Introduction

Public service co-production is an ever-evolving concept, closely context bound and highly dependent on the parties involved. It is a multi-faceted and relational process where end-user engagement is integral to service effectiveness (Radnor & Osborne, 2013). With the exceptional COVID-19 circumstances in early 2020, public service co-production went through a push towards digitalisation and the role of online applications became critical in ensuring service continuity. Carroll and Conboy (2020) argue that technology-driven practices will form a part of ‘the new normal’ and organisations need to normalise the use of technology to accomplish service goals, possibly resulting in long-lasting effects on public service co-production. Even before the pandemic, it had been stated that there is a need for further research into the interplay of co-production and digital technologies (Cordella & Paletti, 2018). One avenue, which has not been studied, is the digitalisation of service co-production inherently reliant on eye-to-eye contact and citizen participation. Restorative practices, which by their nature require physical interaction and active participation, offer a unique opportunity to understand how digitalisation influences the ability of service professionals, individuals and the community to contribute to highly interactive and sensitive public services such as restorative justice. Showing what happens to restorative practices when they are co-produced digitally can shed light on the consequences of abrupt digitalisation of human connection-dependent service processes, which the digital co-production literature so far has not focused on.

Restorative justice is a way of approaching a conflict or crime by actively involving the effected parties—the victim, the offender, and the community—in order to repair harm (Chapman & Törzs, 2018; Van Ness & Strong, 2010) through practices like victim–offender mediationFootnote 1 or restorative conference.Footnote 2 In essence, it is a co-produced approach to delivering justice, with personalisation woven into it for the offender and the victim (Loeffler & Bovaird, 2020). Like other services in 2020, restorative practices switched to digital channels. Restorative practices require active citizen participation in the co-production process (Daly, 2016; Sherman et al., 2005), but due to COVID-19, face-to-face meetings became highly restricted. However, with strict movement restrictions raising stress levels in communities, there was imminent need for conflict resolution. This raised the question whether digitally provided restorative practices were similar in nature and effectiveness compared to analogue services.

In order to understand the immediate effects of digitalisation on co-produced justice and the future prospects for digital restorative practices, an explorative case study was conducted in Estonia, Finland, Ireland and Portugal. As a new approach to analysing considerable changes to co-production of services, normalisation process theory (NPT) (May & Finch, 2009) was used to map the process of change that took place due to rapid and unplanned digitalisation. NPT is a framework that helps to understand key mechanisms behind normalisation, which is a process of implementing, embedding and integrating new practices into routine work (May & Finch, 2009).

Examining this case study could strengthen the wider discussion on using digitalisation in service co-production settings where (a) the change for digital channels is not voluntary and (b) where the nature of the service does not support going digital. This is an avenue not explored so far. Considering that online restorative practices are a recent phenomenon and there are not much data available, the article cannot provide far-reaching conclusions or in-depth analysis on long-term impacts; however, it can shed light on the emerging processes taking place when digitalising a very sensitive co-produced service. In short, the article aims to answer the following research questions:

  • What were the immediate effects of digitalisation on co-producing restorative services?

  • How were the digitally mediated practices normalised while safeguarding the fundamentally co-productive nature of restorative justice?

As far as the author knows, it is one of the first scholarly attempts to explore the immediate effects of an abrupt shift from face-to-face to digitally mediated co-production process in a highly sensitive field. The article begins with explaining the notion of co-production in the context of restorative justice, outlining some possible effects of digitalisation and introducing the framework of NPT. Next, a case study is presented to illustrate how new restorative justice practices were embedded into the emerging context in four European countries. The article concludes with an analysis of the empirical evidence and an interpretation of the findings.

The Multi-faceted Phenomenon of Public Service Co-production and Restorative Justice

Co-production of public services has been a buzzword for decades, describing the changed nature of relations between organisations and people outside the organisation. According to Loeffler and Bovaird (2020), citizens’ role in public service co-production has evolved from the role of a ‘citizen’ in classic public administration to the role of a ‘customer’ under New Public Management and then towards a ‘partner’ in the era of public governance. In the context of this article, the definition by Brandsen and Honigh (2016: 431) serves best to explain service co-production: “a relationship between a paid employee of an organisation and (groups of) individual citizens that requires a direct and active contribution from these citizens to the work of the organisation”. In co-production, citizens and communities provide their ideas, time, skills and other resources (Bovaird, 2007; Verschuere et al., 2012) and service users play a key role in the outcomes of the service process (Tuurnas et al., 2014). For some government services, involving citizens in co-production is a core feature of value creation and an inherent characteristic of service provision, i.e. to receive a service is to co-produce (Alford, 1998; Fledderus & Honigh, 2015). Especially with regard to prolonged, complex services, customer capabilities, understanding of service requirements and perceived self- and provider efficacy are key determinants of success (Spanjol et al., 2015). On the other hand, co-production can lead to a blurring of responsibility and accountability, higher transaction costs, reinforced inequalities, putting pressure on vulnerable service users, or co-producers misusing their role (Steen et al., 2018). In short, public service provision can be seen as a move from services for the public to services by or with the public (Bovaird & Loeffler, 2012), which means that expected outcomes rely on citizens’ input in the co-production process.

