Creating Technology for Social Change

What Baboon Notebooks, Monads, State Surveillance, and Network Diagrams Have in Common: Bruno Latour at CHI 2013

I’m here at CHI 2013, a human computer interaction conference, for the third and final keynote, the sociologist of science and anthropologist Bruno Latour, on the topic, “From aggregation to navigation, a few challenges to social theory.”

Bruno Latour at CHI2013
 
photo by @pstamara

Latour starts by explaining what he calls “the monadological principle,” an alternative to the idea of collective phenomena. Latour offers “a strange argument,” that “there is no collective phenomena… but there exist many collecting devices that generate collected phenomena.” There is no upper level of collective experience or a lower level of individual experience. Instead, he argues, we aggregate experiences into something that we we call collective experience.

To illustrate this, Latour shows us a seven-screen stock trading desk, taken by Armin Linke. This is a “collecting site”– a specific location with two-way data collection flow. It uses a computational instrument to collate the data. It has a visualizing interface, and it requires a team of trained experts.

This collecting interface, what Latour calls an “oligopticon,” is a highly specialised view that shows very focused information. It’s the opposite of the panopticon — which offers the illusion of seeing everything. The control room of a nuclear plant or the monitoring array for a volcano site are oligoptocons — specialised views on data.

Collective phenomena grow out of these collecting sites. “In the expression ‘collective phenomena’ we should underline the word ‘collecting’ and the instruments, sites, data flows and expert teams that do the collecting, display the data and evaluate the resulting arrays.” Latour says that we should replace the idea that there are collective phonemena with a focus on these locations, whether it’s the high tech computer system at CERN or the low tech notebook of Shirley Strum as she studies baboons. They are all collecting devices.

Dr Shirley Strum recording behaviour at Gilgil; photo by Neil Leifer

Change of scale is a myth, Latour tells us, in the study of collective phenomena. What is actually happening is that we’re changing the nature of our instrumentation or the site of collection. He shows us frames from the Powers of Ten Video, by Charles and Ray Aimes. The film purports to move scale from a picnic in Chicago to the Big Bang all the way down to quantum physics. This film is a fallacy– there is no way that you can get a camera out that far or down that close.

Technologies like Google Earth replicate this fallacy– linking different datasets that are not optically coherent into the fallacy of zooma. What actual happens is a shift in the instrumentation and the local of focus.

Latour takes this argument to an even more difficult place (as he puts it): If there exist no collective phenomena, there exist no ‘individual’ phenomena either. How is this similar to the illusion of zoom? An individual, he says, “is not an atom but extends as far as all the entities that it perceives and with which it is in relation.”

What is an individual in this model?

  • A local site (a body)
  • a two-way data collection flow (perception, sensation and reactions)
  • an instrument to collate the data (language, patterns, formats, cliches, cultural frames)
  • a visualizing interface (staging and playing)
  • a team of trained experts (plurality of voices and personalities)

Latour directs our attention to the notion of the panopticon in the case of state surveillance. He tells a story of the Parisian CCTV system as it tracks a woman who is trying to navigate using a map, as presented in the film “Paris Invisible” (which he writes about in greater detail here). Both the woman with the map and the security system are the same– not to be considered in terms of their size but in terms of the nature of the connections in which they are involved. We assume that surveillance is “big” and that the map is “small,” even though the map may extend as far or further than the visibility of the CCTV system accessed by the Parisian police.

Why do we maintain this myth of scale when everything else shows the opposite? What researchers often mean by an ‘individual’, Latour tells us, is “an extended center of perception shaved from all its connections,” a connected site deprived of its dataflows and instruments. What is meant by ‘collective’ is a center that has those flows intact. The myth of these different levels raises a conundrum of reconciliation between the individual and collective. It’s not possible to reconcile them, says Latour, because our analytical frames deprive individuals of their connections. It’s not possible to model the emergence of structure out of individual actions because the idea of the individual is an artificial notion.

What’s the alternative? Latour introduces the idea of the “monad,” one in which the collecting activities of entities are foregrounded. The term monad comes from Leibniz and was later invented by Gabriel Tarde in respect to “an individualizing grasp of the whole universe of relations.” Whenever we use a technology device, we are connected to the rest of society. At this point, Latour describes the debate between Durkheim and Tarde in the formation of sociology as a field.

