Analysis: Episode One

Before proceeding with an analysis of this episode, I would like to�rst call the reader’s attention to the materiality of Walt’s impressive arrangement of computational objects that both enabled and constituted remote pair programming, where a number of important features are discernible. These are rendered in�gure4.1.

Figure 4.1:Anonymised Image of Walt’s Remote Pair Programming Setup

7_{Lazy intitalisation or lazy loading, is a design pattern whereby data, perhaps being loaded from a database,}

is not all loaded at once. Instead, ‘a marker in the object structure [is created] so that if the data is needed it can be loaded only when it is used. As many people know, if you’re lazy about doing things you’ll win when it turns out you don’t need to do them at all’ (Fowler2012, p. 200).

Computational Object Description

Desktop computer The desktop computer is centrally positioned and acts as the primary locus of human a�ention (Guimbretière2002, pp. 15-19,2003, p. 53). It is where programming activities are projected and, in general, it is the rendering of this screen that is shared/mirrored with the remote party in the other location, fundamentally constituting pair programming.

iPad™ The iPad™ is used primarily for FaceTime™ and Google Hangouts™ to virtually connect the remote pair. It is positioned to the le�at about 11:00 and at eye level to facilitate ergonomic viewing. It is also mounted on an articulating arm so that it can be easily adjustedin situ.

Speakers Centrally positioned speakers are plugged into the iPad™ and therefore place the audio of the remote pair spatially inline with the desktop computer.

Unidirectional microphone A tripod-based unidirectional microphone is placed to the right of the speakers and is plugged into the iPad™ so that Walt can move freely within the workspace and be heard by the remote party in the other location. Laptop computer A laptop computer is positioned to the right of the desktop computer. This

computational object is generally not shared with the remote pair and provides Walt with a ‘private’ computational space.

iPhone™ While not rendered in figure4.1, another substantive part of Walt’s array of computational objects is an iPhone™ which he used to send and receive various kinds of messages while involved in the pair programming activity. For example, these can be signals to the other party either before or a�er a pairing session to coordinate specific details, such as communication channel.

Table 4.2:Array of Computational Objects Employed by Walt for Remote Pair Programming Figure4.1depicts a plethora of computational objects that have been assembled by Walt to facilitate remote pair programming. These are enumerated in table 4.2and document the extent of thought and e�ort given on the part of Walt to construct an envir- onment where remote pair programming functions as seamlessly as possible.8 _{Based on} his statements to me and various blog posts he authored on the topic, Walt seemed to take great pride in his work on the development of this sociotechnical frontier. It is through this arrangement of what Latour refers to as the ‘assemblage of human and non-human elements’ (1999, p. 159) or what Barad calls ‘material-discursive practice’ (2007, p. 178) that the phenomenon of remote pair programming—and any attendant leadership that might be observable in it—is actually produced.

Within the episode, evidence of leadership practice is present in a variety of forms throughout this engagement between Walt and Robert. It begins with Walt o�ering direc- tion (Van Velsor and O’Connor2007, p. 32; Uhl-Bien, Riggio et al.2014, p. 90), suggesting to Robert that they take the approach of refactoring and Robert’s tacit acceptance of this suggestion, aligning with the leadership dimension ofprocessas depicted in�gure2.4on

8_{This is not to mitigate the role of the endless legion of human and non-human actants who participated}

in the design, production, and distribution of these computational objects (Latour2010a, pp. 2-3); however, this legion is not the focus of the present study.

page32.

There is also an interesting dynamic that unfolds between Walt and Robert where Robert is withdrawn and uncommunicative, as evidenced by Robert not looking at the camera, his mumbling when asked what he thinks, and his failure to mention that the screen is not being shared between them. It is not until Walt realises he has not shared his screen that this fact materialises. Consequently, until that point, Robert has not been able to see what Walt has been doing and instead was only able to hear it by virtue of Walt’s talking. It is Walt, not Robert, who picks up on this detail of the state of the material- discursive practice.

To borrow Larsson and Lundholm’s terminology, Robert is exhibiting a ‘resistance to closure’ (2010, p. 1102), however, in my use of their term, I extend it to include engage- ments that go beyond talk. Speci�cally, I also include actions relating to computational objects, such as initiating builds, and the dyadic typing, mousing, and watching of com- putational object screens that occurs as part of pair programming. I argue, therefore, that Robert resists this closure in the material-discursive practice of pair programming through his various forms of nonverbal communication and/or inactivity that include not looking at the camera, not responding in full sentences, and not telling Walt that the screen was not being shared.

To be clear, I am not asserting beyond the shadow of any doubt that leadership lies within these practices; rather, I am pointing out that if one looks at the literature, there are numerous precedents to suggest that these practicescanbe interpreted as leadership.