Restorative practices are inherently co-productive, requiring active participation of all parties. Looking at restorative services through a co-production lens helps to pinpoint properties that are necessary for achieving service outcomes like repairing harm, restoring relationships, supporting victims and taking responsibility for the caused harm (Bolitho, 2015; Sherman & Strang, 2007). In the case of restorative justice, the only formal ‘service’ provided is facilitating the discussion between the involved parties, thus creating service value (Willis & Hoyle, 2019). For restorative practices to have meaningful impact, the affected parties have to participate actively and without compulsion, be motivated and supported throughout the process, understand and accept their role in the process (Chapman, 2016). There has to be a trustful and safe environment supporting engagement (Bolitho, 2017). It is important that a professional facilitator assists parties (Bolitho & Bruce, 2017), ensuring that the meeting is safe for all and that everyone has a chance to express themselves and to be heard (Bolitho, 2017). Much of this relies on body language, meaningful silence and eye contact (Chapman, 2016).

In short, restorative practices denote a type of service co-production that relies on human contact, mutual trust and respect, open and honest dialogue between parties, facilitated by an objective, trained outsider. The following section looks at a new layer in these types of services, i.e. digitalisation.

The Possible Effects of Digitalisation on the Co-production of Restorative Justice

Through the process of digitalisation, traditional forms of services are replaced or supported by digital options (Soto-Acosta, 2020). Digitalisation can have a three-fold effect on co-production (Lember, 2018). Indirectly, digitalisation makes it easier to exchange information or provide support functions such as identifying oneself online. It can also transform co-production by introducing new practices or adding a digital element to traditional services. Lastly, digitalisation can replace traditional co-production practices, e.g. by using sensors or algorithm-based decision-making models instead of working directly with service users. Digitalisation is often seen as something normative, a positive push towards accessibility, efficiency and innovation (OECD, 2016). Previous research on digital co-production has focused on the strategic aims of voluntary digitalisation, its long-term effects and possible impacts for future developments in service design and production (Lindgren et al., 2019).

This article focuses on how digital channels transform service processes, diminishing the physical presence of the service provider and the user in the value creation process (Osborne et al., 2014; Rantala & Karjaluoto, 2017). However, it does so in the context of an unexpected and possibly objectionable push towards digitalisation of highly sensitive, fundamentally co-productive services where the lack of human contact is not desirable or can even be detrimental. Due to the pandemic, many organisations introduced ‘tech-driven’ practices in an unprecedented and time-pressured manner (Carroll & Conboy, 2020). The transformation process was not strategically initiated, many of the changes were reactive by nature (Iivari et al., 2020), and due to time constraints, there was no reflection on the long-term sustained use of the new practices created as short-term solutions (Carroll & Conboy, 2020). The aim of this article is to take an explorative look into how initially undesirable changes are embedded in routine processes to ensure service continuity and preserve the co-productive nature of the interactions.

Digital services are usually associated with economic savings, higher productivity, better service quality, increased transparency, access to services, customer satisfaction, citizen participation and empowerment (see also Gelderman et al., 2011; Madsen & Kræmmergaard, 2016; Taherdoost, 2018). However, there are ethical and social implications to consider (Seetharaman et al., 2020). For example, digitalisation can lessen the control citizens have over the results of the services they are contributing to (Breit & Salomon, 2015). There are also worrying signs of ‘technological solutionism’ (Morozov, 2013) whereby all social phenomena are defined as quantifiable problems ‘solvable’ with a technological solution. Furthermore, digitalisation can lead to a new distribution of responsibilities, risks and potentially disruptive effects, not to mention the extra work needed to ensure the high quality of a service provided at a distance (Pagliari, 2007; Vikkelsø, 2005). Digital channels are more vulnerable to breakdown of communications, conflicts, power struggles and mistrust (Sumathipala, 2020). This becomes crucial in restorative services that are dependent on showing and reading emotions, being open to vulnerability and honest responses. There, cognitive-based trust is important, and this develops through social cues and impressions (Turesky et al., 2020). While digitally mediated social environments offer much for communicating with others who are physically distant, they have limitations regarding the manner in which information is transferred, allowing less possibilities for nonverbal symbolic gestures (Carter & Asencio, 2019), but these help to convey emotional expressions, which is a big part of communication (Van Kleef, 2009).

Normalisation of New Digitalisation Practices

In order to map empirically the process of change in switching to digitally mediated restorative justice, the article follows the NPT framework. The components of the NPT were used to pinpoint the immediate effects digitalisation had on the co-production process itself as well as the roles that facilitators and citizens played in setting up and conducting online restorative meetings. NPT “is concerned with the social organisation of the work (implementation), of making practices routine elements of everyday life (embedding), and of sustaining embedded practices in their social contexts (integration)” (May & Finch, 2009: 538). Using the model proposed by NPT to map out the steps taken to digitalise restorative justice can help understand how the abrupt digitalisation of restorative services was embedded in the routine service processes. NPT addresses the necessary factors for successful implementation and integration of new practices into routine work (Murray et al., 2010), and in this article, it is used as a contextual framework to explore and understand the digitalisation of restorative practices. It helps to analyse how and why digital restorative justice becomes a normal practice, or why it does not or should not. NPT as a framework posits that new practices become embedded in routines and integrated in the surrounding context as a result of people working, individually and collectively, to implement them. Implementation is operationalised through four mechanisms: coherence, cognitive participation, collective action and reflexive monitoring (see Fig. 1 for a schematic overview of the components of NPT).

