Digital interface design and power: Friction,threshold,transition

Introduction

Before…[Cash Flash]…was around you had to go to a bank and everything was really complicated…[T]here wasn’t this kind of immediate response, immediate feedback… So the fact that it’s online…you can literally select to the pound how much you want…We put all the controls for the loan at your fingertips so you decide how much you want and how long you need it for and then we tell you how much that’s going to be… it was just a bit of a revolution.

While the designer in the above quote suggests that accessing HCSTC via digital interfaces and devices works to put ‘control…at your fingertips’, which marks a positive development, ushering in an era of increased speed and freedom for borrowers, others have pointed to the harm debt can create, including unsustainable practices of borrowing (CMA, 2014) and mental health issues (Fitch et al., 2011). Accessing HCSTC by digital interfaces can thus be understood as part of broader processes of global finance and particularly fringe finance. In Aitken’s (2006: 480) words:

Fringe finance includes a set of credit practices provided to those populations that exist on the margins of the mainstream financial world. Fringe financial practices typically include cheque-cashing services, payday loan providers, pawnshops, rent-to-own retailers, auto title loan services (credit secured by auto title), tax refund anticipation loans and other services delivered to the ‘unbanked’ or financially excluded.

Recognising the socially and economically damaging potential of HCSTC, theorising digital interfaces in relation to control allows us to better understand how people become implicated in fringe finance systems. By focusing on the specificity of how interface interactions are designed and experienced, the paper deepens and complicates an account of self-governance and coercion in relation to social issues around debt. To do this, the paper draws upon interviews with HCSTC website designers and HCSTC users, ¹ focusing on the techniques utilised by HCSTC designers to guide the user through the process of applying for credit, and exploring how people respond to these techniques.

Parallel to the social and political implications of better understanding digital experiences of HCSTC (in order to argue for new forms of regulation around digital access to credit for example), we use the case of HCSTC design to interrogate accounts of digital power that have become prominent in digital geographies (Ash et al., 2018) and associated fields such as sociology, science and technology and media and cultural studies beyond discussions of debt or fringe finance. Whilst this work has done much to understand how power operates in the background of digital systems, through algorithms and search engines (Amoore and Piotukh, 2016; Cheney-Lippold, 2011; Deville and Velden, 2016; Mackenzie and Vurdubakis, 2011; Pasquinelli, 2009; Striphas, 2015) and interfaces (Ash, 2015; Degen et al., 2017; Rose et al., 2014; Seigworth, 2016; Wilson, 2014), little work to date has focused on how power operates, and is enabled, through actual practices of interface design. As work on the sociology of design (Broth, 2008; Farías and Wilkie, 2016; Mackenzie, 2006; Marenko, 2015) suggests, the risk is that, in ignoring practices of design, assumptions about the smooth manipulation of user action and experience creep back into analysis, reproducing conventional stories about the location and form of power in a digitalising world.

Rather than designers possessing power, which they then exert by creating systems of modulation or manipulation in relation to passive unknowing users, we argue that interface design is an experimental process of managing friction. By friction, we mean a series of bodily and technical obstacles or hesitancies that interrupt, slow or stop a user from completing a task within a digital interface, such as choosing a service or buying a product. Friction is a matter of attempting to not only produce smooth experiences, although that is part of what is being done, but of producing the possibility of a transition between thresholds at the right time. For now, ‘thresholds’ can be understood as key points in an interface that need to be passed through. In digital interfaces these thresholds include, for example, decisions over data entry, purchase application, and so on. By ‘transition’, we mean a moment in which a more or less provisional justification that overcomes or quiets worries, fears or other affective states is made, which allows people to move beyond the thresholds and frictions they encounter in the interface.

Distinct from the smoothness of constant modulation (Deleuze, 1992), power in relation to interface design is, we argue, a matter of producing continuities and discontinuities that work together. Interface design is not about necessarily always trying to reduce friction. ² Rather, as we show, friction can also be productively introduced to help achieve the completion of a task, as long as this friction is carefully managed (much like the ‘planned’ frictions of logistics, see Gregson et al., 2017). In the practices of data-driven design that are central to the making of interfaces, this process of managing frictions and enabling transitions is aided by the use of data analytics. These analytics act as catalysts for experimentation with what works in interface design. Here power, for both the designer and user, is ‘exerted in the drawing together of once incompatible states into compatibility’, with the ‘agency of power work[ing]…alongside and within its subject’, rather than separate or implemented upon it (Hookway, 2014: 23). Our vocabulary of friction, thresholds and transition is offered, then, as a means of understanding how those ‘compatibilities’ between user and interface are (re)produced.

