Home Affairs Committee
Oral evidence: The Macpherson Report: Twenty Years On, HC 1829
Wednesday 12 June 2019
Ordered by the House of Commons to be published on 12 June 2019.
Members present: Yvette Cooper (Chair); Janet Daby; Kate Green; Tim Loughton; Toby Perkins.
Witnesses
I: Oliver Feeley-Sprague, Military, Security and Police Program Director, Amnesty International UK.
Written evidence from witnesses:
Witness: Oliver Feeley-Sprague.
Q145 Chair: Can I welcome everyone to this session of the Home Affairs Select Committee’s evidence into the Macpherson report 20 years on? Can I welcome our witness today and ask you to introduce yourself?
Oliver Feeley-Sprague: Hello, thank you. My name is Oliver Feeley-Sprague and I am the Programme Director for Amnesty UK’s work on policing and security issues. I was part of the team that put together our research and publications and follow-up work on the Metropolitan gangs matrix.
Q146 Chair: Thank you. Mr Feeley-Sprague, could you tell us your reflections 20 years on from the Macpherson report? What would you see as being both its achievements, but also concerns that you would have from the recommendations 20 years on?
Oliver Feeley-Sprague: Before I came to give evidence to this Committee this morning, I was refreshing myself about the kind of things that Macpherson was saying in his summaries and his conclusions and the issues that were raised. What I would say is that from our work on the gangs matrix, the same sorts of issues that were being raised 20 years ago were exactly the same sorts of issues that we encountered when we were compiling our research into this particular policing tool.
As a very short summary, basically if you talk to local communities, especially those coming from a BAME background, and especially those coming from the black and African-Caribbean community, they would feel that tools like the gangs matrix are victimising them and are clearly discriminatory. They have little faith that these tools are policing them fairly and treating them appropriately.
Another thing that I would point out is if you go slightly wider and you look at some of the systematic failings that were identified during the Macpherson inquiry over the way that the police may have managed and conducted their investigations. Whether we are talking about standards and veracity of evidence-gathering, due process and rigour that have been applied to the various processes, the failure to adequately distinguish between the victims and the perpetrators of crime, you see all those sorts of issues that have been at play with the running and the operation and the data use within the gangs matrix, at least in terms of deficiencies. So 20 years on, the same sorts of issues, to my mind, are prevalent in policing tools such as the matrix.
I guess we don’t just work on the matrix. My work on policing touches other areas of policing and while it may not be quite as extreme as expressed within the matrix, the same sorts of issues are apparent in those tools too. We know that 80% of people on the matrix are currently young black men under the age of 24, but if we look more widely at things like the newly-established police use of force reporting tool—which I acknowledge is only a couple of years into the dataset for this, so it is a relatively new area—you are seeing trends of disproportionality against young black men in the reporting of the use of force dataset. Something that we do a lot of work on, for example, conductive energy weapons, our work on tasers, the Government’s own figures show consistently that a higher proportion of taser use is directed against people from ethnic backgrounds, particularly black men. I think the official figures put it at 13% of all taser use is against black men, despite the fact that they only make up 3.3% of the population. I guess my answer to your question is there is not much progress. The same sorts of issues that were identified in Macpherson I am seeing in my work today.
Q147 Chair: Could you just tell us a bit more about the gangs matrix and what is it about it or the way that it operates that is causing you most concern?
Oliver Feeley-Sprague: For people that don’t know what the gangs matrix is, it is essentially a database that identifies individuals on the grounds of whether they are suspected of being affiliated to a gang. It ranks them into three main categories, so there is a red category for the most serious offenders, there is an amber category and there is a green category, which represents the lowest risk. On this database, roughly about 5% of people are in the high risk category of red and the vast majority of the people are in the green category. When I say “vast majority” I am talking 65% of people are in this green category.
What exactly it is, essentially it sounds like quite a high-tech policing tool, the word “matrix” makes sound like a quite a high-tech tool, but from our understanding, it is probably easier to categorise it as more of an Excel spreadsheet type system that is run by each London borough. The 32 London boroughs will each have its own version of the gangs matrix running on its system. It is not one matrix, it is 32 different ones. One of our findings—and those findings were endorsed and essentially are backed up by both the ICO, the Information Commissioner’s Office investigation and review, and the Mayor’s Office for Policing and Crime, MOPAC’s review and report into the matrix—confirmed that one of the key problems was a wide discrepancy between the way different boroughs managed the data and shared the data on the version of the matrix that they were operating.
One of other things that I think was very striking, and one of the criticisms that came out very strongly is that the matrix itself fails to differentiate properly between the victims of gang violence and the perpetrators. There is evidence to suggest that a large number of people on the database are actually the victims of gang violence, not necessarily the perpetrators. When we look at it as a risk management tool, the matrix gives you a harm score of how likely it is you are to be caught up in serious violence. It fluctuates, because the matrix itself, the numbers on it fluctuate slightly, but on average 40% of the people on the matrix who are in the green category—so the 65% that are the lowest-risk individuals—have a harm score of zero, so there is no evidence to suggest that within a two-year period any of those individuals have been caught up in serious organised violence. That is 40% of your database that is made up of people that are incredibly low risk. What is striking is both the ICO and the MOPAC review put a spotlight on that. I am paraphrasing what they say, but they are basically saying that the green category needs a serious overhaul and looking at and lots of people, especially those with zero harm scores, need to be removed from it.
