In tailoring interventions to particular issues and contexts, problem-solving approaches – such as SARA (Scanning, Analysis, Response, Assessment) or the 5Is (Intelligence, Intervention, Implementation, Involvement, Impact) – provide a robust process-based framework through which to specify and better understand the nature of given security problem and guide practitioners towards better-quality interventions and their implementation.

Working outwards from defining the specific crime or security problem and engaging with the end-users and beneficiaries of an intervention is a more effective approach than existing solutions or bureaucracies/organisations available to respond to the problem.

Given the siloed nature of service provision/responses and the segmented nature of knowledge and skills/resources, this demands harnessing multi-sectoral and diverse actors through pooled resources, skills, knowledge and capabilities in interdisciplinary and cross-professional partnerships.

One of the limitations that constrained the implementation of problem-oriented policing is that it focused on the police organisation as the locus of the response to social problems when the levers to the problems often lay far from the reach of the police.

Nothing works everywhere, and a lot of things work somewhere! Context matters – configured in time and space – in the causation of crime and insecurity. Crime prevention and urban security problems are complex and informed by a tangle of interacting causes and interdependencies, which differ across problems and contexts.

There has been a tendency to search for universal solutions under the banner of ‘what works’ which has drawn attention away from the situated and contextualised features of local places. And simultaneously with little regard to which groups of people benefit from particular interventions or design features in a particular place/situation at a specific time.

Early intervention also demands considering the crime and security consequences of change and innovations - in technology, products and services - at the design stage, rather than retrofitting partial solutions after innovations have occurred.

Interventions at the design stage enable up-stream, early opportunities to effect security and harm reduction outcomes, rather than retro-fitting changes after the event. Secured by Design, Crime Prevention Through Environmental Design (CPTED) and ‘defensible space’ theories have all offered important insights that have informed practical and often successful measures. The design of motor vehicle security and the subsequent decrease in vehicle related crime is a notable example.

Designs, however, must avoid being narrowly conceived around security at the cost of other social goods and security requirements need to be creatively balanced with a range of others including, aesthetics, convenience/accessibility, social inclusion and environmental sustainability.

Designing in crime and security features into new interventions necessitates active engagement and responsibility on behalf of the producers of new technologies, services and products, as well as designers and architects. As the example of the Car Crime Index (in the 1980s) demonstrated, this can require significant political and organisational buy-in as designing in crime prevention and security features from the outset may be costly and disruptive to wider commercial imperatives.

Vulnerability-led design responses or too much emphasis on security can promote fear of crime and insecurity and foster social polarisation, with adverse implications for wellbeing.

Human-centred design solutions afford sensitivity to local context, a focus on the nature of the problem(s) to be addressed, an understanding the causes of social problems, the nature of social interactions and the ways in which people use and adapt to solutions/interventions.

Involving communities (or representatives) in the design of interventions creates a sense of (local) ownership and participation, as well as ensuring local context is accounted for and incorporated.

Cost-benefit analyses suggest that resources spent on security, policing and crime prevention might sometimes be better spent on other public services and essential infrastructure - i.e. health, education, transport and culture.

There has been a tendency to prefer technological solutions – i.e. hardware – to human solutions in regard to addressing security concerns, with less regard for the intersection and interaction between social and technological processes; between technology (as hardware) and people.

Social media and the online space is often portrayed as the cause of problems and harms, but its potential as a platform for positive intervention, learning and change should not be overlooked or underestimated.

Urban security demands different data than crime data alone and necessitates thinking differently about – and differently measuring – indicators of ‘success’ and outcomes in the evaluation of interventions. Factors such as levels of perceived unsafety, civic tolerance, social cohesion, trust in authority, community well-being and victim support are salient outcomes in urban security.

Good quality data collection and sharing across relevant organisations, as well as ethically sensitive data management and use: allow for joined-up provision; afford opportunities for joint analysis and coordinated working between relevant agencies; provide the capacity to track and support individuals and families through service provision/diverse interventions, and assess their trajectories; provide an evidence-base from which to assess effectiveness; ensure the best use of resources and facilitate best practice; and afford opportunities to monitor performance and render services accountable and reviewable.

Good quality, shared data are vital in clarifying and defining the nature and extent of the problem(s) being tackled through focused analysis to ensure a properly problem-based intervention.

There is often a confusion between risk factors as ‘flags’ for (or indicators of) causes and casual mechanisms themselves, particularly evident in preventing juvenile delinquency. To distinguish between ‘causes’ and ‘flags’, we need to identify a plausible explanatory process (theory of change) that connects the supposed cause and effect and that actually produces the effect.

Interventions and their evaluation need to be clearer about the causal factors (and the theories of change) that it is assumed will cause a mechanism to produce certain desired outcomes. Hence, we need strong and credible reasons for how and why the assumed cause will produce the effect in question.

