Post No.: 0317
Offender or criminal profiling is about determining the potential characteristics of an offender by examining the characteristics of a crime scene, the victim(s) and the crime itself. This is based on patterns found from past cases, similar recent crimes and from wider demographic data (including sometimes race). This information can be used to predict the future moves of serial criminals in order to catch them and prevent further offences being committed by them.
Racial or ethnic profiling (or profiling by another demographic measure such as religion or nationality) is used to target potential future terrorists or criminals based on their race or ethnicity. This may increase the odds of finding them but it is frequently unfair and unreliable. It can result in a lot of false positives (innocent people being suspected) and false negatives (guilty people being overlooked). Offender and racial profiling are related – the former attempts to determine the characteristics of criminals based on already committed crimes whilst the latter attempts to predict people’s potential to commit future crimes based on their characteristics. Both require assumptions to be made about individuals based on the generalised (stereotyped) characteristics of the group(s) they apparently or supposedly belong to.
The practice of racial profiling can create a self-fulfilling prophecy – if certain groups are routinely targeted and discriminated against by the police even when the specific individuals have not done anything wrong, then these individuals will grow up with a distrust and hatred towards the authorities, which may in turn lead to a greater temptation or push towards criminality. Racial profiling may contribute to certain groups seeing authority figures as accusing them basically from birth, hence some might feel that they might as well join a life of crime if there are no other, better, opportunities for them. This links with what was written in Post No.: 0310 about how holding fears and suspicions against innocent people can cause divisions. In terms of justice, evidence of wrongdoing is a billion times more objective than the pre-emption or suspicion of wrongdoing.
Real criminals could also possibly directly anticipate and circumvent expectations in an act of ‘counter-warfare’ too hence make such profiling strategies self-defeating (e.g. if authorities tend to mainly ‘stop and search/frisk’ Middle-Eastern-looking men then genuine terrorist groups will learn to groom members from other ethnicities, or even use children or women instead).
The cross-race effect also affects stop and searches/frisks when the police find it harder to distinguish between individuals from an ethnic minority group because the police are less familiar with them and think ‘they all look the same’. A broader group of people will therefore possibly be believed to ‘match the description’ of a suspect when it concerns an ethnic minority suspect.
In the longer term and bigger picture, racial profiling will also perpetuate crude ethnic stereotypes that the media spreads and the wider society constantly hears about, which makes it hard for members of stereotyped and discriminated groups to get legitimate jobs, for instance, because fewer employers will want to employ them due to behaving risk-aversely; and so these prejudiced citizens again will (more likely than members of other groups) be pushed into a life of crime to survive.
This could arguably be a ‘chicken and chicken egg’ situation though – did the criminals make a stereotype for their own group or does the stereotype influence who will more likely become criminals? It’s likely bi-directional. But even if the former is partly true, ‘a pattern of some bad people from a particular group’ doesn’t mean ‘all people from that group are bad’ – and really, every single ethnic group has some good people and some bad people in them.
Programmes that attempt to identify and ‘prevent’ individuals or communities who are considered at risk of radicalising into terrorists counterproductively label and vilify such people and their communities, which isolates them and could lead to more distrust against authorities and motivate radicalisation. Well imagine if you were minding your own business then was suddenly and specifically brought up as being at risk of becoming a terrorist – how would you feel? And to expect these communities to ‘sort their own people out’ assumes that they all know each other personally – just like if we assumed that every person from New York should know every other person from New York personally(!)
Geoprofiling or geographical profiling concerns offender profiling. This is about profiling the geographic locations of crimes committed by serial criminals, such as serial killers or rapists, and in turn where they may be based or where they live. In general, they tend to fall into either a ‘commuter’ pattern (travelling to a consistent area away from their home to commit their crimes) or ‘marauder’ pattern (travelling in any direction away from their home at roughly the same psychological distance away each time to commit their crimes). It’s based on the pattern that most serial criminals don’t commit crimes too close to where they’re based yet don’t go out too much further than the distance they feel safe enough to not get identified. This can help narrow down a search, but again these patterns aren’t always reliable. And once more, a meticulous criminal could learn about these patterns and purposely do something different to fool the authorities.
All of these forms of profiling may help narrow down the list of potential suspects or a potential base location for a suspect, and serial criminals may leave a distinct individual signature that links several crimes together, but in most cases they aren’t useful in identifying precisely who committed a crime.
