Fixing a hole? Potential solutions to the problem of disclosure
In the wake of the collapsed cases of Liam Allan and others in late 2017, the House of Commons’ Justice Select Committee has launched an inquiry into the ‘extensive issues’ with criminal disclosure. They are currently inviting written evidence on this issue. One of the central questions the Committee is asking is as follows: Are the current policies, rules and procedures satisfactory to enable appropriate disclosure of evidence and support the defendant’s right to a fair trial?
In our view, they are not. Tom Smith and Ed Johnston look at potential solutions to the disclosure scandal
The recent spate of cases collapsing due to failures in disclosure of relevant evidence arguably represents only a small (if highly impactful) part of a long-term procedural and cultural problem. The Criminal Procedure and Investigations Act 1996 (CPIA) introduced extensive reform of the disclosure process. Yet, in the two decades since, various researchers and commentators (including Sprack (1997); Plotnikoff & Woolfson (2001); Epp (2001); Redmayne (2004); and Quirk (2006)) have been critical of unresolved flaws in the regime, highlighting the failure of the police and prosecution in complying with their duty to share exculpatory evidence with the defence in a timely manner.
Notwithstanding attempts to fine-tune the process through ‘soft’ regulation – such as the Attorney General’s Guidelines on disclosure and the Crown Prosecution Service (CPS) disclosure manual – such issues persist. The review of Lord Justice Gross (2011) and the report of the HMCPSI (2017) continue to demonstrate that the regulatory structure for disclosure is arguably not fit for purpose, which has potentially severe consequences. We would argue that to perpetuate a policy of, ultimately, tinkering with the existing system will likely result in few substantial improvements.
The working theory behind this conclusion is, and has been for some time, that problems with criminal disclosure are not merely caused by under resourcing, poor training, or leadership deficit (which all may play a part). It is underpinned by a problem of culture.
The police (and to a lesser extent perhaps, the CPS) embody the adversarial nature of English and Welsh criminal justice. On a day to day basis, they are charged with detecting, investigating and prosecuting a range of offences, from the commonplace to the most distressing and serious. Whilst the police are conceived, in theory, as objective investigators, it is hard to escape the reality that their role encourages (and possibly demands) a focused, prosecutorial mindset. At the same time, the police and CPS are asked to objectively and neutrally assess whether evidence undermines their case – and by extension the aforementioned investigatory work – and assists suspects of crime.
There is a very obvious conflict of interest between the adversarial tendencies of police and prosecution work, and the objective needs of a fair and accurate search for the truth. This is a fatal flaw in the design of the disclosure process; like any structure, it will stand or fall depending on how robust its foundations are. Giving the police and prosecution sole responsibility for determining whether evidence should be disclosed to, effectively, their opponents is a faulty cornerstone. This is exacerbated by a virtually invisible process; very little attention has, until recently, been paid to disclosure decision-making at the earliest stages of criminal investigations and there is no clear recording or auditing process for such decisions. The paucity of information included in the Metropolitan Police Service review of the Liam Allan case should be noted as evidence of this.
The ‘explosion’ of information available during criminal investigations is certainly a factor. Clearly, when large amounts of material are available to officers in a case, they will feel the need to focus on the most relevant. This already challenging process is exacerbated when time and resource pressures effect officer decision making. It is arguable that the well documented squeeze on police officer numbers, as well as the general desire to progress cases more swiftly (at both the investigative and court stages) may be increasing the risk of important material being missed. However, it is also possible that such an argument – which effectively relieves officers and prosecutors of responsibility for mistakes – may simply disguise a ‘tunnel-vision’ approach to reviewing evidence for disclosure, focusing only on that which supports a prosecution. It is, however, very difficult to establish this due to the lack of transparency in the process and the deficit of evidence available.
With regard to wider policy issues, changes to legal aid may play a role.
For example, limited resources available to criminal defence lawyers may discourage or prevent them from pushing for more extensive disclosure or thoroughly examining unused evidence for relevant exculpatory material. This is, however, speculative. Moreover, if the problem lies with initial disclosure by the police and prosecution, the capacity of the defence to comb evidence is less relevant, as material will simply not be available to them. Another wider policy issue of note, influencing both the investigative and court stages of criminal justice, is the focus on efficiency of proceedings (explored in depth in the Leveson Review).
A general desire (or pressure) to progress cases swiftly may lead to limited early disclosure – either before or after charge, or before a first appearance at court. In research by Cape and Smith (2016) examining detention and bail prior to trial, practitioners commented on the very inconsistent availability of material, evidence and information prior to the first hearings. Moreover, the study found that the cases presented at first hearings tended to rely on police summaries which were relatively brief. Such findings are particularly concerning in light of the push to encourage early guilty pleas. Whilst this policy has its place in saving unnecessary cost and stress for complainants and witnesses, it may be much more difficult for defendants and lawyers to enter an appropriate plea when they have very little information about the case against them. The cliched adage that ‘a defendant knows if they are guilty’ is simply not an answer to this problem. Disclosure is therefore a key stepping stone in ensuring that cases charged and brought to court are effectively dealt with at the first appearance. This also impacts bail decision making, as defendants and lawyers with little disclosed information may feel unable to effectively make representations regarding bail at the first appearance (see Cape and Smith (2016) for more on this).
It should also be noted that Article 7 of the EU directive on the right to information in criminal proceedings is very clear that ‘documents… essential to challenging effectively… the lawfulness of the arrest or detention’ and ‘at least… all material evidence in the possession of the competent authorities’ should be made available to suspects and their lawyers. Whether current practice is compliant with this is questionable. In arguing that leaving the EU will not result in a race to the bottom, the Government has expressed confidence that this jurisdiction will achieve high standards on a ‘voluntary’ basis post-Brexit. If that is to be more than a pipe dream in the area of criminal disclosure, it would seem vital to take action to ensure that this directive is properly transposed into British Law.
In the spirit of productive debate, the authors propose three potential alternatives to the current regime of disclosure at the investigative stage. All involve, in essence, removing some responsibility from the police and prosecution for determining whether evidence should be disclosed to the defence.
Alternative 1: Full disclosure
The first is what might be termed ‘full disclosure’ – that is, providing the entirety of the available evidence (except for sensitive material) to the defence, with an ongoing duty to continue disclosure of all evidence. This approach would address the potential for adversarial ‘game-playing’ on the part of the police or prosecution since it would prevent material from being hidden from the defence (unless classified as sensitive). This solution is not without problems. It does not necessarily combat the possibility of relevant material being missed, particularly in light of the levels of material now potentially available as electronic information. Suspects in police stations (many of whom, according to various studies, do not receive legal advice) would be unable to effectively manage this burden alone, and defence lawyers would not – without a significant change in policy direction – have any greater capacity than before to undertake the work of examining such evidence (currently the role of the police).
Since the police would likely play no further role in assessing whether evidence is exculpatory – and therefore fully embrace an adversarial, prosecution-minded role – it is unclear whether this solution would achieve anything other than over-loading the defence with a haystack of evidence, and few tools for finding needles.
Alternative 2: Judicial Disclosure Officer
The second solution would be to remove responsibility for decision-making regarding disclosure from the adversarial parties altogether. We would propose creating a specific, independent role for determining whether evidence should be disclosed to the defence in a neutral and objective manner. We propose that this should be a judicial figure (either qualified/experienced to the level of District Judge or Deputy District Judge), with a significant background in criminal legal practice. This role would purely be concerned with:
- Initially receiving all evidence from the police, and receiving further evidence on an ongoing basis
- Determining what, if any, evidence fulfils the current requirements under the CPIA for disclosure of exculpatory evidence to the defence
- Determining what evidence is non-disclosable on the basis of irrelevance to the defence
- Making decisions about withholding evidence on the grounds of sensitivity (for example, on Public Immunity or Article 8 ECHR privacy grounds)
- Doing this on an ongoing basis for the life of a case
To be clear, this figure – which we tentatively suggest could be called the Judicial Disclosure Officer (JDO) – would make no decisions regarding the strength of a case against a suspect or what evidence the police/prosecution should use or not use in a case.
