The Politics of Schemas: Part 1

January 31, 2001

Kendall Grant Clark

[XML is] the universal format for structured documents and data on the Web. -- W3C XML Activity Statement
"There is such a tremendous need there," said Eric Schmitt, an analyst at Forrester Research. "The technology world is just crying out for a way to easily pass around information about people." ... Paul Conn, manager of the standards group, said the idea is to create a system that will give businesses a single, encompassing view of a customer. "It's about making the market more efficient," said Conn, who likened the effort to paving over what is now a rutted dirt road in cyberspace. "It's an extremely big deal." -- Internet firms act to ease sharing of personal data, 5 December 2000, Washington Post.
Most people who bother with the matter at all would admit that the English language is in a bad way, but it is generally assumed that we cannot by conscious action do anything about it. Our civilization is decadent and our language -- so the argument runs -- must inevitably share in the general collapse. It follows that any struggle against the abuse of language is a sentimental archaism, like preferring candles to electric light or hansom cabs to aeroplanes. Underneath this lies the half-conscious belief that language is a natural growth and not an instrument which we shape for our own purposes. -- George Orwell, Politics and the English Language, emphasis added


In the first part of this two-part essay, I examine the ways in which the Semantic Web may be political by focusing on the ways in which schemas may be political. But first I need to lay out some background knowledge, describe how I construe some familiar terms, and how I define some other, perhaps unfamiliar, terms.

Institutions and individuals create schemas in working groups of programmers, analysts, domain specialists and others. The schemas they create are usually formalized in meta-languages like DTD or the W3C's XML Schema Definition Language. Some schemas become de facto or de jure standards. The widespread use of a standard empowers those who control it; the more it is used, the less reason to develop or to use an alternative.

Schema making requires the skill and expertise of many expert practitioners and is costly. Schemas are often created by institutions whose sole purpose is financial profit. Institutions create private schemas to use internally, within the institution itself; and they create public schemas to use externally. The number of public schemas increases daily. Schemas are being created that will structure exchange between people, between people and institutions, and between institutions, including all levels of government, corporations, and non-governmental organizations.

The Promise of the Semantic Web

Imagine how broadly the Semantic Web -- as suggested originally and recently -- would alter some people's view of and exchanges with the world. (I say "some" because, as things stand, it will only ever be a fraction of the world's population, not even all those living above bare subsistence, who will use what we in the West parochially call a "universal" or "global" infrastructure.)

Let's assume for a moment that the Semantic Web is political. What does that mean, and why is it a problem? I think it's a problem because of the kind of technology the Semantic Web aims to be. It is a world-mediating system; it mediates between users and a part of the world, often by manipulating machine representations of the world. State changes in the software system may cause state changes or side effects in the real world.

Consider a Web-based banking application. Performing banking tasks by using a Web application is functionally equivalent to performing them at the bank's physical location. There are obvious phenomenological differences to the user in each case, but there aren't any differences to the user's bank account. A $100 withdrawal from a teller is equivalent, in all respects relevant to the bank account itself, to a $100 Web application withdrawal. A Web-based funds transfer just is a funds transfer, as a matter, among other things, of convention and institutional fact.

We're told the Semantic Web will help us track what we think is important; relate to the government, our doctors, financial institutions, the media; protect our privacy; find, manage, exchange and understand information about the most fundamental and intimate aspects of our lives (as individuals, as societies of individuals). We're told, in short, that the Semantic Web will mediate between us and the world in very significant ways. So if it is political, that needs to be discussed and understood like any other widescale, sweeping political change.

Schemas, Interests and Contested Understandings

The phrase "politics of schemas" is curiously polysemous. In this context it could mean industry or vendor politics; the politics of funding and of standards organizations; API politics; or competing technocratic policy in the narrowest sense (Strassmann 1994). The incurious claim is that a technology is political. From its ages-old use in making war, including making war absurd by making absurdly (because uselessly) accurate weaponry (MacKenzie 1993), to the fact that existing social and economic inequalities determine who gets access (and what quality of access) to technology (Schiller 1996), to the history of technological production in Western industrial democracies, which has always already been political, it should come as no surprise that technology serves dominant economic and political interests more often than not (Noble 1979, Noble 1986, Perelman 1998). The only surprise is when it doesn't.

