Monday, 10 June 2013

Computation and Measurement

Yeats said "Measurement began our might". To measure something (certainly in Yeats's sense) is to understand something of its meaning.

Computation isn't measurement. Computation - when treated sensibly - serves to help us discover the limits of our abstractions. We put a formula into the computer, we explore the parameters.  We notice when interesting things happen. By 'interesting', we look for those things which we couldn't have anticipated.

When we measure we bring tools and ideas to bear on the world which is seen to be independent of the measuring act. Measurement is often seen as a means of observing what is there. The fact that our act of measurement changes the world as every act does is a complexity which the scientists of the enlightenment hadn't grasped. When we measure, we look simply look to gather things together.

Measurements produce data. Data can be fed into a computer whose algorithms produce effects, some of which we may find 'interesting': things we couldn't have anticipated. But because those algorithms operate on what we believe to be "measurements" we have a tendency to forget that the interesting things we see are the product of the limits of our abstractions. Instead we believe them to be isomorphic to the world which we measured. We then attribute meaning to phenomena in the world which we have gained through our computer screens. The world is changed as a result - but not in ways we can imagine.

The deep problem here is that whilst out conception of finding something interesting in computation revolves around finding something we couldn't anticipate, we do not accept that measuring something isn't also an act related to anticipation as well. Yet to believe something to be measurable is the result of a way of looking at the world where the phenomena look stable, countable, regular, etc.

But there are many lenses through which to view the world, and each lens produces its own kind of stability, and consequently, a different kind of measurability. Which measurement? Which lens? Which computation?

A typical trick in the social sciences is to use computation to produce the illusion of measurability. Statistics flatten experiences to data which can then be compared. In our financial institutions, the same thing happens: trading systems use prices as cyphers standing for real substances and qualities.

Consequently we are in a mess.

But only because we have forgotten what we are looking at. What we look at are things which we couldn't anticipate; things which are the limits of our abstractions. The obvious thing, given this, is to develop our abstractions! In social science, this rarely happens.

Instead, new policies are created to make observed phenomena fit the existing abstractions. Data operates as a kind of currency, with everyone competing to have the 'best data', everyone trying to implement the policies to produce the best outcomes within the current abstract frame. In the heat of this competition, changing the frame is practically impossible.

The question is "where does the real meaning lie?" Anticipation in daily life requires deep knowledge of each other. It requires the kind of knowledge that doesn't come through computer-processed measurements. It comes through lived experience.

Perhaps not until we see our computation processes and measurements as part and parcel of a lived experience rather than abstract entities, will we be able to situate the power computation and technology within the context of well-lived and dignified lives. 

No comments: