Gem of an online resource, I hadn't seen before: an unedited transcript of Ted Nelson's Where our hyper-media really should go!, from Doug Engelbart's Colloquium (2000), The unfinished revolution, strategy and means for coping with complex problems (via the great site, InfoDesign).

I always find Ted Nelson's insights valuable and interesting, and there are a lot in this talk. One thing he talks about briefly are computer "files", which I have suggested a number of times as being an achilles heal in various information schemes such as the semantic web, FOAF, and RSS. Here is what Nelson says:

Let me talk about files. They say that the fish doesn't see the water, nobody quite sees what an abdominal burden is put on the user by the very nature of files as we know them.

You have to put the data somewhere, and you maybe have to find some way of referring to it. The fact that files are stuck lumps in fixed places, with fixed names obtrudes into every way of the users thought and environment. Always to the detriment of what you are trying to do.

So that we have ever so many work around to the filing system. We have ways of naming small parts, like mailboxes and play lists. We have aliases and short cuts that give access to that whole stuck lump.

Nowhere in the system do we have a clean way of the individual parts that we should be able to refer to. That doesn't even get to the main problem.

The main problem I believe is that we have to deal with contents that over lap. We have to deal with showing and overlapping contents and structures. Interpenetrating connections among contents and structures.

In the talk, Nelson goes on to describe the user interface he has been proposing and developing as the solution. I have always been somewhat skeptical of this suggestion, but I respect that Nelson has working examples of these interfaces and can viscerally experience the difference between it and others. (Still, I am skeptical some level.)

Another Nelson resource online, mentioned in the article, which I also hadn't seen: the open source release of Udanax (as in formerly Xanadu). Looks like development stopped around 1999 (did it?), but it is nevertheless interesting. This may be a way to get real firsthand experience of Nelson's interface design.

Finally, another quote from the talk that I think is also important, on hierarchical data structures:

Parallel interpenetrating structure is the challenge. Yet for some reason, everyone has said computers are hierarchical. Why are they hierarchical because we made them that way.

Hierarchy is something that came down to us from military organizations in the Catholic Church mostly. I was told by one scholar that the reason that Aquinas was backed with the Catholic Church was some yabba-dabba, ding-dong fight. The point is that we have fetishized hierarchy as a kind of structure, thinking that this is a real structure, which is preposterous. It is a degenerate, trivial case, at which we have attempted to force everything else.

The whole point is to deal with parallel interpenetrating structures, and allow hierarchy, sequence, and tables. The three structures we are allowed on computers. To allow those three to be the trivial cases they really are and to deal with the hard stuff. To allow this trivial thing and saying that it is the real thing, and convincing the world that it is the real thing has been an astounding snow job.

This is really key: The whole point is to deal with parallel interpenetrating structures, and allow hierarchy, sequence, and tables.

Note: I added additional line breaks in both quotations from Nelson, to make them more readable than the unedited transcript.