Fig. 1
figure 1

Model of the components of normalisation process theory (May & Finch, 2009)

Coherence refers to the meanings and competencies that hold the practice together, enabling people to share and enact it. Cognitive participation reflects actors’ engagement: initiation of the new practice, enrolment in it and legitimation in relation to existing practices. Collective action refers to the activities used to enact the new practice. On the one hand, collective action is influenced by organising structures and social norms, and on the other, by group processes and conventions. Reflexive monitoring involves formal and informal judgements about the utility and effectiveness of a new practice. Reflections have an effect on social norms and group processes that, in turn, shape practice and coherence.

Method

In order to understand how digitalisation affects the co-productive nature of restorative services, an explorative abductive case study was conducted in Estonia, Finland, Ireland and Portugal. The case study approach enables us to obtain in-depth understanding about an issue or phenomenon in its real-life context (Yin, 2002). These countries belonged to a group of 15 European countries, which in 2020 started to coordinate and exchange their restorative justice experiences during the pandemic. The four countries were selected based on two variables: how established were the restorative services (relatively old vs new services) and how digitally experienced were the countries according to international benchmarksFootnote 3 (see Table 1). The aim was not to explain the differences or similarities between these countries, but to detect the emerging effects of digitally mediated and co-produced restorative practices in contextually different settings. Choosing countries with different backgrounds in terms of restorative practices´ history and level of digitalisation gives a more varied insight into the normalisation process, thus possibly providing a more general overview of the effects of digitalisation on service co-production.

Table 1 Case study country selection

Data were gathered in three phases (see Table 2). First, the author observed online meetings of the European Group for Restorative Justice in 2020. The group met once a month between April and July 2020 and included over 30 European restorative practitioners from 15 countries. The meetings focused on four main topics: necessary changes in practice to go digital (April), restorative practices in the context of COVID-19 (May), supporting staff in returning to contact services (June), and further developments in service design and staff training on online services (July). The author took notes of each meeting and systematised the information to note down reoccurring themes, keywords, challenges and interesting anecdotes, which fed into the next phases of data collection. These were then used to come up with questions and themes for country interviews, enabling systematising the information provided during the interviews (see Table 3).

Table 2 Data collection
Table 3 Reoccurring themes used in coding fioni the online meetings and interviews according to the components of NPT

Secondly, two phone interviews were held with the Estonian mediation service manager in June and August 2020 to reflect on the practical changes made in the co-production process. The manager’s input came from co-vision meetingsFootnote 4 she held from April to July with 19 volunteer mediators. Information obtained during the phone interviews was categorised according to the components and their sub nodes of NPT. Lastly, in February and March 2021, based on the insights gathered in previous steps, group and individual semi-structured interviews were held with country representatives from the selected four countries. From each country, at least one of the people interviewed had taken part in the European group meetings. The interviews were carried out over Zoom or Skype and were recorded and transcribed. The information was coded to match the four components of NPT. Grouping reoccurring themes under the components of NPT allowed to map out the digitalisation process and find similarities and differences between countries, and come up with common changes experienced compared to analogue restorative practices. Altogether, next to observing four European Group meetings, 12 restorative justice professionals were interviewed for this study.

The author acknowledges that interviewing only service professionals (mediators) and not citizens sets limitations on drawing conclusions. However, due to the sensitive nature of the changed co-production process and the apprehension that digitalisation caused in some of the participants, it would have been detrimental to the outcomes of the restorative process to gather citizens’ feedback this early in the digitalisation process. Similarly, since the change to digitalisation in restorative practices was only recent when writing the article, there were no written reports, country analyses or strategic papers available to support or refute the findings brought out in the case study.

Digitally Mediated Restorative Justice in Four European Countries

The Immediate Effects of Restorative Justice Going Digital

When answering the first research question—what changed?, it is important to note that the practical set-up of restorative services by countries varies. In Estonia, victim–offender mediation in criminal proceedings is available since 2007. From 2018, mediation is provided to juveniles in misdemeanour cases. In 2020, a volunteer-based mediation service was launched (ca 120 volunteers). On 12 March 2020, mediation services were shut down due to the state of emergency. Services went online in May 2020.

In Finland, mediation services started in the 1980s with pilot projects. In 2006, the law on mediation came into force and mediation services were coordinated by the Ministry of Social Affairs and Health. In 2016, mediation services were transferred under the Finnish Institute for Health and Welfare (Ojanne, 2017) and there are now over 1300 volunteer mediators (Elonheimo & Kuoppala, 2020). First online mediation pilot was launched in April 2020.

In Ireland, offender mediation and offender reparation programmes have been in operation since 1999 (McStravick, 2015). In 2001, statutory implementation of juvenile restorative group conferencing came into force under the Children Act. In 2019, 433 cases were referred to Restorative Justice Services (RJS) (Restorative Justice Services, 2020), and in addition, 3,500 children and young people were admitted to Garda Youth Diversion Projects (Egan, 2020). In April–May 2020, RJS kept contact with existing clients via phone or online. From July onwards, service provision was a blend of on-site and online.

In Portugal, the law on victim–offender mediation came into force in 2008 and mediation services were launched in 15 courts as a pilot project. Since 2017, victim–offender mediation can be applied as a diversion measure in the early stages of the criminal process and for minor offences. In 2010, there were around 300 cases per year; no statistics have been available since 2017. From March 2020, the law allows online mediation.

