XML.com: XML From the Inside Out
oreilly.comSafari Bookshelf.Conferences.


The Politics of Schemas: Part 2

February 07, 2001

In the first part of this essay, I argued that the Semantic Web will be political because schemas, one of its foundational parts, are essentially political. I claimed schemas are political if they reflect the interests of the institutions that produce them, and they formalize an understanding of strongly or weakly contested world-chunks. Schemas are political when they shut out some interests while serving others, and when they take positions on contested questions and issues.

In this final part, I consider the relevance of political schemas to the XML community and suggest some avenues of further research and consideration.

XML is Not Immune

You may find yourself agreeing that schemas are political but wondering, nevertheless, what it has to do with XML practitioners or with XML itself. XML is, however, a universal data format. If we take the universal claims made about XML seriously, professional schema-makers must ask whether some interests and views of contested concepts might be excluded, perhaps systematically, from schema-making and from schemas; whether such exclusion is socially beneficial or harmful; and, if harmful, what should be done about it.

From the early days of XML's development there's been talk about vendor neutrality, interoperability, and universality. Such talk was part of SGML's appeal since the mid-70s and rightly so. Today that talk fails regularly to take account of politics. XML advocacy often ignores the fact that schemas may be vendor neutral but cannot be interest neutral; that schemas may be universally accessible but formalize a strongly contested understanding of a vital part of the world; or that schemas may distort or impede some people's interactions with the world in ways they find inequitable or inappropriate.

XML schemas are often placed in the public domain and available for anyone's royalty-free use (subject obviously to uncommon levels of knowledge and expertise) -- a state of affairs clearly preferable to proprietary alternatives. But is it enough? What good does it do that one can use, even modify a de facto standard schema, royalty free, when the schema reflects interests inimical to one's own, formalizes an understanding of the world one strongly contests, and is used in a widely deployed, vital Semantic Web application that has no serious competitor? What good does it do to modify the schema to reflect one's own interests and understandings if doing so renders it unusable?

More crucially, what good will a public domain schema do for people, relying on a Semantic Web application to mediate to them a world-chunk, if it does so in a way inimical to their own best interests? They can perhaps stop using it. But no one, aiming to build as ambitiously as the Semantic Web requires, can be indifferent to people eager to use it but choosing not to because it disdains or ignores or disadvantages them.

That a schema is in the public domain or royalty free is largely irrelevant to almost everyone who's not an XML practitioner. XML vendor neutrality may be true but, in a wider context, irrelevant. What really matters socially is the equitable and fair disposition of interests, understandings, and power. It's crucial that some computer experts, especially XML technologists, think about questions beyond vendor neutrality. If XML practitioners with the requisite expertise, knowledge, and curiosity don't think about them, who can and will?

The limits of effective change

Political schemas are conceptually independent of most considerations otherwise relevant to schema-making. The political nature of schemas is independent of the languages in which they are expressed. A political schema will be no less or more political when expressed in XML, SGML, UML, Java classes, RDBMS systems, symbolic logic, or natural language. It follows, then, that the political nature of schemas is conceptually independent of the particular XML applications used to declare them formally, whether DTD, W3C XML Schema Definition Language, RELAX, Schematron, TREX, etc.

Political schemas are also conceptually independent of political perspective. They are not political only if one holds a particular assemblage of political views. Schemas are essentially political because of the way the world is and the way people are. One might make this point another way by saying that the political nature of schemas is conceptually independent of the answers one gives either to the concrete political questions of one's historical situation or to the enduring questions of political philosophy.

The political nature of schemas is conceptually independent of the Semantic Web itself. Any technology that uses schemas, as described in Part One, is a concern for those of its users whose interests and understandings are at odds with those of the institutions that control it. The Semantic Web is intended to mediate to its users a chunk of the world. Recall the Web-based banking application from Part One: it mediates a part of the world, a bank account, to its users such that some state changes in the Web application cause side effects in the real world. It is in that mediation, based on machine representations of the world, and thus an understanding of what the world is like, that the political nature of schemas is especially troublesome. Thus, while I claim that the pitfalls of political schemas are largely independent of the Semantic Web, the more robust and the more strongly mediative the Semantic Web becomes, the more likely it will be that some of its users will find their interests and understandings effectively shut out.

There is a range of appropriate responses to all this, but modifications of XML technology, of specifications, of encoding formats, of political persuasions are not included in that range. The political nature of schemas cannot be avoided by taking up a different political viewpoint or by using one set of conventions for formally expressing a schema rather than another set.

Where Do We Go From Here?

Computer experts have a professional responsibility -- to say nothing of their responsibilities as citizens and as human persons -- to "take a pro-active public role in promoting societally beneficial uses of computers and discouraging harmful ones" (CPSR 2000). Two questions follow from this professional responsibility. First, to what extent has the XML community thought seriously about the responsibility to distinguish benefit from harm? Second, to what extent have analytical tools necessary to doing so been developed? The urgency of the situation is signaled by the magnitude and direction of two vectors of technological change -- the first, to put onto the Web everything that can be; the second, to schematize everything on the Web that can be. Computer experts cannot presume every effect of such far-reaching change is or will be benign or felicitous. What can we do to respond appropriately to the political nature of schemas? In what remains I suggest where we might go from here.

The XML development community -- especially academic researchers and independent developers situated outside large corporations -- should think and talk explicitly about politics and XML technology, particularly the Semantic Web. The issues are too important, the technology too complex, to leave such thinking to others alone. Managers, marketers, politicians have a role, but not the role of XML developers. The conversation should include technically savvy specialists in fields like ethics, philosophy, and law. Those who think very deeply about schemas and about the Semantic Web must consider how competing interests and contestedness matter to the creation and use of schemas.

There are irreconcilable interests and strongly contested concepts central to both the political traditions of Western democratic societies and to schemas already made, being made, or planned. Privacy is one such concept. Economic development, censorship and content labeling, informed civic participation, education, natural resource usage, financial information, the relation of doctor and patient are others. The list is long and varied; and it grows longer and more varied every day. Schema-makers need to come to theoretical and practical terms with the likelihood of there never being widespread agreement among the relevant parties about the contested meanings and competing interests of these and other aspects of public, common life.

Part of coming to terms with this likelihood is theorizing about the treatment and adjudication of competing schemas, including those which may be in principle irreconcilable and those that may not be. Such work is hard for human beings. No one seriously claims that machines will ever do this work for us. We have to face, then, a schema forking problem: the existence of multiple, irreconcilable schemas of the same (or overlapping) world-chunks. I can't think of any good reasons why this won't or shouldn't happen. It's rational for people (or, better, institutions) to consider forking a schema when the standard one doesn't serve or even harms their interests. While I'm certain that it will and in some cases should happen, I am not certain how or whether future versions of the Web can be developed to deal usefully with the fact of multiple, irreconcilable schemas. I suspect that these issues will be subject ultimately to resolution by non-technical, social mechanisms, that is, by legislatures and courts. Computer experts need to think hard and clearly about the politics of schemas so that, when called upon, they have something useful and responsible to say to legislators, judges and juries, oversight committees, etc.

Pages: 1, 2

Next Pagearrow