expert system. ‘In the commercial world they are used to search for mineral deposits, to fly and navigate passenger-carrying aircraft, to land NASA’s shuttles after their journeys through space, and to serve a host of complex electronic systems . . . One famous example in medicine is Mycin, an expert system which diagnoses diseases of the blood.’ How does an expert system work? ‘Well, take Mycin. Each rule in the Mycin knowledge base is an “if-then” statement: for example, “If the patient’s temperature is over 100F, then an infection may be present”. The user, however, never sees the rules. Instead it is the expert system which puts the questions, and uses its knowledge base to diagnose the illness from the information supplied by the user.’
PROFILER employs the same technique in investigative analysis. The artificial information on which its knowledge base is built is similarly a compound of human skills, culled mostly from the NCAVC’s related investigative disciplines. Its initial rules were developed by observing the complete investigative analysis process (based on field reports, crime scene evidence, pathology, forensic science, victimology, etc.), performed in a group setting. Individual expert human analysts further checked and adjusted every rule until the first 150 were ready for the prototype PROFILER trials.
In a joint paper entitled Expert Systems for Law Enforcement published in August 1989, Dr Reboussin and his fellow author Mr Jerry Cameron, Chief of Police at Fernandina Beach in Florida, explained how the basic system worked. ‘Information about the victim and the crime scene is essential to the analysis process.’ After stressing that the analysis is designed to profile the type of offender responsible, rather than the individual, they continued: ‘With detailed information as to age, sex, occupation and daily habits of the victim, the autopsy report, and a specific description of the crime scene, the behaviour of both the victim and the offender during the crime can be reconstructed. The result of this analysis is a description of the (type of) person who committed the crime, which includes physical characteristics (age, sex, and race), behavioural characteristics (whether the offender lives near the scene of the crime, lives alone, or is unemployed), and personality traits and characteristics (the nature of relationships with women, or volatile temper). An intermediate trait would be the nature of the offender’s relationship with the victim.’
Now two years old and fully operational, PROFILER provides an accurate computerised profile of the type of offender responsible for serial violent crime. The depth of the profile is limited, as always, by the number of rules programmed into the system; already these have risen to some 270, and t
his knowledge base is continuously being expanded (see here). Although it is unable as yet to match the human analyst in all areas, PROFILER has already become an efficient working robot member of the ‘A Team’ at Quantico.
Each of the ten senior analysts has his own computer, which he uses to bring up PROFILER on-screen to compare the various aspects of both human and automated profiles. Should any variance emerge both are re-examined to discover why, so that the difference may be resolved. Dr Reboussin and Police Chief Cameron emphasise that the computerised programme ‘will never replace skilled human investigative analysts, nor is it intended to do so . . . Rather, the system will function as an analyst’s assistant or consultant in several ways’.
Laymen pose the question: what if there should be a conflict between man and machine? The professionals do not see it that way. McCrary explains: ‘The human analyst has the last word in profiling: it is his responsibility. You have to say “I’m doing the profile, tnd this is what it is – A, B, C, D and E, right through”. Having said that, obviously there may be areas where queries arise. Take a case in which the offender displays mixed characteristics, a frequent occurrence. The human analysis may read simply “mostly disorganised”, whereas the computer says “Point 75 disorganised, point 25 organised”. Okay, it’s confirming in effect what you’ve said, but has made the point definitively. On other occasions there may be an area where you think, “Why doesn’t the computer see it this way as strongly as we do?” This time you re-examine both profiles and you realise, “Ah! I’d forgotten that”, and adjust accordingly. That, of course, is the great strength of the computer. It never forgets anything.
‘Given that, what emerges ultimately is the sole responsibility of the human analyst. We don’t actually ask the computer if we’re right, for the simple reason that it doesn’t have the final say. Now, I feel comfortable profiling. With experience, and enough successes, I guess you become confident in whatever you’re doing, that’s only natural. But I hope never to feel overconfident: I’m aware that I neither know all there is to know about profiling, nor do I know all there is to learn about profiling. Equally, the computer doesn’t have the benefit of “experience” – whereas we humans do. So I look on PROFILER as one more tool to use when completing what has to be a carefully thought out, fine-tuning process.’
At the NCAVC there is total belief in the fine-tuning process called CIAP, the Criminal Investigative Analysis Programme, which has successfully wedded human and computerised techniques. ‘By the end of this year (1989), we will expect to have dealt with more than seven hundred cases’, says McCrary. ‘That doesn’t mean seven hundred serial murder cases, and in any event not all serial murder cases require profiling. For example, if the law enforcement agency already have a suspect, they don’t need a profile. But they might still ask us for a “personality assessment”.’ (A situation in which, without knowing the identity of the suspect, a profiler uses the behavioural analysis technique to answer the question – is the suspect capable of such a crime?) The BSIS wing at Quantico furnished such an assessment in the case of Robert Hansen, the Alaskan big-game hunter and serial killer who stalked his human victims gun in hand as if they were wild animals (see here).
