The Epoch of Potential Memory

One can see Manovich’s argument becoming true in the development of database technology the 20th century. The first commercially available computer databases were organized hierarchically. If you wanted to get to a particular piece of information, you went to the overarching category and made a series of choices as this category broke down into groups then subgroups until you got to the specific piece of information that you required. This mode of traveling through a database was called “navigation.” The next generation, network databases, followed the same logic, that is, the user had to follow one of a number of predefined pathways in order to get to the data. It was more ordered than a straight narrative archive but it still preimposed a set of narrative structures on the data. The following generation, relational databases, began to break this mold. The underlying database structure is a set of relations or tables, each table having rows and columns. This matrix form allowed a new form of inquiry to be made: Users no longer had to travel the preset pathways, they just had to declare what they wanted to know in a controlled language. Finally, object-oriented databases operate on the principle that you do not need to know either pathways or relationships beforehand. Each data “object” carries its salient history, and pathways and relationships can in principle be reconfigured at will (Khoshafian, 1993). The canonical scientific act for our time (sequencing the genome) resonates with the social and technical turn toward the nonnarrative memory described by Manovich.

To give a name to the current epoch, I call it the epoch of potential memory. To continue Manovich’s trope, this is an epoch in which narrative remembering is typically a post hoc reconstruction from an ordered, classified set of facts that have been scattered over multiple physical data collections. The question is not what the state “knows” about a particular individual, but rather what it can know should the need ever arise. A good citizen of the modern state is one who can be well counted, along numerous dimensions, on demand. We live in a regime of countability with a particular spirit of quantification. Michel Foucault (1991) pointed out that this is one of the principles of governmentality: A modern state needs to conjure its citizens into such a form that they can be enumerated. The state may then decide what kind of public health measures to take, where to provide schooling, what kind of political representation should be afforded, and so on. Uncountables in the West are our version of the untouchables in India: a caste that can never aspire to social wealth and worth. In order to be fully countable and thus remembered by the state, a person needs first to fit into well-defined classification systems. At the start of this epoch, the state would typically— where deemed necessary—gain information on its citizens through networks of spies and informers writing narrative reports. Such information gathering continues today but is swamped by the effort to pull people apart along multiple dimensions and reconfigure the information at will.

But that seems to be quite a jump, from the way in which databases work to the operation of the state. The jump is possible because our way of organizing information inside a machine is typically a meditation on and development of the way we organize it in the world. When the first objectoriented language, Simula, was invented, it was perceived as a way of modeling the way things were actually done in the world. The claim today is still that you take a simple English-language description of system requirements and turn the nouns into objects and the verbs into operations and you are up and running. Object-oriented programming, by this claim, is the ultimate transparent language. At the same time, and from the other end, numerous management theorists claim that now that we have object-oriented programming, we can reconfigure the organization so that it matches the natural purity and form of the programming language. We no longer need hierarchical modes of communication; rather, we can organize according to teams with their own sets of interfaces with management, but where management does not need to know how any particular job is carried out by the team. Thus a programming language that operates as part of an organizational infrastructure can have potentially large effects on the nature of the organization through the medium of organization theory. So object orientation is on the one hand a model of the world; on the other hand the world is learning how to model itself according to object orientation. This kind of bootstrapping process is common when one deals with infrastructures. Generally, I would describe it as the programming language and organization theory converging on a particular instantiation of the organization in which object-oriented programming will furnish the natural, transparent language. This convergence is central to information infrastructures. We make an analytical error when we say that there is programming on one side, with its internal history, and organization theory on the other, with its own dynamic. The programming language is very much part of the organizational history and vice versa. James Beniger (1986) made this kind of connection in his work. Following a robust tradition in cybernetics, he noted that in the late 19th century many things came together to make process control a key factor in management and technology.

We live in a world where we can pull any aggregation of facts out of historical references to produce the aspects of history we wish to explore. It is dynamic and full of potential.