When countries went to lockdown, the first instinct was to wait until lockdown ends and then resume restorative services. However, because there was a heightened need for restorative practices, a digital solution was needed. After the initial experience, the involved professionals perceived that the digitally mediated process itself is not fundamentally different from the physical one. However, the quality and the standard of the co-production process were seen to be somewhat depleted; it was more difficult to build connections, relationships and trust.

We use the phrase “seeing the whites of their eyes” and we see those whites on the screen but it’s not the same. There’s not that connection, I don’t know what it is, to me it’s kind of cold whereas I like to think that when we work with people there’s warmth. (IRL)

According to the interviewees, what was perceived to be missing from the digital co-production process, is informal chatting, resulting in extremely business-like online meetings. The importance of chitchat was only properly realised when it became absent during online mediation (IRL, FIN2). In addition, they found that online mediation requires a longer preparation time and more active input from the facilitator, so everyone has a chance to speak, be heard and understood (Group2). Furthermore, during online meetings it is difficult to make sure that everyone participates voluntarily, which is a prerequisite for the co-productive nature of a restorative process (EST SM).

In accordance with previous research on digital co-production, interviewees mentioned some practical benefits of digitalisation. Online meetings enable to save time (FIN1; FIN2; EST) and enable including specialist support persons more easily (Group2). Practitioners also said that sometimes participants prefer online mediation because it provides a safe distance, as people do not have to physically meet the other person. At the same time, they argued this digital shield could lead to a false sense of security and when victims actually come face-to-face with the offender, the fears might come back (Group4). They also said digital conversation lacks the human touch needed in restorative practices and we should refrain from supporting digital solutions, albeit they are convenient, cost less and help to save time (Group1). In addition, screen fatigue was raised as an issue (IRL) that can hinder the co-production process especially for the participants, resulting in less-than-optimal outcomes for restoring justice.

Interviewees noted that online meetings can be very intimate as they allow to see how the other person lives and what their home looks like (Group2). On the other hand, the interviewees brought out that people sometimes acted too freely at home, e.g. one participant was noticeably intoxicated, and another lit a cigarette during a very sensitive discussion. This raised questions as to whether online meetings should have a different set of rules. Participants discussed having to take into account intruders and interruptions during online meetings, e.g. children or other family members accidentally walking into the meeting, so mediators often ask ‘are you alone?’ before starting the meeting. This is something new compared to regular mediation meetings where facilitators have better control over the surroundings.

When looking at the success in digital co-production of restorative justice with regard to the chosen countries’ advancement in digitalisation and their experience in restorative practices, one can conclude that what plays a crucial role is the readiness to go digital and not so much service experience per se. Finland and Estonia—digitally more experienced—were the quickest to try out digital restorative services, albeit being wary of the effects on the integrity of the co-production process. In Estonia, changes were made in the mediation process itself, moving away from the classical restorative questionsFootnote 5 used in mediation towards a circle format restorative conversation whereby more people would be involved in the discussion, not simply the victim and the offender (EST SM). As was explained during the interviews, they started using this modified format with cases forwarded by the police involving juveniles not adhering to social distancing rules. According to the interviewees, Ireland with its longer and more varied experience in restorative practices was less eager to go digital. Where longer restorative justice experience comes into play is with regard to the approach to the co-production process, i.e. when Finland and Ireland wanted to introduce standardised training, a digital mediation process and solve security issues before going online, then Estonia opted for learning by doing to ensure service continuity. This could be linked to having only comparatively recently launched restorative practices, and thus, adding an online component did not seem like a big change. Where the interplay between going digital and having restorative practice experience becomes clearer still is in the case of Portugal. There, having only short-term experience and comparatively more modest public service digitalisation record, the question of going digital remained only an idea, although the possibility was introduced in legal regulations.

In conclusion, although the mediation process did not seemingly change, there was an intuitive change in the co-production process, which meant a more directive role for the mediators and more responsibility for the citizens in organising the service context, i.e. ensuring a safe and quiet place to talk and the necessary equipment to attend the online meetings. In addition, the process became more business-like, which put some integral aspects of co-producing justice at risk, such as trust-building and feeling connected. In addition, in Estonia, the co-production process itself was modified and new methods were introduced.

Normalisation of Online Practice and Safeguarding the Principles of Co-production

The following sections will answer the second research question— how were the digitally mediated practices normalised while safeguarding the fundamentally co-productive nature of restorative justice—by highlighting the activities implemented to support the abrupt digitalisation of the co-production process that is inherently dependent on human-to-human contact and active interaction. The NPT will be used to showcase the embedding of new activities. It will be shown that the fundamentally co-productive nature of restorative justice was safeguarded when going digital, but a lot more responsibility was shifted on the citizens in order to create a safe environment for the mediation meeting. Although face-to-face meetings resumed once restrictions were lifted, restorative practitioners predict that digital alternatives—also in restorative practice—will remain an option that some citizens might prefer over face-to-face meetings even though it could reduce the quality of a restorative process. This is why special guidelines and more regular feedback were introduced into the co-production process.