Q148 Kate Green: Just pursuing that line of discussion further, if I may, have the police offered any explanation or any response to this criticism of the substantial proportion of people not only in the green category, but with a harm score of zero as well?
Oliver Feeley-Sprague: What is interesting is that when we talking to police colleagues, what I will say here—which is interesting in itself as well—is one of the reasons why Amnesty ended up researching and publishing our report on the gangs matrix is essentially because of people in the police force that we spoke to about another project. We were doing a project looking at the role of databases and algorithms and predictive policing and technology within policing. We were scoping out whether or not these systems were useful, beneficial and so on.
As part of that, as we always do, we went to police contacts and we asked them about this kind of technology. A number of those people said, “No, what you should be looking at is the gangs matrix, because we think it is a problematic tool”. So there are voices within the police, especially those working at local level in terms of community engagement, in terms of looking at areas of serious violence, who believe the tool is counterproductive. It is not just Amnesty that says this, there is a considerable body of opinion within the police that questions the utility of this tool.
Q149 Kate Green: If the tool were reformed so that you did not have this apparently very highly disproportionate representation of people who are both in the green category and with a zero score for risk of harm, if it were reformed to be a more nuanced tool for identifying people at risk of becoming involved in gang activity and gang violence, what would Amnesty’s attitude be to that? Would you welcome it? Would you have concerns about it?
Oliver Feeley-Sprague: We are welcoming of the level of engagement that we have had with the police in talking about the matrix. We don’t always agree—you would not expect us to always agree—but we do have a very mature and good relationship with police colleagues and we talk openly about these things. Our recommendation is we do believe the matrix is a discriminatory tool. We do not believe it is fit for the purpose that it was intended and we do believe that there are serious concerns around the human rights safeguards, including the management and retention and protection of data. At the moment we think the tool is basically not compatible with those human rights and data processing standards. What we are saying is that the tool itself needs a very radical overhaul. If it can’t be brought into line with those standards, then it should be disbanded. That is our position, basically. The onus is on, “Can you bring it into line with what we think the standards should be around due processing?” and so on. If so, it may well be a tool that you can develop.
What is important is it is not just us now, both the ICO and the MOPAC review concluded the same thing. The MOPAC review says that this tool needs a radical overhaul and we need to have specific focus on the large number of people who are green and who have a harm score of zero and they should probably be removed. The ICO said essentially one of the key problems with the tool is that there is no process to properly remove people from the tool in the first place. As far as I am aware, the ICO enforcement notice is a legal obligation on the police. They cannot ignore the findings of an enforcement notice. The limit of the six-month period they had to put the changes into effect expired last month, so as far as I am aware, the police now should have complied with the terms of that notice.
Unfortunately what I can’t tell you today is whether or not they have, because although we are involved in many of the kind of stakeholder meetings and review processes that are ongoing, we have yet to hear from the police what exactly they have done in light of the findings from the ICO and the MOPAC report. I wonder whether that is a role for this Committee, to ask for a situation update or a progress report, because a statutory enforcement notice is quite an important thing.
Q150 Kate Green: To go back to what you were saying a few moments ago, if this tool or another tool could be developed that was able to offer the reassurances in relation to the proper use of data, the proportionate way in which that data was gathered and used—and in the light of evidence that Baroness Doreen Lawrence gave to our Committee some months ago, that the right way to go in relation to responding to some of the challenges we still face 20 years after Macpherson is an intelligence-led approach—would you see the potential for a risk-based, harm-based tool that could assist in that kind of intelligence-led approach?
Oliver Feeley-Sprague: The answer to that question is absolutely, yes. I do not want to give the impression in any way that Amnesty International thinks that the police do not have a public duty and a responsibility to keep everybody safe and having the appropriate intelligence-led tools and risk assessment tools to do that job. It is just that we do not think the matrix serves that purpose, but that is the main point.
I would say also that quite often one of the problems is the ill-defined and amorphous way that the gang term is placed and labelled on certain individuals. You hear the word “gang” in all sorts of contexts. There is already a very sophisticated police database that looks at gang violence in terms of serious organised crime. When we are talking about issues of county lines drug trafficking and those sorts of issues, which are very prevalent in my own region of Northamptonshire, these are serious concerns of organised crime infiltrating local areas. There is drugs, there is people trafficking, there is all of those things. Those are very, very serious things.
There is already a tool called the organised crime group mapping database, which is where these sorts of serious crime networks and individuals are already identified and flagged with the appropriate authorities. To say there is not an appropriate risk-based tool and that the police don’t have that tool is not true, because they do. In the context of gang work, in terms of county lines crime, the tool that they are most likely to be using is the organised crime group mapping database.
