Evolving Information Systems

Browsing the Internet and seeing the variety of indexes available suggests we need multiple indexes because there are multiple valid ways to organize things. Hierarchies are useful, but a DAG structure is more appropriate to allow multiple references to things by alternate paths. In fact, a Web would be better to allow circularity as well - say, there's a novel concept.

Because we don't know how to organize all info, and never will, and personal organizations are valid too, the info needs to be distributed and self organizing, as the WWW is, in fact. Anyone can set up their own view(s) of how to organize things. But in addition, I think we can collectively create better approximations to the truth that satisfy most people most of the time.

The idea is to set up a system that may be organized over time by the users as they use it. News does it now via the groups for discussion of news and news groups. But the process is very slow. I'm thinking of something that changes continuously, incrementally as people use the system. Each time you use it, things would be a little different, but it should not change so fast as to become unrecognizable in a week, although maybe in a month or two, especially in the early days.

The information system would be a combination of hypermedia and news. A semi-permanent but evolvable structure would be at the top layers of the system, and news-like discussion would hang off of nodes at all levels, particularly at the bottom levels. As archivable information accumulates through the news-like discussion, it migrates up to the appropriate level.

The basic structure of the system is as follows: A web or network of information nodes is the most general and appropriate structure to represent interconnected information. Each node is any HTML text node which may reference any number of other nodes. Lists of links might be unordered or ordered alphabetically, or by some other key. Each node references the information content, and there is also a link for all responses to the node.

Given some initial organization, users would make incremental changes to the structure. Responding to a node creates a html node. The kind of response allowed depends on the node being responded to. A news node would simply have responses with no explicit qualifiers as to what kind of response it is. A voting node would have responses that are votes. Another response is no response. In other words, when a node is read without any response, that may be a partial vote to leave it as is. In that case, if you find something that you think should be deleted or moved, better say so.

The possible kinds of response could be customized by the creator of the node. The voting criteria could be customized. This system would be a conversation builder / electronic voting system / news system for all public information and discussion. Private or constrained group discussion should also be possible, thus replacing email.

People would maintain their responses in their own disk space. The archived structure and frequently accessed information would be maintained centrally, or parts of it could be distributed. Distributed caches of all nodes will become very imporant in the near future if WWW is to survive the onslaught of the masses, even without this evolving information system.

Other issues:

• What goes at the top?
  • The top could be replaced by vote from the current top.
  • There might be no particular top. Many alternative tops could be available.
  • Each person could have their own top node that links to what they want to see.
• How to add new public nodes?
• How to respond to groups of things rather than single things?
• How to insert new links at arbitrary points in text you are responding to?
• Contention from multiple simultaneous users. Who gets to insert where?
• Identifying new nodes, and changes to nodes.

Related Stuff

• HyperNews
• Idea Futures
• Self-Organizing the Web
• Hypertext Publishing and the Evolution of Knowledge

1 Untitled by tch3@mixcom.com, 1994, Oct 15

1 Directed Acyclic Graph by connolly@hal.com, 1995, Feb 24

1 Hypernews as a Directed Cyclic Graph by cvarela@isr.co.jp, 1995, Aug 15

...

2 DAG vs ADG by Halliday@ASU.Edu, 1997, May 04

... 1 Message(s)

3 Sources by gsarmien@neptuno.ujavcali.edu.co, 1997, Jul 09

3 Some additional ideas (Ingrid), from Paul Francis by francis@cactus.slab.ntt.jp, 1995, Apr 23

4 Crossposting and shared subhierarchies by cvarela@isr.co.jp, 1995, Aug 15

1 [++] The SELF-DISCIPLINE rules for crossposting by ralph@democracy.tiac.net, 1995, Nov 17

2 Crossposting was Usenet's downfall by cls@truffula.sj.ca.us, 1996, Nov 25

5 Worked on similar idea: see http://www.cee.hw.ac.uk/~calum/AnswerWeb/ by irene@icbl.hw.ac.uk, 1995, Oct 12

6 PEPI "Personalized, Evolutionary, Participatory, Information by bre@globalHome.com, 1996, Feb 13

7 Hierarchies by liberte@hypernews.org, 1996, Aug 18

1 Internet Archive by jthomas@webwonderinc.com, 1996, Oct 29

9 Some unsolicited comments on Messaging... by dlaliberte@gte.com, 1997, Oct 20

10 more: Re: Some unsolicited comments on Messaging... by dlaliberte@gte.com, 1997, Oct 21

1 Re: more: Re: Some unsolicited comments on Messaging... by andre@peak.org, 1997, Oct 23

1 Change of subject helps HyperNews, but not email readers, yet. by dlaliberte@gte.com, 1997, Oct 24

2 How much to archive by cpa199@ecs.soton.ac.uk, 1999, Oct 30

Add	to: "Evolving Information Systems"

Members

Subscribe

Admin Mode

Show Frames

Help

Web www.HyperNews.org