11 September 2011

Decentralized Topology

In a generic way, browser is a term for an interface that allows navigation of objects [1]. Nonetheless, for the topic pursued on this post, the specific notion of browser is uninteresting: new browsers/interfaces can be implemented at any time but the inherent topology of the space transversed by them is fixed and defined by the space itself.

In case of the Web, the navigation topology is defined by the use of anchors and links. Those elements allow a page to reference it's neighbours and enforce a model of local authority where a page controls to where it links to. Problems show up as the inability or hard-time implementing back-links, collaborative information and social web. Additionally, certain content types do not allow links on it own, creating a web of pages rather than a web of contents.

The lack of navegability on web is not a new problem, and most of the times tools are created to circumvent those issues. The self need of search engines, colaborative sidewikis, automatic backlink networks, logging of HTTP referrers, advertisement networks, music suggestion services or even twitter rolls for hashtags. Though the underlying problem lies on the limitations of how Web topology works. Could a service be designed to provide an alternative navigation topology over web? A topology that goes beyond the limitations of local authority? Would such a service be really useful? And deployable?

Challenge: Design a decentralized topology system for URI navigation.

Some side notes:
  • By allowing the system to work with any type URI, schemes such as Magnet Links can be used to provide navigation to Content-only Networks such as BitTorrent.
  • Search engines have been working over several years on technologies to provide "navigation" of content using power of masses, can those ideas and statistics models be re-used to improve navigation quality?
  • Can social spectrum of individuals be explored? Ranking higher/collecting transverse information from your friends/peers?
  • Can such a system be applicable for continuous spaces and usable for AR purposes?
  • Can such mechanism be reliable and usable for different types of users, including presence of malicious users

Disclaimer: It is not like I am aiming at solving the problem, or even claiming it to be a problem, but I think the simple mental exercise is worth the time.

No comments:

Post a Comment