It is important to note that the techniques utilised in HCSTC website design are not unique to these sites. For instance, a whole range of digital interfaces, such as retail, rental and banking websites implement similar designs for their pages and input devices such as sliders and buttons and use analytics to hone and change their interface to drive user engagement. The vocabulary developed in the paper, thus allows us to attend to the precise forms and modalities of power coming into being in an increasingly digitally mediated world (see, for example, on control in a digital world Bogard, 2015; Franklin, 2015) and provide the ability to diagnose how this power operates and in turn better understand the harms it may cause.

The argument about interface design unfolds in four sections. In the next section, we explicitly theorise friction, threshold and transition as a means of understanding how interfaces are designed and condition action and experience without reproducing tropes of the manipulation of passive users. We connect this account to the emphasis on modulation as striation and transformation in work that has drawn on Deleuze’s (1992) arguments about ‘control’. The ‘Designing and testing thresholds’ section demonstrates various techniques designers utilise to attempt to manage the friction of both users and the sites they create by constructing thresholds at key points in the interface. Discussing the use of A/B and multi-variance testing in HCSTC design, this section shows how friction cannot be easily codified and questions the presumption that designers can simply or directly manipulate user actions. The ‘Transition and friction’ section explores how these design techniques are experienced in practice and how people transition through the thresholds of HCSTC sites. In conclusion we point to how the concepts of friction, threshold and transition can be used to understand power in relation to data-driven interface design in ways that refuse simple narratives of manipulation and complicate accounts of modulation.

Friction, threshold and transition

Much work has found resource in Deleuze’s (1992) theorisations of the emergence of ‘control societies’ or ‘societies of control’ to understand power in a digitally mediated world of interfaces, in particular his emphasis on how ‘modulation’ replaces the ‘mouldings’ of disciplinary forms of power. Rather than imposing a form across diverse circumstances, as in disciplinary power, modulations work on and through movements ‘like a self-transmuting moulding, continually changing from one moment to the next’ (Deleuze, 1992: 179–180). Interfaces would, on this understanding, enable the ‘open circulations’ that, for Deleuze, replace the closed milieus of disciplinary power and society. Leaving to one side questions of the relation between discipline and control and the risks of characterising a society according to one mode of power, it is easy to understand how the emphasis on flexibility and continuous change we find in Deleuze’s account of modulation appears to fit with processes of digital mediation through interfaces, especially given the importance Deleuze affords to the digital in the transition from disciplinary power (Best, 2010; Bogard, 2015; Bratich, 2006; Cheney-Lippold, 2011; Chun, 2006; Franklin, 2015; Lazzarato, 2006; Parisi, 2012). If interfaces are understood as ‘autonomous zones of activities’ and ‘processes that effect a result’ (Galloway, 2013: 8), control in Deleuze’s terms would be the key interface effect and modulation the name for the process that achieves it.

But, whilst it characterises a process, invoking modulation can assume a lot about those who are subjected to it. For instance, Cheney-Lippold (2011: 165) argues that web analytics enable a situation in which ‘categories of identity are being inferred upon individuals based on their web use…[and]…[c]ode and algorithm are the engines behind such inference’. In this case, modulation

marks a continuous control over society that speaks to individuals in a sort of coded language, of creating not individuals but endlessly sub-dividable ‘dividuals’…[which]…become the axiom of control, the recipients through which power flows as subjectivity takes a deconstructed dive into the digital era. (Cheney-Lippold, 2011: 169)

Linked to this point, accounts of control and modulation can assume that those who design such systems have complete control over what they do, how they work and can anticipate what their effects will be. For example, Zwick and Denegri Nott (2009: 224) position digital marketers who construct and utilise databases to identify markets for clients as architects who ‘manufacture…customers as valuable information commodities’. In Zwick and Nott’s narrative, the failures, hesitancies and discontinuities that are no doubt involved in digital marketing are not discussed, which reinforces an account of modulation as a matter of smoothness, flow and consistency. Ignoring the specificity of practices with their attendant failures, gaps and contestations can lead to a type of analysis that either dissolves power into the networks, assemblages or ecologies that enable relations of control (such as the database the marketers construct and use) or, often implicitly, re-centres power in the designers of manipulative algorithms or interfaces or systems (such as the software and IT engineers who manage these databases).

Distinct from these accounts, we utilise a vocabulary of friction, threshold and transition to understand modulation as a double movement or management of smoothness and striation that is fragile, continual and ongoing. As Raunig (2016) argues, modulation operates through, on the one hand, the striation of distinct modules and, on the other, constant, permanent modifications and transformations across temporary striations. In this analysis, an interface modulates by creating, albeit temporarily, relations with the user that enable distinct striations and modifications.

Developing this point, we suggest that modulation is a matter of managing the interplay of various bodily and technical frictions that emerge whenever someone or something engages with an interface. Formally stated, frictions can be defined as practical, affective and emotional contestations between bodies and interfaces. Frictions are not something that can be simply overcome or put to one side, but are a necessary part of any engagement with an interface. In Rose’s (2016: 343) words: ‘friction is inherent to interfaces’. Frictions might be blocks or obstacles that interrupt, slow or stop a user from completing a task within an interface, such as a password screen. But frictions can also act as sites of grip, encouraging someone to continue using or engaging with that interface because of the contestation faced by the user, such as when learning to play a videogame. In either case, friction is a key currency of digital power as modulation.

Frictions are powerful because they can activate what we term the infra-individual tendencies of user’s bodies. Following Massumi (2014: 8–9), we can understand infra-individual tendencies as a range of chemical, affective, sensory, motor and memory forces that compete with one another and contribute to how a person thinks, feels and ultimately decides how to act in a situation. For instance, we could understand deciding whether to eat something, not as the outcome of a process of rational contemplation, but as a competition between the rumbling of our stomach, the smell of food in the environment, the colour or texture of food packaging, memories of when we last ate and the promises we might have made to ourselves to reduce our calorie intake in order to lose weight. In relation to interfaces, on an infra-individual level, friction can be invoked by pretty much anything, such as an off putting colour of text, an unpleasant sound effect or a feeling that an image looks unprofessional. Although, as we shall see, some frictions are designed into interfaces, in the main frictions are interruptions that delay, stop, halt or defer some kind of desired interface effect – such as, in our case of HCSTC, the making of an online application for credit.

Whilst frictions may emerge around any aspect of an interface, they become most apparent and pronounced at specific thresholds in interfaces, where a user has to make a commitment to something. By ‘threshold’, we refer to a necessary moment or point in interfaces that a designer needs to encourage the user to cross or move beyond. Always involving some kind of movement, thresholds are occasions of discontinuity within an interface. A threshold is thus what Hookway (2014: 23) terms a ‘liminal condition’, which both ‘brings about a kind of order, stability or balance’ but is ‘also continuously open to the possibilities of disorder, instability, immanence, chaos, and turbulence’. In HCSTC interface design, as we shall see, there are several thresholds within any one interface such as sliders which project the repayment costs of credit that vary according to amount borrowed and loan duration, and application forms. Each of the thresholds within the interface contributes to enabling the user to move towards the key threshold of the ‘submit application’ button.

In our terms, designers are faced with the challenge of managing the frictions that can intensify around the key thresholds that constitute an interface. Or, more precisely, for desired interface effects to happen some kinds of frictions must be minimised (and some enabled) by creating what we call ‘transitions’. By transition we mean a moment or point at which competing infra-individual tendencies within a person are at least provisionally resolved. A transition might feel like a decision and might result in a decision being made, but this is quite different from an account of decision making in which a ‘decision is a singular moment abstracted from the context within which it takes place’ (McCormack and Schwanen, 2011: 2810). Rather, the idea of transition points to a notion of decision that is closer to what McCormack and Schwanen (2011: 2808) understand as space–times that ‘register as transformative events felt through the displacements they generate in bodies’. In relation to interface design, the thresholds of the interface are designed to enable transitions that quiet worries, fears or other affective states that might impede or forestall people crossing the various thresholds that result in an application for credit to be made. What is crucial is that the transition between tendencies does not have to be absolute or irreversible. As long as the tendencies are resolved for just long enough for the user to apply, then the interface designer has successfully managed the friction of the user. While resolving infra-individual tendencies in a user is not always enabled by the interface alone, the interface itself often plays an important role in resolving the infra-individual tendencies of users and enables action to continue. In the case of HCSTC, the continuation of action might be the completion and submission of an application for credit.

In the example of HCSTC, these frictions, thresholds and transitions can be analysed at three sites. First there are various infra-individual tendencies and transitions at play in the user’s body before they pick up their device and access a HCSTC website. Users have some idea that they want to borrow money, usually have a specific reason for doing so and a range of concerns, worries, fears and hopes attached to what the money can do, if their application is successful. Indeed, in some cases, users might already be resolved to borrow long before they begin applying for credit online as they are convinced they need the money and will apply regardless of the cost.

Second, there is a friction inherent in the interface itself, where the designer attempts to organise the relationship between different units in the interface to create a threshold in order to help the user generate the necessary transitions that enable an application. By unit, we mean a modular piece of an interface, such as an image, text box or button. As we argue elsewhere (Ash et al., 2018), these units are experienced synaesthetically across multiple senses, which is how they work to influence infra-individual tendencies in the body of users. As we shall see, the designer can attempt to manage frictions through the design and testing of units that operate as thresholds and the units that surround these thresholds, both through in house testing but also through real time feedback based on user analytics.

Third, there is the specific friction between the user and the interface at the time of application, where the user’s pre-existing infra-individual tendencies meet the frictions and thresholds of a specific interface. Experientially, this plays out through the expression of various bodily states, such as anticipation, stress, anxiety or excitement as users engage with the site and the friction of the thresholds themselves. Through particular design practices the interface can either help manage these frictions so as to enable a transition, which allows a user to cross the final threshold of the interface and apply for credit, or fail to manage these frictions, in which case the necessary transitions are not made and the threshold disappears. In this case, the user may hesitate to apply or not apply at all.

Focusing on the way frictions operate between bodies, software and hardware points to a model of modulation as a continuous interplay and management of different forces and tensions. This management is not always a case of overcoming or minimising friction, but working productively with it, in order to produce a transition or change in state (of a human user, or another object) that suits the designer or owner of that interface or technology.

Analysing friction as modulation requires an empirical focus on the specific form of relations between interface and user while recognising that the achievement of a desired change of state is always fragile and in doubt. To understand how friction is managed we turn to the particular design practices associated with HCSTC websites and interface design more generally before, in the following section, describing how users experience interface design. We organise both sections around the design and experience of the specific thresholds – sliders, buttons and forms – that constitute HCSTC interfaces.

Designing and testing thresholds

User experience designers are keen to manage the friction in their interfaces and help enable transitions that affectively pacify, allay and not unnecessarily delay those that use them. To do this, the designers we interviewed employed a number of techniques organised around the construction of thresholds. In HCSTC website design and website design more broadly, practices of designing and testing these thresholds go hand in hand. These practices range from the development of personas before the interface is designed, to the design of units of the site, such as buttons, sliders and images, to in house testing with users on prototypes of the interface, through to real time testing drawing upon analytics of actual users as they engage with the websites live. These practices are complex, so for reasons of space, in this section we can only focus on the design and testing of units in the interface and how these units were modified using analytics derived from real users as they visited these sites.

From the perspective of a user experience designer, the purpose of the interface design in HCSTC websites is to enable, coax and encourage the user to press, click or tap on the ‘submit application’ button. In this case, HCSTC websites are designed to guide the user through the stages of the application process with as minimal fuss as possible. Thresholds in the interface are designed around a logic of funnelling. The funnel is designed to encourage the user to fill in the necessary amount of information needed to make a loan application, while minimising what designers referred to as customer drop-off, where the user leaves the page and fails to complete the application. The more efficiently the user moves through this funnel, the more effective or successful the design of the interface is considered to be.

In order to minimise the number of drop-offs, designers optimise their websites using real time testing with actual users as they engage with the site. Utilising cookies on their websites and software such as Google Analytics, designers can monitor almost every aspect of user’s engagement with a site. These analytics measure everything from the length of time spent on a single screen, to the page at which a user leaves an application, to the geographical location of the visitor, to the operating system and browser that is being used to access the site. The designer can then use this information to alter aspects of the units of the site and then test to see how this alters engagement with these units.

Such tests take one of two forms: a/b testing and multi-variance testing. In a/b testing a certain percentage of users who access the site in question receive one version of the site and the other percentage receive the other version. Including a key design difference between each site, the designer can then use the site analytics to determine the effectiveness of a single element such as the colour of a button, by seeing if the difference introduced increases or decreases engagement. In multi-variance testing, a similar principle is used, except multiple differences are introduced and all the combinations of these differences are tested to understand which combination of units is most effective (such as two different headlines and two different images that sit alongside this headline). Beyond testing different designs, these sites are also responsive. This means that depending on a range of factors such as location or operating system, the site will present content differently to the user in order to further optimise their movement through the funnel.

Having briefly described the forms of testing employed by HCSTC designers, we can now explore how this testing influences the design of thresholds in these sites. In HCSTC websites we can identify at least three thresholds, which are organised around or linked to particular units: the slider threshold, the apply now threshold and the application forms threshold. The number and order of these thresholds are similar across a range of HCSTC websites, although there are some important differences.