Why should researchers focus on monads? Everytime we invent a new technology, we produce another collecting phenomenon. Instead of seeing a “collective level” and a “individual level” we should look at the different collecting devices on their own terms. The trading desk reflects a collecting phenomena that we call Finance rather than any generic notion of “society.” The meaning of “collective phenomena” is the superposition and overlapping of all the collecting apparatus.

The digital allows us to follow these overlapping phenomena. For the first time, we have the direct experience of the monadological principle: the more you extend the network of relations, the more you individualize the grasping entities (note by Nathan: the more people you follow on Twitter, the more unique your network may be). We are constantly used to visually merging things which we would say are highly qualitative dataset. Latour shows us a chart of Linkfluence and LInkscape. Many of us as individuals are used to experiencing things like search engines and dashboards, making our personal experience much like the experience of the stock traders.

Even if we’re all connected to the same entity in the network, that entity is different for each of us. I am connected to the MIT Media Lab and Ethan Zuckerman is connected to the MIT Media Lab, but the Media Lab means something very different to each of us. Instead of seeing this difference as proof of the Media Lab, we should focus on the overlapping connections of monads. “We should abandon the key principle of impenetrability of entities and understand any ‘whole’ as the highlighting of some of its overlapping items.” Latour then directs us to his group’s paper “The Whole is always smaller than its parts” on this topic.

Biological research has the same problem, Latour claims. We’re able to follow individual cells and their progeny. He shows us a gorgeous video of individual cells in the embryos of drosophila– something that looks like this one:

Why should CHI Care?

Latour gives the CHI community the following challenges:

 

  1. Visual complexity produces opacity. Massive individualizing data produces beautiful, playful hairballs which show us nothing. How do we get filter and focus data while still appreciating monads?
  2. How can we capture the inner narrativity of overlapping monads? Latour shows us the “512 paths to the White House” visualization by Mike Bostock and Shan Carter. The other, the Guardian’s Rumour tracker, following the 2011 London riots. The idea that quantitative is different from qualitative is an artifact of the history of social science and a fallacy arising from the distinction between the individual and the collective, he tells us.
  3. How can we visualize heritage, process, and genealogies? Latour shows us a paper he worked with on “complex systems science” (I couldn’t find it). To be a monad is to establish connections, but timeseries visualizations can focus on structure rather than connectedness (like the paper on Phylomemetic Patterns in Science Evolution by Chavalarias, Cointet et al)
  4. How can we replace models about emergent structures with models that highlight differentially overlapping monads? He shows us a hairball network diagram and talks about the difficulty of moving beyond the hairball to understand the overlapping monads

 

Questions are submitted via card and are read by the chairs:

Question: Isn’t the “monad” another term for “network”? Latour says that “monad” is a synonym for “actor-network” which is clearer, since people tended to separate the “actor” from the “network.”

Question: the boundaries we use as researchers are used for convenience? The scientific process requires reduction? How do we resolve it? Why not keep our boundaries? Latour: we want to choose the reductions that keep the phenomenom alive, we want to get rid of the bad boundaries and replace them with better ones. If we want to be scientific, we need to study the phenomenon, not just make things disappear.

Question: things like social pressure cannot be reduced to individual experience? Aren’t they collective experiences? Latour: whenever you use a new collecting device and create new connections, you modify the world.

Question: If the Internet had existed when you started out, would social theory have developed differently? Latour: No. Tarde developed his ideas before the Internet. It would have been immensely helpful to have the Internet, but it wouldn’t have changed a bit.

Question: does an observation instrument collect data or create it? Latour: it performs them. The trading desk performs the notion of finance. It neither constructs it nor observes it– it doesn’t make the data up. It performs and it is observed. Followup: does it apply to more traditional experiments? Latour: provisionally yes, but it’s more complicated philosophically. Followup: is it easier now that computers are processing that data? Latour: both telescopes and computers modify information as they come in.

Question (missed it) Answer by Latour: the problem with the notion of structure is that it explains what lasts, versus what passes. That’s why it’s important to follow the trajectory and process. How do you represent the things that pass away from an event and the choices and relationships that continue? We now have the datasets to begin to raise the argument, but we need new ways to visualize and describe our notion of time and historicity.