With regard to the Internet's infrastructural technology that will power the Semantic Web, to say nothing of XML itself, the recent overweening concentration on commerce, to the exclusion of other fields of human endeavor, is part of larger cultural and historical trends that place the stateless corporation at the center of human life as lived in the West (Schiller 1991). And yet the earliest promise of the Web (as well as its precedent hypertext forms) was always for a richer, more collaborative, less commercial environment than we face today. No one should be surprised by that outcome.

The triumph of commerce on the Web can tell us much. It cannot tell us how and to what extent the forces of commerce will dominate the Semantic Web. Like every other technology -- like all human tools, including language, as George Orwell said in "Politics and the English Language" -- the Semantic Web is a social production, not an irresistible force of nature. The Semantic Web will be shaped by human choices, which implies that it will be political. To argue that point I suggest that schemas, one of the Semantic Web's building blocks, will be political in at least two senses.

By "schema," I aim more broadly than ordinary in XML circles. A schema is a representation, an expression about or directed at something (Searle 1995). Directed at what? A schema represents an understanding of a world-chunk. The schema may be formally or informally expressed, and the understanding it contains may be formal or informal. A world-chunk is any part of the world of which one can create a machine representation: social, political, economic practices or conventions, institutional processes, and so on. What concerns me here are machine representations of the social world, the world of institutional rather than brute facts (Searle 1995). Making machine representations possible is the immediate goal of schema creation.

Schemas formalize an understanding of what there is in a part of the world, what matters there, how its parts fit together, how it and its parts fit together with other world-chunks, with the parts of other world-chunks. Variations exist across actual schemas, but, generally, they formalize an understanding of a part of the world according to what matters in that part of the world -- where "matters" and "what there is" and "understanding" are always relative to goals, purposes and interests. Understandings of world-chunks are always interest-relative and goal-specific. The relation of understandings to interests is such that there cannot be in principle an interest-neutral understanding of a world-chunk.

Schemas are thus always relative to the interests of a person, an institution, or both. To put it in a slogan -- schemas reflect the institutions that produce them. (Nothing rests on talking about individuals and institutions; in what remains I talk only about the latter.)

Schemas reflect particular interests

Schemas are political, in the first sense, because they serve or reflect particular interests, thus tending to exclude the interests of others. Schemas may be political in another sense by formalizing an understanding of a world-chunk that's strongly or weakly contested. Here I'm adapting the idea of contestedness from W.B. Gallie's classic essay "Essentially Contested Concepts" and William Connolly's The Terms of Political Discourse. The meaning of an essentially contested concept is perpetually contested and recognized as such by the parties of the contest (among several other criteria). Democracy is an example of a Galliean essentially contested concept.

Contestedness is of course a practical reality of rational deliberation, and schema-making is partly constituted of rational deliberation. Some world-chunks are or contain social, economic, or cultural practices, conventions, or resources the meanings of which are strongly or weakly contested. A strongly contested world-chunk is one that is or contains concepts or practices the meanings of which are contested in principle. A weakly contested world-chunk is one which is or contains concepts or practices the meanings of which are decidable in principle, but which have been so long or so intensely contested that the relevant parties can conceive of no mutually satisfactory resolution. If the chunk of the world formalized by a schema is (or contains that which is) weakly or strongly contested, the schema necessarily takes a position not shared by all relevant parties.

When a world-chunk is contested, its schema necessarily formalizes an understanding that is not shared by the relevant parties. In other words, it favors or privileges one understanding among many. Insofar as schemas are developed by institutions with particular interests and understandings of contested world-chunks, the schemas they produce will reflect those interests and understandings, which are likely at odds with the interests and understandings of others. How common are nontrivial, wholly uncontested world-chunks? As one might expect, schemas, like other technological productions, reflect the contested understandings of the institutions that produce them.