Q151 Kate Green: Would there be any value in a tool that identified not just people who are at risk of causing harm—as you have said, this matrix already muddles up perpetrators and victims—but a tool that identifies people who might be more prone to become victims of gang activity?
Oliver Feeley-Sprague: I will answer that by saying again that one of the criticisms that was clear from our research and that of others, but also the ICO investigation, was the purpose of the matrix was not clearly understood by anyone, including some of the people operating it, but very clearly the communities that the tool was designed to help protect. There is clearly a job to explain the purpose of the matrix a lot better. It certainly is the case that the police have argued that the tool is a preventative tool as well as a criminal enforcement tool, so the idea is that, as you say, people who are identified as being at risk of violence are somehow given the opportunity to steer away from that and choose a different path.
The problem is that we have heard virtually nothing from any of the research or the case studies or the people that we have talked to that suggests that the tool is being used in that way. What is much more likely to happen is that the tool is used as a kind of enforcement. If you are looking at a metaphor of a carrot and a stick, it is much more of a stick-based tool than a carrot-based tool. When you are mixing perpetrators and victims and when you are not clear on, for example, the rules by which you are sharing data with a wide variety of intervention services—education, local authorities, housing, jobs, schools, Probation Service, youth offending teams, all of that—when there is no clarity, you are going to end up in a situation where an awful lot of people are on a database with information shared about them, where the implications of being on that database are that you are treated as a risky person, not an at-risk person. That is very clear from the ICO findings, as well as our research.
To answer your question, yes, obviously we hear a lot about the kind of public health approach to tackling serious organised crime and that is obviously something that we need to explore. We need to be looking at what is happening in places like Glasgow, which have applied that approach and have seen dramatic reductions in armed violence. But we have seen no evidence to suggest that the gangs matrix tool has been used in that way.
Q152 Kate Green: One last very quick question from me. You said there is no clear route for people to get off the matrix. Do people know that they are on it or how can they find out?
Oliver Feeley-Sprague: There is no requirement for individuals to be told basically that they are on it. The reason for that, if you were to ask police colleagues, they would essentially say, “The purpose of running this database is the prevention and detection of crime, so obviously we can apply the exemptions within the data protection rules to say we should not obviously tell people that they are being watched and that they are on the list”. But it is with basically scant regard to any due process and rules. This was one of the scary things, that there were no clear—hopefully there will be now, because the ICO mandated this to happen—written agreements on how data would be used and shared with third parties.
Most of our experience in terms of talking to people that have been on the matrix is that they have found out that they are on the matrix because somebody else has told them. We gave an example of a young man called Paul. Obviously that is not his real name, but we referred to him as Paul. He found out that he was on the matrix; essentially a local authority council worker told him. Obviously his data had been shared with the local authority council worker. What was extraordinary about this case was that this guy, Paul, was an anti-violence ambassador, so he was working in the community to prevent armed violence. He was well-respected and known, he was involved in meetings and he may well have even given evidence to these sorts of Committees.
He was staggered to find that he was on this tool and he had no idea why. The only thing on his record in any way was when he was 12 years old there were some sort of marks against his name for some driving convictions, quite minor offences, but he thinks the most likely reason for why he was included on this database was essentially his association, so where he lived, who his friends were and where he grew up. That again was another one of the striking criticisms, that the gang flag was applied to people on the basis of who they knew, rather than what they were doing as individuals.
He tried to get himself off the gangs matrix and he wrote a subject access request to the police to say, “Please give the details” and so on. The police basically told him, “No, because we are exempt from those requirements, so we are not going to confirm or deny that you are on this tool”. But he was told in 2013 that he was taken off it. It then transpired in 2018, a couple of years ago, that he was in fact still on it, he had never been removed. This is one of many, many similar examples.
If you go back to the ICO enforcement notice, what it concludes from looking at the way that the gangs matrix is operated at borough level, by going in and seeing it and talking to the people processing the data, is there is no way of getting off the database and there are no clear rules. Quite often what would happen was that if you were removed from the main matrix, you would be put on a kind of separate informal list of people that were of concern anyway. It is this idea that you can never really get off this database, but there are no clear rules to say how that should be done anyway.
It is one of the clear findings of the two subsequent reviews and I guess the question to ask the police now is, “Have you developed clear criteria for how people are going to be removed? How is that applied and how many are being removed, given that the two main reviews and investigations into your database have essentially said the green people, so 65%, and especially those with a harm score of zero, 40%, there is a case for why they should be on this in the first place?” What we would hope to see would be a dramatic removal of names on the green category.
Q153 Toby Perkins: Oliver, you said that there is already the serious organised crime database, a sophisticated database, and that in terms of the gangs matrix you do see some value in potentially identifying those people who maybe are not the sort of serious organised crime level, but maybe are at risk of becoming a member of a gang, or people that are on the periphery of gang-related activity potentially about whom there is no intelligence at this stage, but it is known that they are involved in those circles. I am describing it as a statement, but it is really a question: you do see some potential value in identifying some of those people, but your greatest concern is about the way that this evidence is being used, rather than the point of the police having this information at an early stage than the point at which they know someone is involved in serious organised crime?
