After struggling for many days, I lowered my ambition for migrating to a personal wiki: I won’t split my existing Notepad files of assembled tiny notes – no separate wiki pages.
1. The idea was very tempting: separate pages for each tiny note would have enabled granular addressing for sophisticated cross reference linking, with automatic back-linking for free. I learned that this radical approach is the right one for new, unreflected notes that do not immediately fit into an established category. But as soon as the periodical reviewing has yielded assemblies of aggregated notes, this procedure starts to become frustrating:
Once the connection between two tiny notes has been reflected upon, the mere link is no longer satisfying, at least in the case I descibed last lime (#115) when the notes and the relationship are still too vague to allow a pregnant title or a cue to reactivate an association chain. Such a connection needs context.
- The context at the link’s target can be reasonably established. With the Wacko wiki I adopted, indivual tiny pages can be assembled using “include” statements (I can even achieve a rudimentary two-dimensional newspaper-like effect by indenting some of them), and I can set anchors to jump to these included units within their aggregated context.
- But what about the link’s origin needed for the backlink ? This context is lost, unless I forget about the backlinking automatism and manually include an anchor and the reverse link for each forward link.
(Digression: This leads me to critically think about the promises of the Semantic Web which claims the transition from document level to content level, and from humans to machine agents focus. As long as the content level (text snippets, named anchors) are still so miserably supported for the humans, I doubt the soon progress. I don’t think we can skip the intermediate step of human linking on snippet level. And for humans, backlinking is still an unresolved problem. (In the first years of the Web, some believed that HTML would only be a transition from Gopher to “Hyper-G” which then already included support for backlinks.) Today it’s still even worse: also the forward link to a named anchor only works if it is not incidentally near the bottom of the page…!)
For my wiki migration, this missing backlinking support caused me to give up the radical plan to split up all notes into separate wiki pages. If I wanted to administer my notes in this atomistic way, I would probably choose a web application such as zettels.info (Zettel = German for paper slip) which looks impressively streamlined for minimalistic effectivity.
2. So what about a higher aggregation level? A surprising affordance of Wacko Wiki is its Table of Contents support. Once the hard work of grouping the items and finding provisional headings for them is done, the reward is an automatically generated TOC including links to the sections (albeit not permanent anchor names), and clicking them has such a nice animation effect showing the scroll-down process that it causes me great (infantile) pleasure each time I watch it. Being fascinated with these headings (although their hierarchical nature does not much suit the wiki network ideal), a temporary idea was to use these headings as a basis for splitting my wiki pages. But the lost context effect of backlinks became even more evident on this lower level. In fact, I did not realize for a long time what made me hesitating to radically split my files until I played with the moderated version.
As a result of my wikifying efforts, I think I have found a way now to deal with new, unclassifyable note items. I can link from several larger contexts to them, use the backlink search to keep track of them (although this requires an additional step of a find-in-page operation), and finally create a new aggregate page for them or include them in an old one, depending on the emerging focus. So, the effort was not in vain.