Navigating Datascapes With a Monadological View
by Adam Robbert
I’ve just finished reading Bruno Latour et al’s paper “The Whole is Always Smaller Than its Parts.” I find myself walking away with several important insights that I think people will find useful, controversial, and, if nothing else, very interesting. Here are some of the main takeaway points:
1) The paper proposes a rehabilitation of Gabriel Tarde’s monadological approach to sociology. Latour et al suggest that it may be the case that Tarde’s appeal to monad’s failed as a sociological paradigm not because of its invalidity as a concept, but because of the inability to deploy monad’s as a research paradigm in sociology given the available research equipment of the day (Note: Latour is explicit here, as always, that “society” and the “social” is read in the Whiteheadian sense of being any coordinated group of actors, rather than just human actors);
2) Latour et al argue that we ought to reconsider the use of monad’s as a useful sociological concept in light of new capacities in research and network technology that allow human researchers to study complex, collective behavior patterns with greater ease than was possible before the ability to navigate through huge amounts of information. We might think of the internet (Web 2.0), GIS, or climate science, as examples that employ these new data navigation techniques.
3) Latour et al distinguish between two paradigms of sociological research they call “1-LS” and “2-LS.” In the 2-LS paradigm sociological issues are researched either through (a) a bottom-up approach that sees individual entities as the primary elements of any system which then engage with one another following specific rules of interaction creating emergent patterns of order within the system as a whole; or (b) a top-down approach where systems or structures are seen as primary and govern interactions that, consequently, generate the types of entities possible within the system. 2-LS is juxtaposed to;
(4) The 1-LS paradigm which actor-network theorists will recognize as the characteristically “flat” approach to sociological research Latour has been pushing for decades. Here the distinctions between “elements” and “individuals” or “structures” and “aggregates” is leveled with the rehabilitated concept of the monad. The writers use the metaphor of the internet to suggest that, just as the user shifts between different kinds of content, media, and platform; or different levels of complexity, scale, and accessibility, the user nevertheless never leaves a single plane of activity. The shifts are between different kinds of monads and not between levels of individuals or aggregates (and this models the 1-LS approach);
(5) Latour et al argue that the 2-LS paradigm has been the dominant mode of thinking in all branches of the sciences since the 17th century. We see variants of the 2-LS paradigm whenever there is an appeal to organic holism or to reductive atomism; they are both essentially inversions of the same set of ontological assumptions and therefore both suffer from the same limitations. The shift from 2-LS to 1-LS has very important consequences, both analytically and ontologically;
(6) The limitations of 2-LS are apparent, for example, whenever the behavior of a phenotype (a whole organism) is reduced to the behavior of, lets say, selfish genes. In the other direction, the 2-LS paradigm can rightly be accused of allowing holistic arguments such as the Smithian chestnut about the invisible hand of the market. In both cases two ontological levels are posited that then feed up or feed down into the derivative structure (either individuals or wholes, depending). Latour et al suggest that the 2-LS paradigm is analytically useful in certain contexts, but ontologically untenable because;
(7) “Why assume that there first exist simple individual agents, then interactions, then complex structure — or the opposite” (p. 7). In other words, what Latour et al are contesting is the unwarranted linearity of the 2-LS paradigm which is based on a confused ontological assumption about the nature of individuals and aggregates. Put in a simple phrase: there can be no individuals that are not also societies, and there are no societies that contain no individuals. In this sense individuals and contexts are the product of perspectives not ontological differences, which leads to;
(8) A critique of the now-popular theory of emergence. Surely new, emergent properties are possible, but what changes with emergence in the 1-LS paradigm is again this critique of linearity which posits individuals –> interactions –> emergent properties. What we get instead is a horizon populated by monads of different intensity, where the more associated (i.e. connected) monads may actually be simpler than the monads which constitute their relations. Thus not all elements of every monad are features of the larger social monad they enact, monads overlap but do not expose one another fully;
(9) I’ll offer my own example to support the above conclusion that wholes can be simpler than parts. Take the relation between airports and airplanes. One might be tempted to say that the airport is more complex than a single airplane since the airport includes multiple airplanes. But the airport, as a monad, doesn’t necessarily access all of the technical equipment on board an individual aircraft. In fact, at any given moment, each airplane may treated as a simple unity, encased in so much metal, glass, and paint, rather than as a technological multiplicity of parts that the monad-airport needs to access at all times.
The airport is thus composed of whole airplanes rather than hugely complicated machines that make up the insides of each airplane. Here airplanes are not atoms in the emergent structure of the airport, but rather monads which partially share their identity or being with the airport as another monad. In this way each monad has a multiplicity of identities (consisting of different levels of complexity), such that each monad is not a collection of smaller atoms or parts that then feed together systematically to form a whole (since monads do not “enter into” systems but are always-already part-wholes or actor-networks partially exposed in specific associations).