For this, we, as always, need to find hard evidence. Focusing on the strength of the evidence when trying to decide if someone is the perpetrator or not, is being deceptive or not, and is ultimately guilty or not, is still by far the primary and best approach for obtaining any truth and delivering fair furry justice. Woof!
Artificial intelligence can work out patterns of crimes in an area and anticipate where to best deploy police resources. Algorithms are increasingly being used to predict crimes and even determine sentencing durations and whether parole should be granted. They can improve upon purely human decisions but they are not free from bias. Rubbish in means rubbish out, biased in means biased out – not just concerning the data that’s fed into teaching these systems but in deciding which variables are deemed to matter in the first place (e.g. people’s age, gender, race, religion, sexuality, where they live, what school they went to, their job, their musical tastes, tattoos and piercings…) i.e. subjective human decisions, and therefore biases, are not going to be completely obviated by using algorithms. And the very minimum bias is that they will assume that past trends always predict the future.
In general, although still evolving, all humans are the same human animal species, and when aggregated, humans do fall into broad, predictable patterns of psychological attitudes and behaviours based on their personalities, mental status and environmental circumstances. So law enforcement and intelligence organisations – as well as private corporations such as social media giants and supermarkets – can learn about and, generally when looking at a population as a whole, predict people’s movements, beliefs, behaviours, fears, desires and the like if they have enough personal data about them (such as their ‘likes’ or purchasing habits). They therefore have the ability to quite intimately shape or nudge people’s choices too. (So note that ‘nudging’ is not just some paternal government tool to get people to choose certain options over others – private for-profit firms have been doing it for decades to shape consumer behaviours, and not usually for the benefit of the consumer but for the firm too e.g. clickwrap agreements that are lengthy and full of technical jargon to discourage careful reading and have large ‘I agree’ buttons and small ‘decline’ buttons, one-click checkouts, placing the most profitable and/or impulse-buy items on the gondola sections or near the checkouts in supermarkets, exiting through the gift shop, etc..)
But these patterns are only present in general i.e. with a large enough sample or target market there’ll be a lot of people who’ll conform to the predicted psychological and behavioural patterns but not everyone will. So this can either be interpreted to mean that profiling and targeting is incredibly useful overall, or pernicious overall, when potential future criminals are identified or when adverts are specifically targeted at certain groups to influence their purchases or votes; where companies or politically-motivated groups don’t really care specifically who it works on as long as it works on enough people, which means that even anonymous data is useful in shaping profits and political outcomes. (This has been dubbed as ‘surveillance capitalism’ – making money from what data companies collect about us. This data is used to make predictions about us and to shape our behaviours e.g. Pokémon GO and similar games herd masses of people towards places where the company wants a lot of footfall – and players will perform this herding behaviour without knowing it too.)
Because of this hardly-perfect predictive reliability, it’s less reliable to use a generalised psychological profile to try to predict the attitudes, moves and so forth of a specifically watched individual, such as a specific criminal suspect; although it could still act as a guide when no better information is available about a specific individual – as long as it’s understood to be just that to help narrow the search and that this guide is always subordinate to more concrete and specific evidence for the present case at hand. Another important thing to bear in mind is that situational factors (e.g. the poverty and desperation of a person to steal) frequently play a larger role than dispositional factors (e.g. the inherent risk-appetite of a person to steal).
So psychological profiling and prediction based on personality and personal data is more useful if we don’t care who we correctly predict as long as enough people’s attitudes and behaviours are correctly predicted, but less useful or reliable when trying to predict a specific individual based on his/her personality and personal data (at least with current personality tests, understanding of people, data and algorithms).
In summary, it could boil down to this – it’s arguably rational to use potential suspect profiling from a consequentialist perspective if the overall benefits of it outweigh the costs to society. So how many discriminated innocent people for how many caught or prevented guilty or would-be-guilty people would make its use acceptable? Would even just one incorrectly suspected innocent person put under secret surveillance or stopped and searched/frisked against his/her liberty be too many to make such profiling strategies unacceptable?
It could be a case of short-term/little gains but long-term/big costs? It’s kind of like if we pre-emptively locked every adolescent or adult male up in prison then this would reduce crime, but the overall cost to society would be worse. (This is one of the debates regarding mass surveillance – it could be reducing crime, but at what cost?) And profiling does not merely predict whoever likely is, but can also possibly shape whoever will more likely become, a criminal, because certain ethnic or religious groups in society have become crudely discriminated against – which should not be what fighting crime is about!
Woof. What do you think? Please share your views by replying to the tweet linked in the Twitter comment button below.