And, as I will argue, it is the material-discursive practice ofcitingleadership that brings

leadership into existence within a member’s experience, just as my citing it within this dissertation brings it into yours. In this sense, talking of ‘leadership’ is always performat- ive and any ‘leadership’ that exists canonly do so because some person or entity evokes it in a way that is convincing or acceptable to the participant in that material-discursive practice. I elaborate further on this point in section4.4.

For the present moment, the point I want to make, according to Larsson and Lund- holm’s, one of the functions (and therefore ‘markers’ of) leadership is the overcoming of such resistance (2013, p. 1022). The preceding episode shows the dynamic of resistance going on for some time until just before 10:50 a.m. when Walt o�ers Robert control of the keyboard. Such negotiation of control is enmeshed within extant meanings of leadership (Alvesson and Spicer2012; Barker1993; Collinson2005; Hales1986).

Moreover, within the human pair programming dyad there are particular con- straints that derive from a practiced relationship with the computational objects, one of which is that only one person can type code at a time. In practice, this generally means one person holds the keyboard and ‘drives’, as it is commonly called by the practitioners I

observed. In this sense, it bears a close resemblance to conversational turn-taking (Sacks et al.1978). The person ‘driving’ is the person actively writing the code. In this and other instances of practice I observed, this took the form of the ‘driver’ talking out loud about what they were typing as they typed it.

Based on these various precedents, to say that these are leadership practices is within the realm of possibility by means of the articulation of what the driver’s think- ing or rationale is regarding what they were typing. In other words, by simply think- ing/talking out loud about how they planned to solve a particular software engineering problem, they were also engaging in the articulation of a vision or strategy towards some speci�c business goal (Guastello2007; Montgomery2012; O’Reilly et al.2010). This per- spective corresponds to the leadership dimensions ofpositionandprocessas depicted in

�gure2.4on page32.

Thus, when Walt o�ers the keyboard control to Robert, this appears to engage Robert at a higher level in that he demonstrates the active typing of code, the initiation of builds, and the utterance of fully articulated words. As with Larsson and Lundholm’s study, through ‘repeated a�rmative responses [...] agreement is gradually developed and the persuasion thus accomplished’ (2013, p. 1022).

The ‘persuasion’ here is to convince Robert to participate more fully in the material- discursive practice of pair programming. This is evident in Walt’s various o�ers and sup- port to Robert. Accordingly, I characterise the approach that Walt took here as ‘empower- ing leadership’, ‘creative process engagement’, resulting in an increased ‘employee cre- ativity’ (Zhang and Bartol2010, p. 117). This is a mode of leadership that aligns with the dimensions ofpurpose andpersonas depicted in �gure2.4 on page32. My claim is that it was evidently Walt’s aim to engage Robert in pair programming and the means by which he did so were through the capabilities of his personality—as enacted through computational objects.

Then, at 10:50 a.m., Walt takes over ‘driving’ without any discussion—Walt simply starts typing in what might be thought of as an ‘open space’ where Robert was not typing, analogous to jumping in at a conversational turn (Sacks et al.1978). This is a phenomenon I observed quite often between many pairs. This phenomenon has also been closely ob- served in terms of the negotiation of authority in Linde’s research on helicopter�ight crews (1988), however, I extend her argument here to make the claim that such negoti- ation can also constitute acts of leadership along the dimension ofpositionas depicted in �gure2.4on page32.

I make this claim because within this passing back and forth, theroleeach mem-

ber plays also changes. According to Beck (1999, p. 58), the ‘driver’ talks out loud, making audible their thinking process whilst writing code whereas the ‘observer’ o�ers strategic

feedback, which can be interpreted as a leadership practice along the dimensions ofposi- tionandprocessas depicted in�gure2.4on page32.

Moreover, the kinds of practices identi�able in this role of the ‘observer’, according to Hernandez et al., involve ‘strategic choices [...the] behavioural processes by which strategic leadership in�uences organizational outcomes’ where such leaders ‘can guide interactions among other organizational members and channel knowledge’ (2011, p. 1179). Such an approach is prefaced on the leadership dimension ofresultas depicted in�gure2.4

on page32.

Thus, the passing of keyboard control, I argue, provides markers for leadership practice within the material-discursive practice of pair programming across a number of dimensions depicted in�gure2.4on page32. First, when one party spontaneously ‘takes over’, they enact through theirpersonality, and with a speci�cpurpose, a change inprocess.

Second, in so doing, the material-discursive practice is dynamically restructured such that the roles are reversed and the person formerly ‘driving’ is now placed in a role where they enact what can be considered the leadership practice of o�ering strategic guidance. In this fashion, I argue that there is evidence for speci�c leadership practices that can be identi�ed with each role, suggesting leadership in the plural (Denis, Langley and Sergi

2012).