Transition and friction

We could understand the process of designing thresholds to manage friction as one mechanism through which the doubleness of modulation is produced – frictionless thresholds allow for striations and for movements. Not only is friction an ever present problem for designers to attempt to manage, but as we have seen a relation of friction is also designed into units in order to enable the continuation of a desired relation with the user in another form. Designing interfaces is, then, best thought of as a means of producing the continuation of a relationship in order to achieve some kind of desired end. Now we have examined how friction is managed around certain thresholds on these sites, we can step back and consider how these frictions and thresholds are experienced by customers and how the process of transition did or did not take place. Again, we focus on the thresholds that must be crossed to enable the desired interface effect: the completion of an online application for HCSTC.

Rethinking digital interface design

This paper has examined how HCSTC interfaces are designed and experienced in the context of attempts to understand power in a digitalising world. Expanding on existing analysis of how interfaces modulate life, we have developed the concepts of friction, thresholds and transitions to understand the user–interface relations necessary for desired interface effects to happen – in our case submitting an online application for HCSTC. Designers seek to enable the transition between competing tendencies of users at key thresholds in the process leading to the completion of an online transaction. In relation to HCSTC, these thresholds were key moments in the application process that had to be completed to apply for a loan. The frictions were the various obstacles that caused hesitancy or blockages that stopped a user from applying but were necessary passage points to the application itself. The transitions were the points where infra-individual tendencies within a user were momentarily resolved, which enabled users to convince themselves to cross thresholds in the interface.

Focusing on the concepts of threshold and friction of the interface is not to say that other factors outside of the interface were unimportant. In some cases, users had already resolved to cross any threshold they were going to encounter in the interface regardless of the friction involved and were absolutely set on applying for a loan. With our respondents this was often because the loan was required for essential purposes such as to purchase food or pay bills. But in many cases, the designers did manage friction in ways that enabled thresholds to be crossed in spite of hesitation or doubt from the user. In this regard, the paper has suggested that social problems around accesssing credit digitally need to be understood in relation to the thresholds of the interfaces and websites of specific companies and the infra-individual tendencies of particular users, rather than just analysed by the type of credit these sites offer. More broadly, while Aitken (2006) suggests that fringe finance operates through a dual process of self-governmentality and coercion, the paper has suggested that accessing fringe finance such as HCSTC digitally cannot be understood as simple self-governance or coercion alone. Rather, it may be more useful to describe becoming indebted digitally in terms of active contestations between bodies and interfaces, where many of these contestations remain backgrounded and implicit to the users of these sites.

Beyond the example of HCSTC interfaces, this paper has provided a vocabulary for describing and analysing the operation of power in data-driven design practice in digital interfaces. As we have seen, the management of friction in interfaces is responsive. Using analytics and data from users, designers can optimise the thresholds in the interface and automatically customise the interface depending on when and where the user is visiting the site from in an attempt to manipulate friction in a way that maximises application rates. In other words, what users do with an interface folds back into the practices of design and testing. As such, power becomes an negotiation between designers and users of these products through processes of experimentation that are ongoing and subject to change.

Returning to Deleuzian inspired accounts of modulation through digital technologies (Best, 2010; Bogard, 2015; Bratich, 2006; Cheney-Lippold, 2011; Chun, 2006; Franklin, 2015; Lazzarato, 2006; Parisi, 2012), we can understand the management of friction as a kind of power that is related, but not equivalent, to control as first delineated by Deleuze (1992). Whereas systems of control are assumed to set limits that ‘confine and enable’ what people do, while allowing them the freedom to roam within that system, data-driven interface design involves constant attempts to manage a changing set of continuities and discontinuities between entities that are ultimately irreducible to their operation in that system. Control, here, is not smooth or continuous. Instead, it is about working with and attempting to produce thresholds and transitions in interfaces and managing the resultant friction of these thresholds and transitions.

Furthermore, rather than suggesting that power is becoming increasingly exteriorised into non-human, technical systems of algorithms and AI, this empirical work cautions against an account that overemphasises the autonomy of technical systems (or the interpretative creativity of users). Instead, in relation to interfaces, we suggest that it is crucial to focus on the ongoing exchanges between human and non-human objects as key to understanding new forms of power in relation to digital interfaces. This work thus offers an important rejoinder to accounts of control in digital systems, which sometimes too easily assume that power operates automatically and without human negotiation. With this point in mind, we hope that the concepts of friction, threshold and transition provide purchase for analysing other forms of digital interface beyond the example of HCSTC.