Oliver Feeley-Sprague: There are two parts to your question here. One part is about the standards of evidence that are required to identify whether you should be on this risk-based tool or not. That is one set of questions that I have some concerns about.
The other one of course to remember is that it is a risk management tool and it does work on a traffic light system, so you do have within it green, amber and red. Obviously the amber and the red categories are the people for which the most concern is generated, and specifically the red group, who are identified as people that are not at risk but are risky. If you talk to our colleagues in the community, we did not just write this report, we worked incredibly closely with local community groups and partners—and I know they can’t be here today, but StopWatch—and we could not have done this work without working with those individuals. Nobody is going to say that people that are in the red category, if they are on there through due process and that the evidence is sufficient, are not dangerous people and that the police have a duty to protect us from that. There is a clear case for us to be looking at the red and the amber group. We need to have tools that prevent violence from these individuals, because they have already been identified as risky.
The second part of your question is the standards of evidence that are required to be on the database in the first place. This is where again we have concerns, but not just us. As I said, the two subsequent investigations found deficiencies here as well. There is no standardised way of basically corroborating or confirming intelligence and evidence. Across the 32 London boroughs who were operating their own version of this spreadsheet matrix document, there were wholly different ways of interpreting and analysing things like social media data, looking at YouTube videos or Facebook profiles or those kind of things. That to me—and I think the ICO flagged this too, as well as MOPAC—suggests that the rigour is deficient, by just trawling social media and using that as an evidence base, so we need very clear rules on how things like social media evidence-gathering is used.
The other thing that came out quite strongly in our research, and this comes to the level of victims versus perpetrators, is that quite often if you were recorded in a police notebook, say there was a record of a stop and search incident and your name was part of the notes that were gathered as part of stop and search, say you happened to be in a car that was stopped and searched, the fact that your name appeared on those notes appeared to us to be in some cases grounds for being put on the matrix. Again, we would question the rigour of the methodology that suggests that that is a form of evidence.
In short, to answer your question, yes, high-risk people obviously we need protecting from and the police have a duty to protect us and to do all they can to prevent those sorts of issues, but the data to identify those people has to be robust and they have to be scrutinised and they have to be accountable. Going back to the original question about 20 years after the Macpherson inquiry, I think that was one of the lessons from the Macpherson inquiry, that the police did not have enough rigour in the way that they used their process in evidence-gathering standards. It is exactly the same thing here with the matrix. Does that answer the question?
Q154 Toby Perkins: I think so. I totally understand what you are saying about the most serious criminals, that people need protecting from them. I think everyone would sympathise with what you have had to say about the level of rigour required. I guess the question about whether this might be useful, as opposed to an addition to the other more sophisticated database that you were referring to previously, is in terms of identifying and trying to keep information about those people for whom there may be a belief or the start of a belief that they are involved in something more serious, but as of yet the evidence is not there. Obviously that would not include people identified as having been stopped and searched and nothing being found and it would not include a community worker who was working alongside people in gangs as part of their work in the community, not at all, but it might be useful in that regard.
I wonder if you could say a little more about what the consequences are for people who end up on the matrix, because while many people would have an expectation that the police would have developing levels of intelligence about gang members, even those who they had not yet identified as being involved at a serious level, the idea that that gets given then to a council worker or gets spread much more widely than the police I think would be alarming for people. Can you just tell us a bit about what the consequences are individually for people when they find themselves on the database—and collectively for the BAME community—and of the thought that this database exists and what it might mean to people who get identified without any evidence against them?
Oliver Feeley-Sprague: This goes back to the previous point I made about the way of looking at this tool as a carrot and stick kind of thing. The consequences appear to us to be much more stick-orientated than carrot-orientated. One of the things that we talk about in terms of consequence is this thing that is referred to as the Achilles heel tactic that the police have used. It is referred to in the ICO report as a kind of Al Capone technique, but it is the same issue. It is essentially where evidence of gang association and serious violence is maybe not quite as full as you might require and there is a range of other enforcement techniques to put pressure and influence those individuals. Typically that involves—I am confident to say from our research that this a widespread tactic—the police sending letters threatening evictions to families of gang members, so a direct relationship between being on the gangs matrix and your housing, your right to housing.
I know that you were hoping to take some questions on stop and search. So you don’t see these in isolation, one of the core relationships between the gangs matrix and stop and search is being on the matrix is obviously grounds for increased targeted stop and search. Our partners at StopWatch did a follow-up report on our matrix report, where they took a load of testimony from a large number of people that were on the matrix, obviously all anonymised, and as you might appreciate, it is quite difficult to get people to openly talk about these issues, because they are deeply personal data and all of that.
