Retrofitting the Web

<p>Hypertext, which we generalize as hyper<i>media</i>, is text plus <i>links</i>: an extra degree of freedom to move around, <i>other</i> than starting at the beginning and reading all the way to the end. By this definition, the ubiquitous World-Wide Web is a mediocre specimen. True, the Web is a cosmopolitan system of unprecedented scale, with instantaneous global reach, that is characteristically easy to deploy and extend. <i>Hypermedia</i>, however, is nothing without links, and the Web’s links are <i>brittle</i> (they routinely break), <i>one-way</i> (you can see links <i>out</i> but not links <i>in</i>) and <i>untyped</i> (anemic support for telling what a link <i>means</i>). This results in big clumsy documents with few if any links (that often go nowhere), no backlinks, ambiguous, ad-hoc data semantics, and poor reuse of content. The <i>sparsity</i> of the Web, as it comes off the shelf, means:</p> <ul> <li aria-level="1">Readers have to read a lot per unit of information gleaned</li> <li aria-level="1">Authors need to maintain separate documents for distinct audiences</li> <li aria-level="1">Developers have to continually reinvent the same basic functionality</li> <li aria-level="1">Information duplicates and goes stale, causing people to be misled</li> </ul> <p>All these points aside, this bias spells a missed opportunity for building out an entire <i>universe</i> of creative expression.</p> <p>Not only are there copious <i>practical</i> reasons for addressing these issues, doing so would dramatically boost the Web’s capacity for communicating complex ideas, making subtle arguments, and fostering innovative methods of storytelling. Turning the Web into truly <i>dense hypermedia</i> entails being able to <i>reliably</i> point to a very large number of very small objects, and doing <i>that</i> in turn means designing a system—a <i>protocol</i>—for maintaining the stability and continuity of the addresses (URLs) exposed.</p>