They would solely be responsible for determining disclosure. We feel this would fully address the cultural problems which currently hamper fair and relevant disclosure. There would, however, clearly be resource implications. Since the JDO would make decisions regarding disclosure in a large number of cases, this would require full-time and experienced staffing, located in every police force area to ensure adequate coverage. Whether this can (or would) be resourced is a difficult question to answer, as it would likely need to be centrally funded to protect independence. Indeed, the idea of a semi-independent police-based figure is not recommended. This concept underpinned the introduction of the Custody Officer role under the Police and Criminal Evidence Act 1984, which has and continues to have problems of independence (see Dehaghani (2017)). This would be a very significant step, as it would effectively be creating an entirely new figure in the English and Welsh criminal justice system, perhaps most closely comparable to an investigative judge in the inquisitorial systems of various European jurisdictions. We consider this solution to have the most potential to ensure the long-term fairness of the disclosure process.
Alternative solution 3: AI
A third solution would be to harness technological innovation to assist the police in undertaking their current role. The most obvious method of doing so would be the use of Artificial Intelligence (AI) or algorithmic decision-making in the disclosure process. There is some history in the use of such technologies in criminal justice practice. From January 2017, New Jersey became the first state in the US to almost eradicate the use of bail completely. Judges now use a nine-factor algorithm to examine whether a defendant is either dangerous or likely to abscond prior to trial.
The defendants who are classified in either category can be detained and the remaining defendants are released albeit with various monitoring conditions attached. Such techniques have emerged closer to home. In May 2017, the BBC reported that Durham Constabluary, with the assistance of academics from Cambridge and Winchester Universities, were preparing to ‘go live’ with an artificial intelligence system designed to assist custody officers in deciding whether or not to detain suspects at the pre-charge stage. At a basic level, the system employed an algorithm called the Harm Assessment Risk Tool (dubbed ‘HART’) which would classify suspects as being a low, medium or high risk of reoffending. This, in turn, would help custody officers determine whether or not it is necessary to remand the suspect in custody or release them, with or without bail. The decision-making of the algorithm is based on data which Durham Constabulary gathered between 2008 and 2013. This data includes behaviour predictors such as the criminal history of the suspect as well as age, gender, two forms of residential postcode and existing police intelligence. In 2013 the tool was tested and was found to be ‘correct’ in 98% of cases, with ‘high risk’ forecasts correct 88% of the time.
Durham Constabulary have stated that the decisions will only be ‘advisory’ during the experimental use of the tool. The latter fact – that humans remain the ultimate arbiters – has been welcomed by critics who complained that reliance on factors like gender and postcode undermined the fairness and reliability of the algorithm. That being said, if the tool continues to make ‘correct’ decisions in 98% of cases, there is some argument to be made that humans cannot necessarily do any better. These are examples, and apply primarily to risk assessment and prediction. But clearly there may be some scope for adopting or utilising a similar approach in assessing what material is suitable for disclosure, and therefore presents an opportunity to make better, fairer decisions. A key benefit of this solution would be to tackle the issue of high volumes of evidence. With well-designed and reliable technological assistance, the police could shift into an oversight and fine-tuning role in ensuring disclosure is undertaken properly, rather than attempting to grapple with very large amounts of material with varying levels of relevance or usefulness.
Equally, if designed properly, an algorithm which determines whether evidence needs to be disclosed or not would be entirely neutral and not subject to the bias that some police officers (and, quite frankly, human beings generally) will inevitably succumb to. Of course, this raises serious questions about how such a system would be designed, if it could be designed. This would take significant time and investment to achieve.
Despite long-term anecdotal evidence and individual case examples demonstrating issues with criminal disclosure, we believe that a greater body of research evidence would greatly inform future reform. At present, the authors are preparing a bid for funding to undertake independent research into disclosure practices in police stations and courts. We believe that, in the long term, empirically tested evidence would help to establish the true extent of the issues, their causes, and help inform potential solutions. This article is designed to kick-start a conversation which is solutions-focused. The authors recognise these solutions are imperfect, but would welcome input and dialogue to take them forward.