All my best ideas are thought of by someone else first. Dave3 uses the term Tree of Knowledge this morning like I did on Halloween. He ties it to distributed directories as is picked up as MetaDirectories at WIFIblog. I think this is great but there is one problem. They appear to be talking in the plural. To be very fair, Dave Winer is either there or close - he wants the directories able to talk with each other through a standardized interactive format. He wrote on 20 October:

1. Decide on a format for a directory. It should be XML-based so people can use any text tool to edit them. I designed OPML for this purpose, but if you want to use another format, I won't fight you on it. This is too important to have the usual fight over the bits on the wire.

2. Build software that renders data in this format as if it were a Yahoo or DMOZ directory. All environments should have well-tested efficient renderers, commercial and open source.

3. When this software encounters a node that includes another directory, include its hierarchy in that directory. These inclusions are what determine page rank, just like links in HTML pages.

4. If you run a search engine, index these files. Use page rank to determine which is shown first. Don't segregate these files, include them in the returns for HTML and all other formats you support.

5. Evangelize. Get academics, librarians, researchers, etc to produce data in this this format. Link and organize.

What is required, however, is not a bunch of people setting out directories of what they know but a single centralized place where links can be placed to anything of interest based on a universal recognized taxonomic system like the Dewey decimal system. Individuals creating their own directory format will cause chaos. Automated keyword RSS bots may overcome this as they will fills the index of directories with content according to the taxonomic scheme rather than the names associated with the files by the creators of the files. What is important in the use of a standard taxonomy of ideas is the ability to browse for associated knowledge categories on an equal level to the information already known or to move up the hierarchy, go laterally and then dive down into a neighbouring specific field of knowledge.

What we do not need a forest of knowledge. We need one tree. If not, we will perpetuate the central failure of the web - disorganization. It kills lateral thinking.