Yet within the dyad described in the episode above, I rarely observed Robert play- ing the role of observer as a strategic guide. Instead, what was more readily observed was Walt playing both of these roles in ways that were more closely identi�able to Beck’s description. Thus, when Walt o�ers commentary at close to 10:50 a.m. about limitations of the approach Robert is taking, he injects an alternative perspective in line with Beck’s vision of pair programming (1999, p. 58). And again, at approximately 10:53 a.m., Walt advises Robert while he is writing code that he might want to approach the writing in a particular fashion, providing strategic direction (O’Reilly et al.2010). In these engage- ments, while the roles shift dynamically, we see how particular roles are played, that align with the dimension ofposition, as depicted in�gure2.4on page32.

During this engagement, we see an increasingly�uid dynamic between the pair where control passes freely between the two of them and where tests are being consist- ently written and run as part of the material-discursive practice of pair programming. This kind of engagements is typical of the instances of pair programming I observed, although it is clear that, in this remote pairing context, the computational objects play a constitutive role in enacting the material-discursive practice, for in addition to providing the context of a shared programming environment (which is also the case in face-to-face pairing), the computational objects also play a constitutive role of providing the communicative link between the pair.

I therefore argue here that within the short space of 45 minutes, the empirics provide evidence for all�ve dimensions of leadership as depicted in�gure2.4on page32. These�ndings are summarised in table4.1.

The question then arises, ‘Was what I observed leadership?’ I could argue that it was, based on my citations of the various dimensions of leadership and the extant liter- ature that align with the observed behaviour, but even if that is the case, that is not my aim here. My point in drawing the reader’s attention to the way in which the observed practices align with these dimensions is to make visible what may well be the ‘raw ma- terials’ members employ when they materialise leadership through a citational material- discursive practice. Thus, within a context such as Eta, the way ‘leadership’ comes into beingqualeadership in practice is through the citation of these dimensions bymembers

and their application of these citations to observed material-discursive practice: the en- actment of a citational material-discursive practice that references some other enacted material-discursive practice.

Verily, withinEta, evidence of such citations is present. For example, in analys- ing an internal electronic discussion board that is materialised by and accessed through computational objects, a number of statements are made by both employees and senior leaders explicitly ascribing leadership to the material-discursive practice of pair program- ming. I choose this data source in particular as the statements found therein precede my engagement with the site. In other words, they precede my questioning members about ‘leadership’ and looking for itin situ. Thus, they provide here a particularly trustworthy

form of supporting evidence for my argument that precedes any activities I undertook that may have biased or led the informants in any way to highlight leadership towards the ends of my research. For example, one employee asserts with respect to a client engagement that the practice of pair programming:

[C]learly led to a [positive] change in both engineering leadership and process (Eta Internal Discussion Database).

Here, the employee links leadership on the part of engineers as a management function resulting in e�ective changes in business process and client engagement, binding this citation with the leadership dimension ofprocessas depicted in�gure2.4on page32.

The founder of the company o�ers similar remarks, linking the desired outcomes (results) of pair programming with the leadership dimensions ofperson,position,process,

andpurpose. He asserts that:

The key to sticking with a process—or perhaps a better way to express it would be to continue to *care* about process, since it doesn’t really matter if it maps

exactly to what we do—is support at the executive level (Eta Internal Discus- sion Database).

The founder cites explicit links between producingresults(‘sticking with it’) through

aprocess(pair programming) and apurpose(continuing to ‘*care*’), all of which he claims

is underpinned by speci�c people (associated with the leadership dimension of person) who hold certain executive leadershippositionswithin the organisation. Thus, all�ve di-

mensions of leadership as depicted in�gure2.4on page32are mobilised to underscore what makesEta ‘special’ or ‘great’. What the founder is pointing out here is that this ‘sticking with a process’ is somethingEtadoes particularly well in relation to the material- discursive practice of pair programming. The statement therefore points towards what is viewed as important or valued atEta, that which is associated with the category ‘leader- ship’.

Thus, I claim, not only embedded within the material-discursive practice between Walt, Robert, and their computational objects, but also within the broader sphere ofEta

itself is an enacted and performative ideology that involves the association of various material-discursive practices that includes humans and computational objects and their relation to what is valued within the organisation and cited as forms of ‘leadership’. All of these suggest the possibility that the production of ‘leadership’ is itself achieved through material-discursive practices that may involve citing one or more of the�ve dimensions of leadership as depicted in�gure2.4on page32and its association with some other enacted material-discursive practice. To develop this argument further, I present a di�erent episode within the same site where the possibility of machines ‘leading’ is investigated.