I Coherence—Defining the Gist of Digital Restorative Services

Coherence refers to a set of ideas about the meaning, uses and utility of a new practice that is made possible, shared and enacted by socially defined and organised competencies (May & Finch, 2009: 542). For a new practice to persist, it needs to be given meaning and the meaning has to be understood. Taking the digital leap for restorative practices required a shift in thinking from facilitators and citizens alike. For example, the interviewees noted the facilitators had to be more explicit in guiding the conversation, which caused them stress; they were anxious whether the devices and connections would work and what happens if technical failures take place when someone has just shared something meaningful and deep (FIN2).

Restorative meetings inevitably had to be more directive; the facilitator took a much more active role then they might if they’re face-to-face. That’s really hard for RJ people to get their heads around because so much of their training is about trying to take the backseat as a practitioner but maybe this has to be different. (IRL)

The facilitators had to redefine the way they prepare citizens for mediation, focusing a lot on technical capacities and how it would feel to go through mediation online (FIN2). Independently of their country or digital platform, everyone interviewed agreed that having mediation fully online, instead of a blended model where some are present and others online, is better for the quality of the meeting. In relation to technicalities, signing mediation agreements became cumbersome when electronic signatures were not possible. In Estonia, every citizen has an ID card enabling electronic signing. As was said in one of the Finnish interviews, the mediator would print the agreement and send it for a signature round via regular post (FIN2). It takes a long time to get signatures from everyone, resulting in a less than optimal mediation outcome.

All the interviewees stressed that restorative practices have to be implemented in a safe environment to enable desired results, i.e. repairing harm and taking responsibility for your actions. This, according to the interviews, is more difficult online. Similarly with ensuring confidentiality of the meetings:

It’s difficult to ensure people’s safety during a meeting if you’re behind a screen and not in the same space. It’s impossible for you to intervene, and there’s a higher risk of re-victimisation than we would have in physical meetings. (EST)

Mediation depends on trust and confidentiality and if someone is secretly recording the meeting or taking pictures, we don’t really have a way of controlling that. In the beginning of the meeting, we need to remind everyone that you cannot record the meeting or take pictures. It’s maybe not the best start to say straight out loud that this would be a crime but it is important to strongly emphasise that this is not allowed. (FIN1)

Based on the interviews, online mediation in cases of domestic violence had additional stress factors for the mediators. Especially when parties were sharing a household, it would not enable the mediator to ensure that the victim can feel safe during the discussion (EST SM). As voiced by an interviewee,

How do we as mediators know that these people aren’t being pressured on the other side of the screen? (EST)

In traditional circumstances, unlike online, one could sense the change in temperature or energies in the room and if necessary, stop the mediation meeting, noted an interviewee (EST SM). For those reasons, in Finland, for example, online mediation is not possible for domestic violence cases (FIN2). It was also said that body language is an important tool to understand the situation and if need be, have a small break when one party is clearly uncomfortable or nervous (GROUP1; FIN3). On the other hand,

It’s more difficult to talk over each other in an online meeting and people are more polite during online meetings and they wait until the other has finished speaking. People are kinder and waiting for the turn because you can’t honestly understand anything if everyone spoke at the same time. (FIN2)

Interviewees noted that it is more difficult to get online meetings under control and mediators have felt powerless in situations where citizens lose control over their emotions (FIN2):

With online mediation, what’s scary is that once it goes off the rails, there’s no way of getting it back under control. There you really should have closeness, presence. (FIN1)

It is more difficult to follow the energies in the meeting, play with meaningful pauses and make sure that everyone feels safe and supported, the interviewees said. This can become especially difficult when someone has bad internet connection (EST SM). In real-life meetings, silence is normal, whereas in online meetings, silence could also mean that someone has lost their internet connection (GROUP4).

II Cognitive Participation—Initiating Digital Restorative Practices

Cognitive participation helps to frame a practice by enrolling and engaging people in collective action. This means initiating a new practice, people participating in it (enrolment), buying into new ways of working (legitimation) and lastly operationalising the new practice (activation) (May & Finch, 2009: 543–544). Initiating digital restorative practices started with mock processes (Estonia), pilot phases (Finland) or thorough exchange of practices before launching digital mediation (Ireland). In Portugal, a project was launched for a 4-step restorative prisoner re-integration programme, which had a digital component because of restrictions caused by the pandemic. Enrolment to the new practice was supported through guidelines, training and mentoring. For example, a list of preparation instructions was developed for the citizens: find a peaceful place to talk, make sure the device has enough battery to last through the meeting, etc. (FIN1). In addition, a short guideline was developed for the mediators, outlining the service process, including how to make sure citizens are comfortable with having an online meeting and that they participate voluntarily (FIN2). Mediators started mentoring citizens with regard to the technical side of digital mediation. A lot of the process refining took place as learning by doing. There were constant attempts to make mediation as client-friendly as possible, but this also induced discussions about how far to go with customisation.

Sometimes clients wanted to contact via FB Messenger, they wanted to have the mediation there and this is where we drew the line. This would’ve meant that you would’ve had to use your personal account and that’s not allowed. (EST)

Initial training for staff and citizens, buying the necessary equipment to improve the quality of the sound and drafting process guidelines supported the activation of online mediation. However, there is a need to legitimate the process so everyone gets accustomed to it, enabling more focus on the content of the conversations and less on the technical side of things.

