Internet Technologies
StandardiSed Instant Messaging
E-mail may be one of the Net's killer apps, but in coming years, significant growth in interpersonal communications will be driven by instant messaging (IM). As with other Internet applications, IM's transition from a popular niche product to widespread use will be accelerated by standardisation efforts. An open IM standard known as IMPP (Internet Messaging and Presence Protocol) is on the table today, but AOL, which is estimated to control 90 percent of the total IM user base, is conspicuously absent from the list of supporters.
Recent scrutiny from the US government, including the Federal Trade Commission, has pushed AOL to submit an im standards proposal of its own, but it differs enough from IMPP that some time will likely pass before the various players can agree upon--let alone implement--universal IM capabilities. Nevertheless, we can be confident that in the next few years instant messaging--like e-mail, or the voice phone network--will permit users to communicate with one another regardless of the particular hardware or software vendor.
New Apps Exploit Connectivity
As the Internet continues to grow, entirely new classes of applications that take advantage of the massive number of connected machines are emerging. Consider Napster, a tool that allows users to share MP3 audio files on their systems and to search for and download MP3s from other Napster users' machines. The concept, of course, needn't be limited to MP3 files; programs such as Wrapster make it possible to share any kind of data files. Gnutella provides capabilities similar to those of Napster but without any central directory server.
A natural complement to distributed file-sharing capabilities is distributed computation. The idea behind distributed computation is that a really big problem gets split into discrete, independent chunks, which are then parceled out to individual computers whose owners have volunteered their idle processor time to the cause. In aggregate, the users' computers form a sort of distributed supercomputer. The concept was first popularised by U.C. Berkeley's SETI@Home project, a a piece of code that's now been downloaded by more than two million users. Though SETI@Home is a single-purpose tool designed solely to scour radio-telescope signals for signs of extraterrestrial transmissions, you can expect to see general-purpose mechanisms for distributing all kinds of massive computations. United Devices, for example, is a company that will use distributed computing for projects in areas such as bioinformatics research, drug design, and climate studies.
Making the computational resources and information on millions of computers available spells potentially big rewards for search engines in particular, because they face the daunting task of indexing an already enormous, fast-changing Web. Startup Gonesilent.com, with its InfraSearch technology, offers an intriguing example of how distributed search capabilities might evolve: rather than crawl Web pages as most search engines do, it lets site developers actively link their internal site-search capabilities to a growing, Gnutella-based, distributed search engine.
Other search technologies exploit the connected nature of the Web in different ways. Google's search engine, for instance, determines the relevance of results not by simple keyword matching but by analysing the network of links that reference a page. DirectHit's search technology uses popularity--tracking which result links people actually click on--to influence subsequent results. The Webhelp.com site provides live, human-assisted search. And researchers in the search space are examining more sophisticated ways of performing metasearches (searches of multiple search services); tailoring capabilities to the unique keyboard and display formats of hand-held and wireless devices; searching nontextual (audio, image, video), proprietary (PDF or PostScript), and multilingual material; providing natural-language search interfaces; and, of course, exploiting the structural information that XML can provide about a document's content.