But it was quite extraordinary to hear that there were individuals that were subject to stop and search—and they put the figure in their own testimony—in the hundreds, so people that were stopped and searched more than 100 times because they were on the gangs matrix. To me, that is an almost incomprehensible amount of police interaction. I can’t possibly even comprehend what that might be like.
In at least one of those cases, the individual ended up with a criminal conviction for a public order offence because he was stopped and searched so many times that he became so frustrated and aggravated and so alienated by that process that he kind of lashed out against it and ended up with a conviction for public order because of that. I do not know all of the circumstances there, but I am just explaining the interrelation between these different tools. Other witnesses would basically say that for them, stop and search is as routine as getting dressed in the morning. These are incredible levels of intrusion that we would find difficult to comprehend.
Just outside things like housing, there were numerous examples—and this goes back to the question about at risk versus risk—where because there was no clear guidance on how data was shared and the conditions by which it had been shared, the blurring of the lines between risk and at risk was missed. In educational establishments, for example, young people were often removed and excluded from courses because they were on the matrix and that they were referred to as being risky rather than at risk. So you have things like housing and education as a consequence.
We did not touch on this much in our research, but I think it should be of concern to this Committee, given that we now have structural failings of this tool and it is a problematic tool with serious issues around the management and control of data. The tool is used in other aspects of state interactions. It is used in the criminal justice system and it is used in things like sentencing decisions. It is also used in other areas of the Home Office for immigration decisions, so either citizenship applications or removals and deportation, either administrative or enforced removals. It is not an area that I work on, but obviously Amnesty has a core interest and a core stream of work looking at immigration and detention issues.
We are very concerned that this tool may well be used in those settings without much scrutiny of the findings of the two reports. I guess the question is has there been any new guidance within say the Crown Prosecution Service that, “This is a problematic tool, so you need to be aware of that when using it in those environments”? Similarly for immigration decisions. It is not just about individuals and their access to housing and health, it also potentially has quite serious implications for other bits of the state.
Q155 Toby Perkins: When you say it is used in sentencing and in court cases, just expand upon that. Can you give us specific examples of where it has been used?
Oliver Feeley-Sprague: I would probably have to come back to you in writing on that, because essentially we work a lot with lawyers and people who represent people who may well find themselves on the matrix. They have certainly told us that in the work that they do, they are aware that the matrix has been used in those sentencing decisions. Sorry for my ignorance here, but I have no idea of the rules around disclosure of that kind of information, whether it is possible for me to present that to you or not, but I can certainly come back to you with more information.
Q156 Toby Perkins: Yes, because I think many people would be alarmed by the breadth of usage that you have outlined there, so I think any evidence you could provide about it being used to prevent people being on a college course, it being used to increase someone’s sentence unrelated to that or any of the other things you have just referred to would be very useful.
Oliver Feeley-Sprague: On the housing issue and that sort of general issue, yes, of course we have numerous interviews and case examples that we can pull together. Obviously they are anonymised, as you would expect, but yes, I just have to come back to you on the sentencing issue.
On the immigration issue, the area is under-researched. All we do know is it is clear that the Metropolitan gangs matrix is a tool that is used in that process. That is all we know.
Q157 Tim Loughton: Just to be clear, and I am sorry if I missed it at the very beginning when I was not here, but who has access to the matrix database?
Oliver Feeley-Sprague: That is the point. It is incredibly broad. We did a series of freedom of information requests. Going back to what I said, one of the main problems is that the 32 boroughs that operate their own version of the matrix all had different rules about how the data were shared with third parties, but typically the data are shared with local authorities, with youth offending teams, with the Probation Service and housing.
Q158 Tim Loughton: Who owns the data?
Oliver Feeley-Sprague: It is a Metropolitan Police tool, so I guess it is their data.
Q159 Tim Loughton: What information-sharing regulations are they subject to? Because you are saying it can be used by housing, educational institutions or whatever, so would I be able to access it?
Oliver Feeley-Sprague: That is a very good question. Certainly following the ICO enforcement notice, the answer to that question should be no, unless you had a very good reason, a proportional and legitimate reason to have access to it and that there was a clear written data-sharing agreement. I think that is a legal requirement under the Act, which is that if you are going to share data, there has to be a legal written agreement on how that data is shared and the purpose for which it is going to be used.
Q160 Tim Loughton: You have just given some examples of where people have been stopped and searched over 100 times and that is linked to the fact that they are on the matrix. How exactly does that work? It is not as though it is a car registration, where the police will routinely stop a car because it comes up on their vehicle recognition as not insured or whatever. How do police know to target certain individuals, given the relative demise of community policing? Do they go out with a list of who is on the matrix and then effectively seek them out in order to stop and search them? Because once you have stopped them to ascertain their identity, then you are a stop and search statistic, so how is it being enforced, that somebody could be stopped 100 times because they are on the matrix?
Oliver Feeley-Sprague: I have never been out with the police on an operation, so I can only go by what I have been told by police and by the people that have been matrixed about what their experiences are. My understanding is, yes, you are entirely correct. Essentially the police have you on a list with various details, which may well include vehicle registration details, and therefore will target you specifically. They basically have you on the list and therefore will pursue you because you are on that list. That is how it works.