This isn’t going to go away now. I hope this isn’t going to be a significant part of what we do but we’re going to the future. We can accommodate people who may want to work remotely with us but we’ll have to put in a framework and structure for this, proper cause and guidance. (IRL)

III Collective Action—Organising the New Practice

Embedding a new practice involves collective purposive action aimed at a goal and it is dependent on the work that defines and operationalises it (May & Finch, 2009: 544). For a new practice to endure, people should have the necessary skillset and the practice should be integrated in the existing, surrounding context. With the urgent need to keep restorative practices running in lockdown, it became obvious that more structure is needed in the co-production process. In Finland, all mediation staff were trained on how to organise online mediation and issues related to data protection, e.g. how different programmes work, which data they collect, how foreign cloud services work and which risks are affiliated with them (FIN2). The mediators consulted data security experts and based on their advice, drafted standard operation procedures that cover everything from the initiation phase up to writing and signing the mediation agreements online (GROUP2). It became necessary for the mediators to be familiarised with the app they were using so they could give guidance to the participants if necessary. They reworked the practice of working in pairs and had the mediators sit in the same room because it is important for the facilitators to read each other’s body language while leading the conversation. As put by an interviewee,

When one of us gets jammed, the other one can take over, but things would be extremely difficult to observe when we’re both like small pictures on the computer screen. (FIN1)

In Estonia, there was no special training developed for mediators; it was more about learning by doing and having one-on-one instructions prior to facilitating a session. It was more important to have ‘some sort of a service up and running and developing full training programmes later on’ (EST SM). Although this lack of proper training can be seen as a negative, Finnish colleagues commented on it positively: ‘Estonians are always so quick to try out new solutions, whereas we need to have a full system available and tested before we can launch anything new.’ (FIN1). Estonia provided continuous support for the mediators helping them set up Skype meetings and having feedback sessions afterwards to empower them for new cases (EST SM). There was a lot of mediator discretion. For example, once the weather got better in spring, they exchanged screen-mediated meetings for outdoor mediation sessions in the park while keeping a safe distance between participants (EST). In Ireland, a lot of work went into learning from other jurisdictions and colleagues from abroad:

We had no documentation or referencing to go from. We were getting support and information from other jurisdictions and talking to colleagues from abroad but when you’re doing things on your own, you’re learning every day, there were operational, logistical and practice issues that we were trying to record and learn from. (IRL)

There were many regular exchanges of practice also inside the team to ensure a smoothly running process. One of the recurring themes in all the interviews was keeping online mediation as linked to other processes as possible and maintaining its similarities to the analogue version. This helped to maintain the integrity of the co-production process. For example, the restorative programme created for prisoners in Portugal would include a possibility to carry out some parts of the programme online, but that would not change the content of the activities, only the channel (POR). As restorative practices are often part of criminal proceedings, it is necessary that the police, prosecutors and courts accept the agreements reached by the parties attending online mediation. That led to service providers in Finland organising co-vision and reflection sessions ensuring a standardised, yet flexible co-production process that took into account citizens’ needs, capacities and competences, but on the other hand, produced results that the authorities accepted.

IV Reflexive Monitoring—Keep Developing the Process Until It Works

Reflexive monitoring stands for communal and individual appraisal of the practice and where necessary, practices can be reconfigured (May & Finch, 2009: 546). Although online restorative practices are a relatively new phenomenon, there is already continuous reflection on the practice. Some of the interviewees found it easier to organise online rather than regular meetings, because not only does it save travel time and make it easier to agree on the meeting date and time, but also because it was easier to lead the conversation and because participants did not interrupt each other as often. In contrast, it was said that:

Online mediation meetings seem easier because we miss out on things we would notice in face-to-face meetings. We don’t notice things so we don’t have to take them into account but it’s really important to notice. You get a false impression that the session went really well. (EST)

In addition, there were doubts about the suitability of online mediation for every case.

If it radically speeds up the process or the people live in really distant places then of course, online mediation helps. The more sensitive the case, the less I would even consider online mediation. (FIN1)

One of the most common themes where online mediation was seen as questionable were cases of domestic violence, where the balance of power between the parties makes mediation difficult even in regular circumstances. In an online environment, it would be doubly difficult for the mediator to make sure the situation is safe for the victim (EST; FIN2). Since each case is different and oftentimes complex, the mediation process cannot be fully standardised, and hence, the mediator’s discretion becomes important. In some cases, the mediators had pre-meetings online and based on the impressions they got, decided to postpone mediation until it was possible to meet in person again.

From the citizens’ side, feedback on online mediation was good, according to the mediators. For example, people feel safer during mediation when they are participating from home, especially young people (FIN3). However, mediators were conscious that people who voluntarily opt for online mediation are inclined to like it (FIN2). The mediators were worried that perhaps online meetings do not have the same effect on offenders as regular meetings; they do not nurture the feeling of regret, guilt or responsibility. These are important from the viewpoint of a successful restorative process. However, as was mentioned in one of the interviews, it is:

Important to acknowledge that we’re never going to replicate in-person meetings, so the question becomes more like ‘what can we achieve online?’. Can we get people to feel like they’re heard? (IRL)

Lastly, it was said throughout the interviews that off-the-shelf applications available today have not been developed for facilitating restorative processes and there is a market for applications developed for mediation. Ideally, these could enable the mediator to pre-set the meeting space as a facilitator so the process itself ran smoothly once the clients are in.