‘We may also be asked for specific investigative strategies, and/or interview techniques,’ he said. ‘At other times we may give on-site assistance during the investigation. Some cases we deal with will be over quickly, others will prove more difficult. All of it is time-consuming, a heavy workload at times. One analyst may find himself involved in fifteen or even twenty cases – and that’s pressure. But we’re in the young stages of this whole analysis process as yet; and I firmly believe that we’re headed in the right direction, and that this is the way to do it.’
We catch a glimpse of the pressure that faces everyone in the ‘operational wing’ the moment we step inside McCrary’s office. A mass of paperwork relating to one more homicide – police report, photographic evidence, pathologist’s report, forensic details, and victimology – waits on his desk alongside his computer, telephone, notepad, legal and investigative reference works, et al. This new investigation centres on a torso murder, one in which the nude, headless, legless body of an adult white male was discovered – wrapped in a tarpaulin and dumped at a lay-by for truck drivers – on Route 8, south of Torrington in upper Connecticut. The police investigation has encountered problems, and they have submitted a VICAP crime report requesting a profile; a task assigned to SSA McCrary. After studying the relevant documents, discussing every feature of the case at daily group conference, and consulting with various NCAVC resident advisory specialists, McCrary furnished the police with a profile. The ‘Template’ is still awaited from the master VICAP computer at FBI headquarters in Washington.
Because this is an ongoing case, no details of the profile can be disclosed. We know, however, that it will have listed some twenty or more components, including the age, sex, race, marital status (or equivalent), IQ, school and college grades, and the ‘rearing environment’ of the type of offender responsible; plus personality traits, life-style, demeanour, appearance and grooming, the kind of job he holds and past employment history, his behaviour during the murder, from Phase One (his meeting with the targeted victim) to Phase Four (disposal of the body, after dismemberment and mutilation); where the offender lives in relation to the body disposal site, his motive for the murder – everything bar his name.
Even so, the early signs indicate that this may prove to be one of the ‘more difficult’ homicides to solve. Connecticut, where the body was dumped, covers more than 5,000 square miles in area: not nearly as big as Texas, say, or California (or most states), but still the size of Northern Ireland. And if in fact Connecticut turns out to be where the unknown killer lives – one of the profiling decisions McCrary has to make – to track him down is never going to be easy. In this case the task is further complicated by the fact that the murdered man (whom the police have identified as a known homosexual, aged twenty-six, with convictions for both prostitution and burglary) was also transient, and on both counts a ‘high risk’ victim – a homosexual hitchhiker. He set off on his last hike from Nevada – almost the entire breadth of the United States away – nine days before his body was found, hitching lifts mainly from truck drivers. At what point along the route he was picked up, propositioned, attacked and murdered is uncertain: a vital profiling component, which will have called for a very finely-tuned decision by SSA McCrary.
Every twist and turn in this torso murder analysis would seem to provide seemingly obvious behavioural clues, yet equally evident contradictions. For example: the victim’s head and legs are both missing, while the torso has been mutilated by emasculation and removal of the nipples. Even to the untrained observer, it would seem logical to assume that this combined dismemberment and mutilation must have taken hours to perform. In turn this would appear to be a clear indication that the murder was pre-meditated and methodically thought-through, even though the victim was one of opportunity. To execute such a crime successfully would necessitate a secure base, where the offender knows he will be able to work undisturbed during those hours of surgery; and he will need more time afterwards to wash away all traces of his sawbones task – and to hide the incriminating body-parts – before setting out to dispose of the torso.
The place best suited for such a lengthy and potentially risky chore would obviously be the killer’s home and/or workplace. As we have shown, one of the profiler’s first tasks is to calculate where the offender lives in relation to the scene of the crime. In this case, only the body disposal site is known thus far: in itself a not uncommon feature in homicide investigations. But what is doubly baffling (in behavioural terms) is why the murderer first ‘de-personalised’ his victim (by severing and retaining the head), only to dump the body in a truckers’ lay-by on a main road, thus ensuring it would be found with the least possible delay. Furthermore, why compound the felony, literally, by deliberately leaving the victim’s arms (
and more importantly, his hands) attached to the torso – as if to invite identification by fingerprinting?
There are a number of possible reasons why a lust killer – which is clearly what this offender is – may remove and retain his victim’s head. One might be that the murderer is a ‘souvenir’ collector, a psychopath who stores selected body parts to re-live at will the violent fantasy which first inspired the crime. But that would not explain the apparent behavioural contradiction in first ‘de-personalising’ the victim, only to ensure his identification at the earliest possible moment. By the same token, no matter why he beheaded his victim it cannot have been to thwart identification. Nor is this apparent contradiction the product of a confused mind. The modus operandi shows this to be a meticulously planned murder, patently the work of an ‘organised’ offender – which points to a causative decision, and one more riddle for the analyst to fathom.
At this moment, the telephone rings to involve McCrary in yet another serial murder enquiry – thus rendering all questions on the torso case superfluous. (The ten senior analysts in the BSIS wing are each allotted a separate national area of responsibility. In normal circumstances, all serial cases reported in that area are automatically assigned to the analyst concerned; while the initial report of each new case will come in one of two ways – mostly from VICAP crime returns, but sometimes through direct calls to the NCAVC.) Today’s direct call comes from another state in SSA McCrary’s ‘area of responsibility’ – Rhode Island, Connecticut’s eastern neighbour.