Q161 Tim Loughton: But somebody who is not driving a car, walking along the street, how do they know?
Oliver Feeley-Sprague: I assume that in a borough level matrix, by the time you break it down, it does vary, but say there roughly are 3,500 people on the matrix, by the time you have divided that into 32 boroughs, you are talking about a much smaller subset of people. It would be much easier for police officers keeping watch over a couple hundred individuals that they know they need to target. I would say at a local level, it is—
Q162 Tim Loughton: This is a form of surveillance, you are saying?
Oliver Feeley-Sprague: Yes. The matrix tool is an intelligence tool and surveillance is part of intelligence, so yes.
Q163 Tim Loughton: But the police are not operating facial recognition technology yet on their person, so they have to be openly targeting a set of individuals, just as they would target known criminals.
What is the age qualification for being on the list? What is the youngest person on the list?
Oliver Feeley-Sprague: Twelve. A significant number of people are under the age of 24, the vast majority.
Q164 Tim Loughton: Do you think anybody under the age of 12 is unofficially on the list?
Oliver Feeley-Sprague: I have no idea, I am sorry.
Q165 Tim Loughton: Do you suspect it? The reason I ask is because the All-Party Children’s Group did a report on police stop and search and the relation between the police and other people a couple of years ago. There was an alarming postcode lottery as to the level of stop and search, but particularly the level of stop and search for under 12 year-olds, where there are separate procedures anyway. They showed huge differentials between the way the various forces were operating there, so hence my question as to whether you think there is some unofficial matrix for under 12 year-olds, sort of mini gang members or fledging gang members.
Oliver Feeley-Sprague: There clearly is an informal version of the matrix that is operated in some boroughs because the ICO found that there was. On the issue of whether people are removed from the matrix, they were very clear that they found evidence of people being removed from the main matrix, but kept on a separate at-risk list. There is a distinct possibility that that is true.
I cannot say one way or the other whether there are children under the age of 12 on the matrix, but what I can say is that the risk that that is the case is high, given the incredible deficiencies in terms of oversight and data collection methodology that was used to put people on the matrix in the first place. The whole tool is kind of broken in that respect anyway. I guess that would be a question that you could follow up with the police. I think you might be due to take evidence on this issue from the police running the matrix. Am I correct?
Chair: We will set up further evidence sessions in due course. Janet Daby.
Q166 Janet Daby: Thank you so much for all the information you have presented us with so far. If I can just talk about the matrix a little bit further, have you ever seen the matrix?
Oliver Feeley-Sprague: One of the colleagues that we work with in a partner organisation has seen copies of the matrix. I have seen his copies. I have not held it in my hand but I have seen him hold it up, if that is what you mean. That is why I know that it is essentially an Excel spreadsheet type of tool. Lots of people who we interviewed have seen it because they are either the people who are processing the data on it or they are part of the wider information-sharing group that get access to see it at meetings and that sort of thing.
Q167 Janet Daby: How many people would you say would be on this matrix? I know it would obviously vary from borough to borough, but how many would you say, roughly?
Oliver Feeley-Sprague: I think when we published our report there were about 3,800 people on it. I should say one of the improvements has been the overall transparency around the Met Police. One of the things now is that the Met Police has provided much more information about the matrix. There is a page where a lot of these documents are now stored. The latest March 2019 figures for the number of people on the matrix put that number at 3,134. It is a little bit lower than when we published it last year.
Currently there are 3,134, but this is the key thing here. I think this is slightly alarming because we are talking about a March 2019 publication, although the caveat is that the data relates to the last quarter of 2018, so there is a little bit of overlap there. Given the ICO enforcement notice and the MOPAC review, which is basically saying, “Green people are problems. We need to be very clear on why”, if you look at the latest 2019 figures and the breakdown of the 3,134 people, 65% of people are still identified on that as being green. It is still the same proportion.
Q168 Janet Daby: On that point, would people who are listening to this Committee—I would like to think the Policing Minister is also listening to this—be horrified to hear the high percentage of victims who are on this matrix list as well?
Oliver Feeley-Sprague: Yes, I think that was surprising to everybody. It was certainly surprising to us and it was certainly surprising to the ICO that there was no clear distinction between perpetrators and victims. That seems quite astonishing to me that that could be the case, but there was no rigorous, robust or even lawful—the ICO have ruled this unlawful with the breaches—information-sharing methodology and agreement in place. By that I mean there was no way of the people who were being given copies of the matrix—and that varied between different boroughs, different boroughs could share it differently. One borough, we were told, would basically e-mail a copy of the matrix in advance of meetings. Another London borough did not do that but would project copies of the matrix on to a big screen for people to see. Those are two completely separate ways of interpreting the requirements on data-sharing across the boroughs, which points to the fact that data-sharing agreements were essentially non-existent.