If you think how the whole mediation community in the world is struggling with this it would make sense to develop an app that suits our needs. (FIN1)

People are desperate to facilitate good interaction right now, there’s a lot of money in that and a lot of competition. It’s a question of what we do with that technology, how we re-imagine the interaction. (IRL)

One of the concerns regarding user-friendly online restorative practices was that people opt for the easiest thing and if it is easiest to participate remotely that might become the default option. However, the quality of restorative justice would be depleted and high-quality online options were seen as going in the wrong direction. As a step further, algorithm-based options were discussed whereby the mediator could be replaced by a bot. As the mediator is an impartial party in the process, this option is not wholly impossible. However, a mediation process has so many nuances one cannot simply replace the mediator with an algorithm that proposes articles to the agreement based on answers to certain questions during mediation. There are too many details for a ‘robot’ to detect. The practitioners agreed that a mediator has to be neutral, but he/she also has to notice the moments where the victim is struggling, does not understand the questions asked or there is a risk of re-victimisation—‘a real mediator could stop the process but a robot would not notice these slight changes.’ (EST).

Discussion

As was shown above, in a rather short timeframe restorative practices went online due to the need to keep services going, not from a desire to increase efficiency and/or effectiveness (see also Lindgren et al., 2019). In accordance with Kuipers et al. (2019), the uncertainty brought about by COVID-19 increased the complexity of restorative practices. The case proved, as has been shown by previous research (Gelderman et al., 2011; Madsen & Kræmmergaard, 2016; Taherdoost, 2018), that digitalisation has its benefits like economic savings from not travelling or higher productivity because it enables to organise more mediations per day. However, no noticeable increase in effectiveness was seen in this case. On the contrary, practitioners mentioned depletion in the quality of the co-production process.

It was also shown that co-producing restorative justice was more vulnerable to breakdown of communication due to technical constraints and lessened possibilities for exchanging nonverbal symbolic gestures online. Setting up digital restorative practices required an open discussion about the meaningful qualities of restorative justice. For example, active and voluntary participation and a trustful and safe environment are crucial for mediation to bring about results (Bolitho, 2017; Chapman & Törzs, 2018), but creating and maintaining those circumstances in a digital setting is challenging. They were ensured through a more active and directive role of the mediators compared to normal circumstances where the facilitator’s role is more laid-back (Bolitho & Bruce, 2017). Much of the facilitation relies on body language, meaningful silence and eye contact (Chapman, 2016), but these become somewhat obsolete in a digital setting. For example, a ‘meaningful silence’ online might be caused by glitches in internet connection, rather than a person contemplating. This vulnerability to breakdown of communications and not being able to communicate in nonverbal symbolic gestures has also been demonstrated in previous research on digitalisation (Carter & Asencio, 2019; Sumathipala, 2020). This also meant that some aspects of mediation had to be redefined, e.g. the increased role of citizens in creating a safe environment for themselves in the restorative process. In Estonia, the mediation process itself changed from victim–offender mediation towards a circle format restorative conversation.

In summary, the answer to the first research question is that there were changes in the roles that mediators and citizens play in the co-production process, as the mediators had to be more directive than is recommended in restorative practices and citizens on their part were much more responsible for creating the needed environment for restorative justice to take place. As has been said before, customer capabilities and perceived self- and provider efficacy are key for successful co-production (Spanjol et al., 2015) and the abrupt change to digitalisation put these to the test because the technological solutions were not familiar to many. In addition, the digital channel significantly depleted the quality of conversations. This led the practitioners to agree that face-to-face meetings are preferred over online mediation. For the quality of the co-production process in terms of having a safe environment for honest discussions and being able to speak and listen without interruptions, face-to-face mediation was deemed infinitely better. However, there should be an option to meet online when people are not able or willing to travel, provided the mediators carefully consider the specific case and assess the risks and benefits of a digital mediation process, including the possibility of depleted quality.

When looking at how the digitally mediated practices were normalised while safeguarding the fundamentally co-productive nature of restorative justice, one could notice that there was hesitation in going digital because restorative practices require active communication and close interactions. For each country, the process started with creating a common understanding, goals and meaning of digital restorative justice. In the words of NPT, a new practice is made possible by “a set of ideas about its meaning, uses, and utility; and by socially defined and organised competencies” (May & Finch, 2009: 542). That is, through creating coherence. With the new practices introduced with digitalisation, the facilitators attributed an identity to the digital version of mediation as something that is not as good as face-to-face interaction, but something that is needed to be able to sustain restorative practices during the pandemic. Making sense of digital restorative practices required defining crucial aspects of digital mediation and drawing comparisons with the ‘normal’ process. This helped to make sure that the essence of restorative justice remains the same even in digital form.

Bringing the meaningful understanding of digital mediation into action meant a sudden and noticeably flexible service initiation process. In Estonia and Finland, the uptake of digital channels was swift compared to Ireland. In Portugal, due to restorative services implementation having been weak prior to the pandemic, digital restorative services remained only an idea. In practice, the launch of digital restorative justice in Estonia, Finland and Ireland meant supported enrolment for both mediators and citizens; they all had to engage in the new practice. Legitimation for the new practice was supported by agreed-upon guidelines. What was additionally important was advocacy for service continuation—alternative option being the suspension of the highly needed services—and organising technical equipment, training and mentoring. With regard to advocacy, not only was it important to convince the referring agencies in case study countries, e.g. the police and prosecution, that digital restorative practices produce outcomes, but also the European group of restorative practitioners became a forum that allowed an honest debate over the usefulness and risks of going digital with restorative justice. The need to keep the core elements of restorative justice alive also in digital form was considered crucial.

