A Taxonomic Framework for Social Machines

http://​www.​galaxyzoo.​org/​.

http://​www.​ushahidi.​com/​.

See http://​sociam.​org/​.

http://​www.​youtube.​com/​.

http://​www.​flickr.​com/​.

http://​www.​alexa.​com/​.

For an overview, see http://​en.​wikipedia.​org/​wiki/​Wikipedia:​Bots/​Status, retrieved in December 2013.

http://​picbreeder.​org/​.

The notion of a Web-extended mind draws its inspiration from work that goes under a variety of headings, such as ‘extended cognition’, ‘cognitive extension’ or ‘the extended mind’ [8, 9, 28]. Smart [44] defines a Web-extended mind as a system in which some of the informational and technological elements of the Web can be seen to constitute part of the material supervenience base for (at least some of) a human individual’s mental states and processes.

The use of the term ‘mechanistic realization’ in the definition is intended to highlight the importance of this mechanistically-oriented explanatory account [59].

Similarly, it is the complementary nature of biological and non-biological resources (in terms of their contrasting representational and computational capabilities) that is often seen as lying at the root of the advanced forms of intelligence exhibited by extended cognitive systems. Sutton [49], for example, writes that “in extended cognitive systems, external states and processes need not mimic or replicate the formats, dynamics, or functions of inner states and processes. Rather, different components of the overall (enduring or temporary) system can play quite different roles and have different properties while coupling in collective and complementary contributions to flexible thinking and acting” (p. 194).

Clocks may provide one example of a social machine that is independent of the Web. In their book, ‘Anti-Oedipus’, Gilles and Guattari [16] suggest that clocks are a form of ‘social machine’: “The same machine can be both technical and social, but only when viewed from different perspectives: for example, the clock as a technical machine for measuring uniform time, and as a social machine for reproducing canonic hours and for assuring order in the city” (p. 155). Interestingly, clocks have been seen as providing the technological impetus for the transformation of society. A number of theorists have emphasized the way in which clocks enable the large-scale scheduling and coordination of both individual and collective action, and the way in which the transition from fixed, centralized clock towers to portable wristwatches paved the way for new forms of social interaction and engagement [23]. The invention of portable time-keeping devices, argues Landes [23], made it possible to organize and synchronize activities in a way that had never been possible before, and on the back of this new capability there emerged a new social and economic era. The clock, in this case, can be seen as the technological element of a social machine in the sense that it is influencing social interaction via the delivery of machine-generated temporal representations. These representations serve to structure, sculpt and scaffold forms of social interaction and engagement that progressively shape the contours of the social, economic and cultural landscapes in which we live.

https://​www.​linkedin.​com/​.

http://​www.​google.​com/​recaptcha.

https://​myspace.​com/​.

http://​www.​reddit.​com/​.

https://​www.​mturk.​com/​mturk/​.

The Web site of the SOCIAM research project lists a large number of additional examples of social machines—see http://​www.​sociam.​org/​social-machines.

We are grateful to Ségolène Tarte (University of Oxford) for pointing this out.

https://​diasporafoundati​on.​org/​.

This corresponds to the tier termed ‘Causes/Groups’ in Fig. 1, which builds on a selection of Web-based systems that, through their large-scale user bases and general character, have reached a level of popularity that turns them into frameworks for the development of more special purpose social machines.

It is also possible to imagine one or more social machines being ‘incorporated’ into a larger social machine. In the same way, perhaps, as the neurological subsystems associated with memory, attention and perception merge to form part of the integrated mechanistic substrate that realizes more ‘macrocognitive’ functions such as sensemaking (see [21]).

In fact, as we mentioned in Sect. 3, the technological subsystem is only considered to be one part of the social machine; the human participants are also deemed to be part of the social machine.

https://​www.​linkedin.​com/​.

http://​www.​planethunters.​org/​.

See http://​gigi.​cpsc.​ucalgary.​ca/​.

https://​www.​zooniverse.​org/​.

The constructs identified in the context of the repertory grid exercise ultimately drive the generation of dimensions associated with the taxonomic framework. A construct such as ‘Heterotelic vs. Autotelic Usage’ (see Sect. 4.1), for example, is ultimately used as the basis for the ‘Motivation Type’ and ‘Form of Motivation’ dimensions listed in Table 2.

Note that although two dimensions may be similar, they are only regarded as identical if the set of characteristics associated with the dimensions is the same in each case. In the absence of shared characteristics, a dimension mapping is regarded as ‘partial’.

Such consistency is evidenced by the way social machines are described in a number of papers. We thus encounter descriptions of social machines as “purposefully designed sociotechnical system[s] comprising machines and people” [10], as systems in which “the human and digital parts…[form] a machine in which the two aspects are seamlessly interwoven” [43], and as systems that involve “the co-constitutional involvement of humans and technologies” [50].

Note that in the light of our definition, the ‘engineering’ of a social machine entails more than just software development and deployment; it also includes the assembly of mechanisms that enable and encourage user engagement.

The reliability of the framework is indicated by inter-rater reliability metrics. Poor measures of inter-rater reliability may indicate that some dimensions are more difficult to interpret, understand or discern than others. This may call for the dimension to be refined or removed from the framework.

The process of identifying categories or classes of social machines is supported by the use of statistical methods that are applied to the social machine morphospace. Cluster analytic techniques are typically used to support these analyses (see Geiger et al. [14] for an example of such techniques applied to crowdsourcing systems).