By the time it got to data-sharing among partner agencies, this differentiation between victim and perpetrator was missing. Again it goes back to this risk versus at risk. A large number of people were treated as being risks even though they were likely to be victims because there was no robust methodology in place to scrutinise the data. That is very worrying.
Q169 Janet Daby: Am I clear from what you are saying that it sounds like people are being discriminated against in various areas of their life because of being on this matrix?
Oliver Feeley-Sprague: Yes, absolutely, definitely. Let us remember that the figures on disproportionality for this tool are startling. In our report we put the figure at 78%, so 78% of people on it were essentially young black men. The March 2019 figures from the Met put that figure higher. It has gone up, not down. It may not be statistically significant because it is only a couple of percentage points out, but they are putting the figure, as of their last published data, at 80% of people on this database being young black men. That is out of all sense of proportion to the demographics or even the relationship between ethnic grouping and crime. It is way, way out of proportion.
Q170 Janet Daby: If I could just move on to stop and search, obviously the police have a variety of stop and search powers available to them under different legislations. That could be under PACE 1984, the Misuse of Drugs Act, Section 60, traffic stops or obviously under the Terrorist Act. You have mentioned already that when people are on the matrix there is a link to them being more likely to being stopped and searched and you have obviously mentioned over 100 times in some cases.
Oliver Feeley-Sprague: This is what those who were on the matrix said, yes.
Q171 Janet Daby: If the police have stopped people to that level, obviously there would be some level of recognition that people who are on the matrix are therefore being stopped and searched. What is your view on that?
Oliver Feeley-Sprague: I need to state that even though they are related issues and I do not want you to see these in isolation, we do not work on stop and search as a specific issue. I have no clear mandate to speak to issues of stop and search, but obviously there is a close relationship between those issues and the matrix, as I have already described. It clearly is one of the enforcement tools that the matrix process uses. One of the ways enforcement is taken is increased stop and search of people who are on the matrix and we know that is true.
I would also say that obviously we work very closely with StopWatch. We could not have done our matrix report without their help. I am fully aware of the kind of arguments and the kind of things that they would talk about and I can see a lot of similarities between what I would say and what they would say on stop and search. On the disproportionality of stop and search against black people, you are essentially eight times more likely to be stopped and searched if you are a black person as of now, which is worse than it was 20 years ago.
Q172 Janet Daby: Obviously you are not the expert in stop and search and I appreciate that, but what would be your views in terms of how that affects black communities? Obviously you have touched on that, but do you think stop and search is an effective use of legislation? Do you think it is having a beneficial influence on stopping crime and how do you think it is affecting the BAME community?
Oliver Feeley-Sprague: Again with the caveat that this is not an area that we have specific competence to talk about, what I can say is that in the methodology that we apply, Amnesty is very much an evidence-based organisation. We try to present our findings and our conclusions on the basis of evidence that we research. I think it is true that if you look at the evidence from official bodies, whether I am talking about HMRC or the College of Policing, the relationship between stop and search and the reduction of crime is often overstated.
Last year the College of Policing produced probably one of the most in-depth studies, looking at records of stop and search over 10 years. I might be paraphrasing the findings, but what they concluded by looking at the actual data was that if you increased stop and search by a magnitude of 10 times, so you did 10 times more stop and search, you were likely to see reductions in crime of the magnitude of 0.1% to 0.3%. They acknowledge that stop and search may in some cases lead to reduction in crime, but there is an incredible mismatch between what the evidence says about how effective it is and the statistics that show its effect on reducing crime. This is the College of Policing, not Amnesty. This is the College of Policing in a report saying that if you increased it by 10 times, you are likely to see a reduction in crime of 0.1% to 0.3%.
One of the things that they do acknowledge though is that the use of more targeted stop and search may be beneficial. Where you have specific intelligence around hotspots or those sorts of things, it may well be the case there that enhanced stop and search might be a beneficial prevention tool. Where it links to the matrix is that we know that the matrix is one of those tools that is used to decide whether or not to put someone on a list for enhanced stop and search powers. What I am trying to say is that the matrix is a very problematic tool, so if you are using that as a basis for enhanced stop and search, as a kind of rationale for doing it, there is a problem there.
Q173 Chair: I have a few follow-up points, just to clarify. On this issue about the potential relationship between the matrix and stop and search, is there any data or is that just the anecdotal evidence?
Oliver Feeley-Sprague: In terms of numbers, I do not think there are factual numbers that would say, “Here are X stop and searches that were a result of being on the matrix”. It depends how you want to refer to evidence. It is certainly the case that there is evidence in terms of witness testimony from people who were on the matrix and their experience of stop and search, who have said that being on the matrix has led to enhanced stop and search.
We also know from the police’s own confirmation through our correspondence with them that enhanced stop and search is one of the tools at their disposal when looking at the wide range of measures to do with the gangs matrix. In this Achilles heel suite of measures, enhanced stop and search is one of them. Those two are factual.
Q174 Chair: Is there any evidence on the differential for red-listed, amber-listed and green-listed by stop and search?