Evaluation is important for development (to help strengthen institutions), for knowledge (to provide a deeper understanding of specific questions or fields) and for accountability (to measure the outcomes and their effectiveness/efficiency).

Methodologically, the ‘what works’ movement - through its emphasis on quasi-experimental methods and random control trials - has (deliberately) focused attention on single interventions and sought to remove contextual factors and the analysis of the implementation processes, in order to highlight constant conjunctions.

Programme evaluations need to play greater attention to both the context and the processes of implementation in informing what works, where and for whom.

For evaluations to be meaningful, the aim of the intervention needs to be clearly defined, as do subsequent outcome measures by which the success of the intervention can be assessed.

Rather than seek to evaluate the presence or absence of a successful crime preventive effect, there is a need to explore the causal mechanisms (or ‘theories of change’) that are believed to underlie and produce those effects/outcomes (or their absence). Understanding how something works or is intended to work, enables more focused design of interventions that also take account of contextual factors.

Knowledge about failure and of undesired side effects is as important as learning about success. Urban security evaluations tend to focus on success stories and in policing interventions too often appear ‘doomed to succeed’.

The overwhelming lesson from the last 30 years is that the institutional context and resistant organisational cultures have often undermined the implementation of research-informed urban security and crime prevention. It is not that the science is poor with regard to crime prevention and urban security – although it is inevitably incomplete, in some places inadequate and shifting in the light of technological and social change - but rather that it is not being implemented or implemented in inappropriate ways, circumstances and situations that constitute the most basic contemporary challenge.

The importance of political leadership, public trust and institutional commitment, support, appropriate levels of resources and buy in from relevant stakeholders are all pivotal to the success of interventions.

Communicating the successes of crime prevention and the effectiveness of up-stream early interventions in ways that elicit long-term political commitment and organisational change remain a considerable challenge.

There is a long history of successful experimentation in urban security with robust evaluation to support their effectiveness and impact, but the lessons from which are not mainstreamed and realised in routine organisational practices or not appropriately transferred to other places and populations.

Demonstration projects may provide interesting insights and learning but will result in little change if they are not embedded within infrastructures that align with cultural values, underpinned by sustainable funding and supported by long-term organisational commitments.

Effective multi-stakeholder partnerships require: shared ownership; clearly defined outcomes and expectations of each contributing partner; acknowledgement of asymmetries of power differentials; constructive negotiation of conflict; mutual understanding and regard for difference; trust and information-sharing; and meaningful engagement with end-users and beneficiaries.

Developing shared values in collaboration demands that partners understand each other’s priorities, values, positions and limitations well enough to have meaningful dialogue about the different interpretations of the problem, and to exercise collective intelligence about how best to seek to resolve it.

Insufficient regard has been accorded to understand the diffusion of innovations and the structural features of organisations, including their propensity to take up new knowledge (absorptive capacity) and the presence or not of a receptive context for change, including things like organisational culture and environment.

Responding to public perceptions of insecurity by providing additional security interventions, technologies or hardware may fail to engage with the issues underlying these demands. It may also miss the opportunity to subject these demands to rational debate and local dialogue. Hence, the need to engage local publics, stakeholders and user communities in genuine problem-solving processes that investigate beyond the immediate appearance or superficial expression of security problems.

Seeking solutions to problems of local order through security alone may serve to exacerbate population’s fears and entrench perceived lines of difference within and among local communities.

Predicting future crime and security trends and developments, given their dynamic nature is intrinsically difficult.

All evaluations produce knowledge of what worked (in the past) for a particular population, under specific circumstances, at a particular time and may not hold for a future population at a different place or time. The inferences that can be drawn are contingent.

The knowledge base with regard to causation and the causal interactions between multiple factors remains limited.

The role that social, educational and welfare provisions play in shaping the propensity for crime and criminal behaviours remains poorly understood.

Too little is known about and insufficiently robust data are collected concerning the processes of implementation that influence the effectiveness of urban security interventions.

There is insufficient understanding of the ways in which context shapes successful outcomes and the nature and extent to which particular preventive mechanisms are context-determined or context-dependent.

More can be learnt comparatively about the ways in which urban security interventions and their effectiveness are shaped by differing culture, social practices and legal, political and administrative frameworks.

There is a need to better understand the extent to which crime prevention lessons from the physical world translate into cyberspace and their possible application (or not) to online environments.

The implications for urban security of artificial intelligence (AI), machine learning and algorithms build into products, services and utilities are largely uncharted, as expert knowledge and processes of interpretation are replaced by machine learning and automated decision-making. What we do know is that these algorithms are not impartial but embed with different assumptions about behaviour and risk that are opaque and obscure. As such, they raise fundamental ethical and normative questions about the values that inform the future of urban security.

Climate change, an ageing population and growing social polarisation, diversity and inequality are all likely to interact with wider social and technological change in ways that are more complex, interconnected and interdependent, raising new challenges for the tense relationship between liberty, security and other social values.