though they are not good general search terms. On the other hand, in a specific social context (such as a particular person’s photos), the same tag can be useful since it can designate a particular individual. The use of a tag as metadata often depends on such a context, and the “network effect” in these cites is thus socially organized. ¹⁹

A more ambitious use of metadata involves recent applications of semantic Web technologies⁷ and represents an important paradigm shift that is a significant element of emerging Web technologies. The semantic Web represents a new level of abstraction from the underlying network infrastructure, as the Internet and Web did earlier. The Internet allowed programmers to create programs that could communicate without concern for the network of cables through which the communication had to flow. The Web allows programmers and users to work with a set of interconnected documents without concern for the details of the computers storing and exchanging them.

The semantic Web will allow programmers and users alike to refer to real-world objects—people, chemicals, agreements, stars, whatever—without concern for the underlying documents in which these things, abstract and concrete, are described. While basic semantic Web technologies have been defined and are being deployed more widely, little work has sought to explain the effect of these new capabilities on the connections within the Web of people who use them. ²⁸

The semantic Web arena reflects two principle nexuses of activity. One tends to involve data (and the Web), and the other on the domain (and semantics). The first, based largely on innovation in data-integration applications, focuses on developing Web applications that employ only limited semantics but provide a powerful mechanism for linking data entities using the URIs that are the basis of the Web. Powered by the RDF, these applications focus largely on querying graph-oriented triple-store databases using the emerging SPARQL language, which helps create Web applications and portals that use REST-based models, integrating data from multiple sources without preexisting schema. The second, based largely on the Web Ontology Language, or OWL,

looks to provide models that can be used to represent expressive semantic descriptions of application domains and provide inferencing power for both Web and non-Web applications that need a knowledge base.

Current research is exploring how the databases of the semantic Web relate to traditional database approaches and to scaling semantic Web stores to very large scales. 1 In terms of modeling, one goal is to develop tools to speed inference in large knowledge bases (without sacrificing performance), including how to exploit trade-offs between expressivity and reasoning to provide the capabilities needed for Web scale. ¹⁵ A market is beginning to emerge for “bottom-up” tools driven by data and “top-down” technologies driven by Web ontologies. Creating back-ends for the semantic Web is being transitioned (bottom-up) from an arcane art into an emerging Web application programming approach, as new open-source technologies integrate well with traditional Web servers. At the same time, new tools support ontology development and deployment (top-down), and tens of thousands of OWL ontologies are available for jumpstarting new domain-modeling efforts. In addition, approaches using rule-based reasoning modified for the Web have also gained attention. ⁴ Engineering the future Web includes the design and use of these emerging technologies, along with how they differ from traditional approaches to databases, in one case creating back-ends for the semantic Web, in the other new tools for ontol-ogy-based applications.

The semantic Web is a key emerging technology on the Web, but, also, as we’ve discussed, there are different opinions as to what it is best for and, more important, what the macro effects might be. Our lack of a better understanding of how Web systems develop makes it difficult for us to know the kinds of effects the technology will produce at scale. What social consequences might there be from greater public exposure and the sharing of information hidden away in databases? A better understanding of how Web systems move from the micro to the macro scale would provide a better understanding of how they could be

developed and what their potential societal effects might be.

conclusion

The Web is different from most previously studied systems in that it is changing at a rate that may be of the same order as, or perhaps greater than, even the most knowledgeable researcher’s ability to observe it. An unavoidable fact is that the future of human society is now inextricably linked to the future of the Web. We therefore have a duty to ensure that future Web development makes the world a better place. Corporations have a responsibility to ensure that the products and services they develop on the Web don’t produce side effects that harm society, and governments and regulators have a responsibility to understand and anticipate the consequences of the laws and policies they enact and enforce.

We cannot achieve these aims until we better understand the complex, cross-disciplinary dynamics driving development on the Web—the main aim of Web science. Just as climate-change scientists have had to develop ways to gather and analyze evidence to prove or disprove theories about the effect of human behavior on the Earth’s climate, Web scientists need new methodologies for gathering evidence and finding ways to anticipate how human behavior will affect development of a system that is evolving at such an amazing rate. We also must consider what would happen to society if access to the Web was denied to some or all and to raise awareness among major corporations and governments that the consequences of what appear to be relatively small decisions can profoundly affect society in the future by affecting Web development today.

Computing plays a crucial role in the Web science vision, and much of what we know about the Web today is based on our understanding of it in a computational way. However, as we’ve explored here, significant research must still be done to be able to engineer future successful Web applications. We must understand the Web as a dynamic and changing entity, exploring the emergent behaviors that arise from the “macro”