SAMPLE ABSTRACT #1 TITLE: "Internet Governance In Crisis: The Political Economy of Top-Level Domains" PRESENTER: Milton L. MUELLER Department of Information and Systems Management Hong Kong University of Science and Technology Clear Water Bay, Kowloon, Hong Kong E-mail: Tel: 852-2358-7633 Fax: 852-2358-2421 TRACK(S): Social and Policy ABSTRACT: This paper connects the debate over content regulation of the Internet with the controversy over the creation of new top-level domains (TLDs). It attempts to explore the question of whether domain naming will become a way of classifying content, and, if so, whether such an approach would make it easier for system administrators, users, or governments to block access to unwanted content. The paper is not concerned with the political or moral case for or against censorship. Instead, it focuses on the inherent practical problems facing attempts by any user or government to take a more selective approach to accessing Internet content. The paper begins by assessing the inherent limitations and problems associated with the two existing methods of content control: 1. User-defined software blocking filters; and 2. Government-imposed censorship laws, licensing regulations, and proxy servers. The first method is represented by the Platform for Internet Content Selection (PICS) and RSAC standards. The main problems with this method are that they are difficult for ordinary users to implement and most sites are likely to be unrated. If unrated sites are not blocked the filter will be ineffective; if they are all indiscriminately blocked the value of the Internet as a site of innovation and free flow of information will be greatly diminished. The second method is being used by various governments. Government attempts to categorize and regulate content at the source are ineffective because of the transjurisdictional nature of Internet communication. Attempts to block content via proxy servers are problematic because they impose a uniform set of standards on all users; moreover they face the same problem as unrated sites. The paper then considers the Postel draft and other proposals for TLDs. The Postel draft proposes to create many new TLDs. The issue here is not the merits of that specific proposal, but the more general idea of associating TLDs with certain types of content. The possibility of an ".XXX" or ".PORN" TLD is one obvious example. This raises the intriguing possibility that TLDs could -- perhaps -- simplify problems of content classification and rating. The rest of the paper is devoted to exploring and modeling the feasibility of this idea. We can assume that content providers would have a strong incentive to cluster into TLDs that would offer fairly clear signals or reputations to potential users. On the other hand, there would be borderline cases or disputes about the appropriateness of content for certain audiences. In these cases one domain that some users found objectionable might lead to widespread blocking of the entire TLD by users or governments -- a negative externality. The administrator of this TLD might have an incentive to expel the objectionable subdomain holder to protect the interests of others within the TLD. There is also the problem of domains that straddle several categories of Internet content. To what extent will a natural, voluntary "invisible hand" process lead to coherent, useful categorization of Internet content, a categorization that could then be used by parents, sysops, and others to guide their access to the Internet content? That is the question the paper seeks to answer. KEY WORDS: Internet, Internet Governance, Content Regulation, Top-Level Domains, Postel draft, Censorship, Platform for Internet Content Selection, domain naming, Censorship, Content Classification, Ratings Systems. SAMPLE ABSTRACT #2 TITLE: "An Interactive Prefetching Proxy Server for Improvement of WWW Latency" PRESENTERS: Ken-ichi CHINEN (* Lead presenter) Nara Institute of Science and Technology Information Network laboratory 9816-5, Takayama Ikoma City, Nara, 630-01 Japan E-mail: Tel: 81-7437-2-5342 Fax: 81-7437-2-5349 Suguru YAMAGUCHI Nara Institute of Science and Technology Information Network laboratory 9816-5, Takayama Ikoma City, Nara, 630-01 Japan E-mail: Tel: 81-7437-2-5342 Fax: 81-7437-2-5349 TRACK(S): Applications Technology ABSTRACT: As the Internet gains in popularity, demand for high-speed information services has increased, especially since the World Wide Web (WWW) is now the standard Internet. The time required to retrieve information depends on the latency of the network, so reducing latency is becoming increasingly important. Prefetching technology can reduce latency, and many prefetching systems have been designed and implemented for high-speed information services. Performance of a prefetching system depends on its strategy. Our system is designed to prefetch, through the proxy server, the information that surrounds the information that the client has requested. The prefetched information can then be shared through the proxy server. This paper describes the various aspects of proxy server prefetching: its strategy, implementation, analysis using actual information and requests, and evaluation. The WWW is an interactive service. Therefore, prefetching should also be done interactively to achieve quick response and the retrieval of fresh information. People access the WWW through a client (also called a browser) that issues a request to the server or proxy server. The user can then access other pages linked with the requested page. If, for example, a client program can retrieve a linked page before the user accesses it, the retrieval time is reduced. Retrieving linked pages at the time of each request is called interactive prefetching. The link linked page is retrieved by parsing the pages HTML. Because the modification and reinstallation of clients are servers are complex and annoying, we designed our prefetching systems as a proxy server that stores its results as a cache. By prefetching in a proxy server, the prefetching results can be shared among users, and conventional client software can be used. The prefetching proxy server, called WWW Collector (Wcol), is implemented in C language, generally under UNIX. To reduce the load on the host from prefetching processing (CPU, I/O and others) implementation is by the pre-fork technique. The program is separated into modules that communicate by fd-passing. We forecast the performance of interactive prefetching by a simulation using the access log of the proxy server for the campus network of the Nara Institute of Science and Technology (NAIST). The content of the log was 8519 requests. The pages referred from the requested page was 0 to 1272. The included information in referred pages was 0 to 424. It is difficult to calculate an actual latency and response time. We simulated performance according to the hit-rate. As a result, the hit-rate increases as the prefetching number increases. The hit-rate roughly followed an exponential function. We evaluated whether the prefetching proxy server was useful by testing its operation in an actual environment, replacing the proxy server in the NAIST campus network with our prefetching proxy server. The proxy server prefetched 10 referenced pages on the requested page and prefetched 10 units of information included in the referenced pages. The hit-rate of the prefetching proxy server was 64.3%, 1.67 times the hit-rate of the caching proxy server. The prefetching proxy server was 2.46 times as fast as the non-caching proxy server, and 1.62 times as fast as the caching one. In conclusion, we describe an interactive prefetching scheme for the Web to reduce the latency perceived by users. In this scheme, the proxy server attempts to guess the user's next requests and prefetch them. The results of our simulation show that with interactive prefetching, one can expect a large hit-rate and a hit-rate function approximating an exponential function. Our evaluation shows that interactive prefetching reduced latency and that the prefetching proxy server was faster than a caching proxy server.