title_test.gif

New archiving system!

Inspired by Mark Bernstein's site, J. Nathan Mathias' Notebook of Sand, and others, I considered my archive system and realized that it wasn't going to scale.

Scalability Problems

  • The "archive" note in Tinderbox was going to keep growing, leading to an archive.html that was both broken in its template usage and expanding with each entry
  • I would have no easy way to make a calendar with month-by-month summary pages like those on the two blogs mentioned
  • As the archive grew in my Tinderbox file, any time I opened it in outline view I would get the whole thing. To work with notes after the archive, I'd have to scroll past a list that is already longer than a screen.

Solution

So I set up a system like J. Nathan's, with year and month notes, and entries are children of the months. I'll just keep the latest month open, and since the years are children of the archive note (thus keeping descendedFrom(archive)=true), my agents still work. The challenges were twofold.

Legacy Links

Problem: Others have linked to existing entries. While Tinderbox easily accomodates this dramatic architectural shift, the static links of the web don't. I have to leave pointers from the old system.

Fixed: I made aliases of all the existing notes (not that many, really) and left them in the archive note. I exported the whole kit and kaboodle one more time. That created proper pages like I used to have. Then I tossed all the aliases in an unexported note. That keeps them around, their exported html is still hanging out in the export folder for the site, and if I ever forget and delete the legacy files I can pull the aliases back out again and re-export.

Cascading Actions

Problem: The old archive note had an action: publicationDate=today;hot=false; where publicationDate kept the note organized by blog publication date and hot marked entries which needed attention. But now I had a system which would expand with new year and month notes, and the months needed to have that action. Obviously I can't be bothered to remember to type that every month.

Fixed: I set up a few new prototypes. The prototype *year* has the action Prototype=*month*, and the prototype *month* has the action publicationDate=today;hot=false;. Now whenever I make a new month and drag it into a year, it becomes a *month* and gets the action. In point of fact I also gave the old archive note the action Prototype=*year* in case I forget to assign that, too. It took me a while at Tinderbox Weekend Boston to get my head around how to use cascading actions to enforce a data structurein that manner.

The Main Page

Problem: The main page is an agent that draws the first 7 children from the archive. After the change the children of the archive note are the years and months.

Fixed: This took a bit of kludge. I set up an agent called "pseudo-archive" with the following agent query:

  1. descendedFrom(archive)&!(Prototype=*year*|Prototype=*month*|Prototype=*note*)

... got that? It's in the archive but it's not a year, month, or subtopic. None of those should be listed by themselves.

Then I changed the main page agent query from #first(archive,7) to #first(pseudo-archive,7). Naturally I left pseudo-archive not exporting its children and using textOnly.html as its template so that its presence on the site is masked.

Bill Bly

Bill Bly wrote one of my favorite hypertexts so far. And he seems to present hypertexts that work in a way that makes sense to me... as ways to connect fragments; tools for creating the gossamer spiderweb strands that bind and separate our more substantial thoughts; as masses of information for which the discovery of the structures is as integral to learning as is the information itself. There must be a better way to articulate that. I'm sure I'll find it as I read on.

I just discovered his blog within his personal site, and that's part of what has me writing this entry. I realize now that the people section of t.org is growing with a bias toward blogs. The blogosphere is fertilizing its growth, and it is responding ... blogotropically?

Bill has a page about his work with hypertext. He approaches it from a refreshing angle. Rather than looking into hypertext systems as tools for new work, he is approaching them as the lesser of evils: as attempts at software which are with regard to the way we think perhaps less flawed than the word processor, spreadsheet, or operating system. That's another take on one of the things that keeps me interested in the field-- I think hypertexts more accurately represent the ways we think and communicate.

Memex and Beyond Web Site

The Memex and Beyond Web Site seems to be closed but valuable. It's a snapshot, circa 1996, of the key people, institutions, papers, and conferences in the field of (academic?) hypertext at that point. It's very much what this site would aspire to be-- a thorough index of the field. Based on what's there and the stated goals, it would be lovely to see it finished, but its structure is well-done even by current standards, an internet geologic age after it went quiet.

It's interesting because it shows very vividly what I call projections of a hypertext. Every page of the site is a multidimensional structure flattened into linear form, and each item on the page is a vertex, a link which allows you to see that same item in another projection. Each link serves to highlight the multidimensional structure of the text by serving as the corner between facets, the juncture of flattened images. The way you navigate the site keeps the structure in mind.

I have to post about it now as I keep finding myself coming back to it, sometimes for material and sometimes to consider its form. As I dig through the site I find myself frustrated by the web, wishing that the links would yield two or three destinations each: that item in each of the other planes of the site and as an item by itself.

memex

The Memex is probably one of the more influential machines never to have existed... certainly as far as hypertext is concerned. The memex was a theoretical analog computer, a sort of microfiche library which would track the activity of the user and allow the user to create navigable links between works. It was first described in the 1945 Atlantic Monthly article "As We May Think". Wikipedia has a good entry with more factual information. Dynamic Diagrams has produced an excellent Interactive Animation of the Memex. The memex is often cited as an origin of hypertext alongside Project Xanadu.

I think it's interesting as a case study in the articulation of nascent concepts. Watching the video of Vannevar Bush demonstrating the concept of the memex is like watching the silent movies of early airplanes. The ideas are there but the technology hasn't caught up yet.

(In terms of t.org, I suspect that this note will be more useful as a connective and referential node than it will be informative on its own.)