Just started reading. Most of the items on your list of things to fix I was thinking about for quite some time. Besides the comments, maybe you could write a blog post about your system and experience with it? I for sure would be interested.
Do you think that solution to the whole personal knowledgebase problem should be solved from a single app perspective or is it better to think about this as a sort of ecosystem - one "module" for adding new things to the system, other for statistics[0], sync etc.? Also, did you consider semantic part of such system? So we can write programs that can make sense of all the things we have in the system?
[0] I was actually thinking about using machine learning to somehow finding the most important things in the wiki and suggesting tags/place in hierarchy of new entries
Actually my main takeaway is that organizing information shouldn't occur in a single app. It needs to take advantage of the web as much as possible. The model I want is basically small "apps" joined by hyperlinks -- exactly what the web is.
The primary app is a wiki with unstructured text, outlined/indented text and hyperlinks. But, as mentioned, information comes other forms too: structured data like bookmarks, spreadsheets, images, etc. The web lets you express all those things (spreadsheets being the weak link, since it involves executing code). Reinventing it all within a single app would be a lifetime of work.
The web already has common patterns and facilities for log analysis and so forth. If you just make it a web site, you can track all your actions with web log analysis. Whereas if you make it some "app", then you have to reinvent the analytics.
I definitely considered doing more textual analysis, stuff like PageRank, and even semantic analysis. I think it's interesting to explore but I don't the cost/benefit works out right now.
I have over 2000 pages, but probably 500 or less are "active". That can mostly be managed with full text search, and nothing smarter. You just have to make the search fast, which isn't hard with that amount of data.
But, if you want to write more code and programs against the data, a web site is a good format for doing so. I basically use the Unix philosophy of independent programs and data, rather than the "app" philosophy of data locked up behind code / user interface.
My philosophy is just to optimize for speed and ease of both writing and reading. And then you use it more. And the more you use it, the more valuable becomes. Just like with learning itself, you start to make more connections once you have some existing knowledge/content to graft on to it.
I think the most important thing is the content, and not the tool, and as long as the tool is lightweight and fast, you will keep up the habit of entering content and (just as importantly) retrieving it.
That link is full of wisdom, thank you for posting it! I never met anyone who wrote their own wiki and used it for so long.
I was surprised the biggest thing for you was speed and full-text search, and also by this comment: "I think people forgot how to make a simple web app with a form and plain buttons."
The biggest thing for me is learning. Not so much categorizing everything correctly.