Digitalisation faced reluctance from all sides. For the facilitators, hesitancy was caused by the fear that the essence of the restorative process might be lost when there is a screen separating people. To accommodate this, the facilitators’ role in the meetings became more directive to enable a good flow of online discussions. Restorative practitioners also voiced their discomfort and a feeling of insecurity with getting the technology to work properly and feeling helpless in situations where the citizens did not have adequate internet access. The feedback from practitioners additionally showed that online restorative practices are not ideal, as they do not enable reading emotions nor ensure the same feeling of security as do face-to-face meetings. As these are important components in co-producing restorative outcomes, one could argue that this is a real problem in highly sensitive or complex matters, such as domestic violence cases, which is a hidden and physical social problem. From the citizens’ side, they sometimes lacked the necessary equipment and good internet connection to co-produce restorative justice, thus hindering the outcomes of the process. This phase of contemplation and developing practical means for digitalising restorative justice is understood as cognitive participation in NPT: the initiation of a new practice, people enrolling in the new practice, legitimating it and leading to the activation of the new practice, including providing the necessary resources enabling its implementation (May & Finch, 2009: 543).

In order to roll out the digital practices, collective action was needed. This was supported by the social context (the actual need for restorative practices) and group processes (learning within and between organisations, jurisdictions and internationally). Collective action through goal orientation became the mechanism used to safeguard the underlying principles of restorative justice. Being able to try out new solutions in the existing framework, being open to learning from mistakes, sharing best practices and setting up guidelines and training programmes for mediators were behind the successful launch of the new practices. Each country in their own way showed how the existing skillset was used and further developed and how the new practices were integrated to the existing co-production framework. As May and Finch (2009: 544) argue, collective purposive action aimed at some goal can embed a new practice by reshaping people’s behaviours and reorganising relationships and contexts.

Lastly, new digital practices were regularly reflected upon during co-vision sessions with practitioners, but also internationally, through the online meetings between European practitioners. In NPT, reflexive monitoring refers to formal patterns of monitoring the normative elements of implementation. “These frame how things ought to be, rather than the conventions that frame how things are worked out in practice. The shift from explicit to tacit appraisal by participants is an important signal of the routine embedding of a practice.” (May & Finch, 2009: 545) The constant reflection and apprehension about going digital in all studied cases (except Portugal, where digital restorative justice did not actualise) was a natural sign of embedding a new practice. Learning from each other enabled to develop online mediation more quickly and embed digitalisation in the overall co-production process. Honest discussions about possible changes to the co-production process due to digitalisation helped to pinpoint the risks to the essence of restorative justice.

In reflection, concerns were voiced that online restorative practices should not become the new norm, even though they are more convenient, cost less and help to save everyone’s time. It was discussed whether the facilitator—a neutral person asking questions and guiding the conversation—could be replaced by a bot and concluded that the role of a good facilitator is to understand the balance of powers between participants and no app can do that. In the end, a bot lacks the human touch that is considered necessary in any restorative process. In short, restorative practitioners in all case study countries prefer face-to-face interaction because it helps to better safeguard the intrinsic nature of the restorative justice co-production process. However, when face-to-face meetings are impossible, the online version is an option that should be used when all risks have been assessed and the mediator can be sure that no re-victimisation can take place during the restorative process.

Conclusion

The article looked at the immediate effects of digitalisation on service co-production, the mechanisms used to safeguard the underlying principles of restorative practices and how new practices were embedded in the existing co-production process. It was shown that the roles of service providers and citizens changed in the co-production process. The former took a more directive role in mediating and the latter became responsible for creating a supporting service environment. The main mechanisms used in launching digital mediation were learning-by-doing, active exchange of best practice, and integrating digitalisation with existing service processes, which helped to embed the new format alongside the old one. One thing the case study showed was that the launch of digital restorative practices depends most of all on the readiness to try digital solutions and somewhat less on prior service experience. Where the latter comes into play is with setting up digital co-production, i.e. with older, established services, there is a need to introduce some sort of standardisation in mediation preparation, delivery and feedback mechanisms. Where co-production had less rooted processes, like in Estonia, people were more apt to try new solutions and learn by doing. Based on the opinions of the interviewed mediators, face-to-face interaction is, regardless to the abundant digital opportunities, still the norm to strive for. However, what the case taught us is that setting up work practices that aim to keep alive the inherent nature of the co-production process of highly sensitive services, even though there is depletion in interaction quality, is worthwhile if the other alternative is no services at all.

As the current study focused on the immediate effects, it would be interesting to see in future research that if time was dedicated to more meticulous process development—ideally co-created with citizens to match their needs—would it be possible to develop a digital restorative practice that could really be a viable alternative to face-to-face meetings. Due to COVID restrictions, sensitivity of the service (especially where service co-producers are crime victims) and the scope of this article, it was not possible to interview citizens during data collection. However, this could further strengthen the understanding of going digital with highly sensitive and co-produced services.