Schemas confer power

But politics is about power and coercion as much as it is about interests, discourse, and deliberation. Schemas that are de facto or de jure standards empower the institutions that produce and control them. What is the nature of that power? The answer lies in the nature of social power as asymmetrical access, an idea developed by the feminist philosopher Marilyn Frye in "Some Reflections on Separatism and Power" (Frye 1983). "Access is," Frye says, "one of the faces of Power" (Frye 1983, p. 103). If I have power over a person I have greater access to that person than the person has to me; and this asymmetry of access is directly proportional to the power difference between us. The more power I have, the less access the other has to me, and the more access I have to the other. As Frye says,

Differences of power are always manifested in asymmetrical access. The President of the United States has access to almost everybody for almost anything he might want of them, and almost nobody has access to him. The super-rich have access to almost everybody; almost nobody has access to them. The resources of the employee are available to the boss as the resources of the boss are not to the employee. The parent has unconditional access to the child's room; the child does not have similar access to the parent's room ... The slave is unconditionally accessible to the master. Total power is unconditional access; total powerlessness is being unconditionally accessible. The creation and manipulation of power is constituted of the manipulation and control of access (Frye 1983, p. 103).

Applying Frye's notion of social power to schema-making is easy enough. Standards are sites of social power. Controlling them confers power. Access to standards is largely asymmetrical. Institutions that control a schema qua standard have access thereby to the understanding of a world-chunk formalized by the schema. But few who use, directly or indirectly, the schema have access to it. Schema-makers have access to those who use the schema by influencing, even determining, the ways in which they interact with parts of the world.

Institutions that control schemas that structure widely used Semantic Web applications will have power over users. A sign of that power will be patterns of asymmetrical access. Users will have no access to the schema, except when controlling institutions are democratic, and no power to control, change, or improve it to reflect or better reflect their interests or understandings of the world. The controlling institution has significant access to the ways in which these users interact with the world through the mediation of machines and machine representations of the world.

A schema is never neutral ground upon which the relevant parties meet, on equal terms of symmetrical access, to agree to machine representations of a world-chunk. Like the child's room, the worker's shop floor, and the slave's body, schema users are likely to inhabit territory they do not control, territory controlled by institutions with interests and understandings perhaps inimical to their own. Children are not employees, neither children nor employees are slaves, none of whom are Semantic Web users; but the idea of power as asymmetrical access applies to each.

So by the "politics of schemas" I mean two senses: first, the interests reflected in the schema, which are the interests of the institution of production; and, second, the adjudication of strongly and weakly contested world-chunks. In short, schemas are political when they shut out some interests while serving others, and when they take positions on contested questions and issues. Having determined the ways schemas may be political, the conditions under which they may be apolitical are clear as well. Schemas are apolitical when all relevant (that is, all affected) parties have the same interests, and the world-chunk formalized by the schema is not contested. Neither is a sufficient and both are necessary conditions of apolitical schemas.

I have shown two ways in which schemas, and schema-employing technologies like the Semantic Web, may be political. In the second part of this essay, I will explain the relevance of political schemas to the XML community and what the XML community can and should do about them.

Works Cited

William E. Connolly, The Terms of Political Discourse, Princeton University Press, 1993.
Donald Davidson, ``On the Very Idea of A Conceptual Scheme,'' Inquiries into Truth and Interpretation, Oxford University Press, 1984.
Marilyn Frye, The Politics of Reality: Essays in Feminist Theory, The Crossing Press, 1983.
W. B. Gallie, `` Essentially Contested Concepts,'' in Proceedings of the Aristotelian Society, 56(n.s.), 1956.
Donald MacKenzie, Inventing Accuracy: A Historical Sociology of Nuclear Missile Guidance, MIT Press, 1993.
David Noble, America by Design: Science, Technology, and the Rise of Corporate Capitalism, Oxford University Press, 1979.
--------, Forces of Production: A Social History of Industrial Automation, Oxford University Press, 1986.
Michael Perelman, Class Warfare in the Information Age, St. Martin's Press, 1998.
Herbert Schiller, Information Inequality: The Deepening Social Crisis in America, Routledge, 1996.
--------, Culture Inc.: The Corporate Takeover of Public Expression, Oxford University Press, 1991.
John Searle, The Construction of Social Reality, Free Press, 1995.
Paul Strassmann, The Politics of Information Management: Policy Guidelines, Information Economics Press, 1994.

Want to comment on this article? Email the author.