Oliver Feeley-Sprague: Not as far as I am aware, but I could look and see.
Q175 Chair: That would be really helpful because obviously if there is a difference between the issues around the green traffic light you have raised and the red and so on, that might raise some additional questions.
Similarly on the traffic lights, on the disproportionality, which as I understand it is even recognised in the Met’s own equality impact assessment, is there any evidence on the disproportionality on red, amber or green, the different traffic light categorisations, or is it just the whole thing?
Oliver Feeley-Sprague: My impression is that it cuts across the whole database, but I am pretty sure that if you wanted to interrogate that data then the police or MOPAC, who also have access to that data, would be able to provide breakdowns on that for you.
Q176 Chair: Fine. Looking at recommendations to address this, you have obviously talked about the overhaul in terms of the gangs matrix more widely, but what about just dealing with the disproportionality issue? Do you have any specific recommendations on that?
Oliver Feeley-Sprague: We did not answer the question why in our research. We basically uncovered what we considered to be serious human rights violations. The next job is to ask, “Why is this the case?” We point to it in some of the ways, for example, that social media may well have been misused as an evidence-gathering tool. We can point to why it is that an outcome of the matrix might be disproportionality because people have not understood the cultural environment in which social media is used. There is that sort of thing, but we have not answered the question as to why. That is what needs to happen next.
That is why we have been engaging with the Equalities and Human Rights Commission, because we think there is a real case, maybe even wider than just the matrix, to do a proper evaluation now into why. Even the police now acknowledge that there is a problem here. We need to get to that answer.
Q177 Toby Perkins: Just for context, when you were speaking about the disproportionality you said that 80% of the people on the matrix were black. How does that compare with the best available intelligence in terms of the proportion of people who are members of gangs who are black, do you know?
Oliver Feeley-Sprague: For the taser work, with figures that I am aware of, in terms of demographics 3.3% of the population would classify in that group. From taser statistics, that is what it is. It is much lower. If you are looking at crime statistics, it is 27%. You have 80% to 27%.
Q178 Toby Perkins: Is that all crime or gang-related crime?
Oliver Feeley-Sprague: All crime. Well, serious violence, which is characterised as—
Q179 Toby Perkins: Serious violence?
Oliver Feeley-Sprague: Yes. What I want to say here, something I did not say but I think is quite important, is that the matrix is often conflated with issues to do with knife crime and serious violence. I know we also put in written evidence to the serious violence inquiry. One of the things that is quite striking, and this is why we think this tool is not really fit for the purpose it is intended, is that the official data—for example, the Mayor’s Office’s 2017 knife crime report—suggests that only 5% of knife crime is flagged as gang-related activity. That is only 5% of the problem. If you are looking at serious violence through the lens of gang membership and knife crime through the lens of gang membership, you are almost looking at the wrong thing because the relationship between gangs and knife crime is statistically low. MOPAC, looking at official figures, put it as low as 5% in their 2017 report.
There are different ways of looking at the impact of serious violence and knife crime depending on how you characterise the data, but even on the other end of knife crime, wounding rather than fatalities and that sort of thing, it is still vastly disproportionate. In those statistics, which are also available from the police, 80% of knife crime is non-gang related. If your argument is that the gangs matrix is important in your fight against serious organised violence and knife crime, then I would suggest the evidence says it is not. That is one of the reasons why we are saying it is not fit for purpose.
That is not to say that knife crime and serious violence are not enormously problematic issues that we all have a duty and a responsibility to deal with. We are just looking at effective policing tools and this tool is not related to those statistics.
Q180 Toby Perkins: But would you accept that in terms of the disproportionality, unless we know the proportion of gang membership among this community it is impossible to say whether a matrix is proportionate or not? We would need to know what proportion of people from that community are members of gangs in comparison to the rest of the population, yes?
Oliver Feeley-Sprague: Yes, although I think on any yardstick an 80% make-up of a database is disproportionate. There are serious questions there about why that figure is what it is. That was my point to Ms Cooper, that the question now is why. We need to get to the answer of why.
Q181 Toby Perkins: I accept that, but it is also a key piece of information in terms of the best available intelligence, what proportion of gang membership is in which community, is it not?
Oliver Feeley-Sprague: Yes.
Q182 Janet Daby: If the crime prevention multi-disciplinary team were working with people involved in gangs to exit the gang life through that route, would that person be added to the matrix?
Oliver Feeley-Sprague: It is difficult to say because there is not much clarity on how you get added to the matrix in the first place. This is one of the major problems with the tool. It is arbitrary and it varies massively across different London boroughs. From what we know, what you have just said is incredibly likely, given our case study of the individual that we named Paul. He was clearly working on violence prevention and well-established as being one of those people on a pathway out of violence. He would claim he has never been associated in any way with crime or violence but has been put on the matrix, he thinks through association.
Chair: Thank you very much. We appreciate your evidence and your time. There were a few points that were raised, so if you are able to send us any follow-up evidence on those, that would be very much appreciated. Thank you for your time.