Saturday, March 22, 2008

Google and Attractors (Part 3)

If we are to believe Claire Colebrook, who claims, in Understanding Deleuze, we are in post-linguistic era, we need to develop theories and approaches that are not language dependent. This means we cannot use discourse analysis or propose a model for communicative action. I do not claim that these approaches are not helpful or should be abandoned. Instead the problem of these approaches is that language is always assumed to be present. In other words, language becomes the transcendent principle and according to Deleuze and Guattari we need to think immanently instead of transcendently. Where can we then turn? In this post I would like to suggest complexity theory can help us out the language impasse. To achieve this I will concentrate, once again, on Google’s search engine. A topic I have discussed previously here and here. In this post I will argue that the language we enter into the search engine functions as a chaotic attractor.
In simple terms an attractor helps to explain the behaviour of a (real) system. For example, if there is a bowl and we put a marble in the bowl, then the point at where the marble rests is referred to as the attractor. However, this example could suggest that attractors are deterministic, in the sense that the outcome is always the same, or that attractors are singular, in the sense that there is only one attractor in a system. Both of these claims are untrue and we need to be careful of proposing determinism. Another example can demonstrate that attractors are not deterministic and multiple. Imagine a game of football/soccer. In the game there are two goalposts situated at each end that act as attractors. As Brian Massumi writes: ‘The field is polarized by two attractors: the goals. All movement in the game will take place between the poles and will tend toward one or the other. They are the physical limits.’[1] Importantly, the goals do not decide the result of the game, but they do determine the movements and variations within the game. The goalposts, as attractors, literally feed into the game.
So how then is this relevant for understanding how Google functions as a search engine when we enter keywords? From a language perspective we could suggest that we are creating what Sassure would term a linguistic sign. This would argue that the keywords entered into Google are the signifier and what is returned from Google is the signified. Both of them together compose the linguistic sign. However, this is exactly the type of thinking we are trying to avoid. There are two clear problems with this type of thinking. The first, as I have mentioned above, is the failure to represent a post-linguistic mode of thinking. To think in terms of linguistic signs means language is a necessary component. The main issue I have with this problem is that language is not always a component of life. In other words, things occur without language (e.g. geological stratification). The second problem Sassure's linguistic sign does not inform us how Google functions as a machine. This is the advantage of replacing the linguistic sign with the idea that the keywords are attractors. Attractors have the benefit of helping to provide an explanation for when a system reaches a stable point. If we return to the football example, we can see how the goals explain how the game settles after 90minutes. Of course, the goals, as attractors, do not fully explain the result of a game. There are other relations of exteriority that require recognition: players, ball, referee, rules…However, the attractors in a system are not there to offer a full explanation, and instead the focus of the analysis should concentrate on how the attractors affect the behaviour of a (dynamic) system. In the case of the keywords entered into Google these influence the behaviour of the sorting machine and explain the emergence of a settled state. In this system the settled state is the website created after a search is executed (i.e. the return page). The point is each time we enter keywords into we are providing the sorting machine with an attractor. The attractor enters into the system of Google, which can be thought of as the algorithm(s) Google use to download, index, and rank online documents. On the whole, maybe it would be more accurate to realise we are not entering a signifier into Google, but actually entering an attractor. In addition, attractors, in a nonlinear world, continually influence the behaviour and movements of systems throughout the world.

[1] Brian Massumi, Parables For the Virtual: Movement, Affect, Sensation (London: Duke University Press, 2002) p72

No comments: