Developing a personal schema

posted by Jason Kottke Feb 09, 2002

I need some advice. I've been trying to develop a personal schema/taxonomy** and I don't really know where to start. I'm not looking for tools or anything like that, just some general advice. I guess another way to put the question is: if you had to organize all the stuff that a person comes into contact with, how would you do it? Do you have any experience in doing this, say, within the context of The Brain, a Wiki, or categories for a personal weblog? How did you go about doing it? Got any advice?

** After asking a few folks what the difference between a schema and a taxonomy is, I think what I'm really looking for is a schema of my world, that is, a system of describing the people, places, and things that I come into contact with. Once I develop that schema, I can then use it to build a topic map. (And even after all the searching and asking I did, that's probably not really what I'm talking about here, but all the info I could find on these subjects was very specific (mostly related to XML) whereas I'm thinking about this stuff in the most general sense possible.)

Reader comments

scoFeb 09, 2002 at 1:11PM

Have you looked at humanmarkup.org? It's not exactly what you describe, but they are trying to create a markup language that could describe one's "emotions, physical descriptors, proxemics, kinesics, haptics, intentions, and attitude".

leonardFeb 09, 2002 at 2:04PM

Well, if you're looking for developing a schema, there's always Dublin Core (yeah, it's over the top). Less over the top may be the dmoz RDF Dump.

How useful is any of this for personal use? Dunno, I recommend going with a Wiki, as it's semi-structured and much more convenient (ie, it keeps track of relations not hierarchy - most data one keeps track of rarely fits into one category in a schema anyway). Tavi and PhpWiki both have improvements over the original like better DB support, better versioning, page lockdowns/editing control, and other such whatnot.

Bonuses over say The Brain or Touchgraph (which admittedly is way cool) is that while they are not as visually impressive, the Wiki's are viewable and editable completely from any browser, much more flexibility with mapping, useful versioning, and of course, the data is fully available in a relational manner (the DB) for other uses.

Aaron of MontrealFeb 09, 2002 at 2:11PM

You realize, of course, that you're peering into the abyss, right?

That said, do not try to define an all-encompassing schema. It will only end in frustration. Rather, try to think about the relationship, and more importantly the commonality that one "thing"has to another. [1]

If you're try to superimpose some kind of default structure on top of things, better than it be as abstract and generic as possible. From there, you can begin to build case specific taxonomies as they occur without breaking the basic "skeleton". [2]

[1] http://groups.yahoo.com/group/weblog-devel/message/76

[2] http://aaronland.net/weblog/archive/3891

graywyvernFeb 09, 2002 at 2:52PM

my stuff is organized chronologically: first, in order of
frequency of referral (daily, weekly, several times a year),
then the long-term files, regardless of subject, in the order
that i came to them.

thus i can find everything i was thinking about, writing,
or doing in 1983 in one box. they are associated in my
mind, & if i can't remember exactly when i can usually
remember something else that can be dated.

but my books are shelved according to size.

VoyagemanFeb 09, 2002 at 4:05PM

Chronology, Priority, Subject , Relevance - I go back and forth at whim so flexibility is key. Two software tools that make it a breeze:Correlate and Mindmanager

Sam TrychinFeb 09, 2002 at 5:19PM

Ugh... I agree with Aaron of Montreal... most of the
taxonomy-creation/display tools seem to force you
into a generic structure of their own flavor.. the results
usually seem unsatisfying. Plus, the navigation interfaces
often get in the way of the actual content. (I find The
Brain annoying, for example.)

I'd just sit down with a piece of paper and a pen, and
start drawing little circles & lines. Do you want an automated viewer? Seems like you can get quite far
with lists of links, or some very simple flash and/or
image maps if you want it to be spatially meaningful.

Of course, if you want to enter things into a general-
purpose database on the fly, and cross-reference them
according to your category system, to have your
taxonomy inteface autogenerated later... then you're
back at sqaure one again, coding a taxonomy-viewer
by hand.

I guess I'd just keep it simple, if you want it stored
in a DB and autogerated... stick to a very simple
HTML interface, or some simple Flash. I'm mainly
arguing against starting with someone else's taxonomy
system and trying to fit your info to it.

Guess that didn't help much....


Rikard LindeFeb 09, 2002 at 5:33PM

Jason, your blog is probably a very good taxonomy. It's sorted chronologically and there's a good search. Guess you could add other sorting mechanisms, like alphabetical sorting and amount of traffic or even better... by incoming links (mmm Google). I'd love to see your top- ten postings (as judged by the rest of the blog community) as a sidebar.

KylerFeb 09, 2002 at 5:58PM

Check out Foundations of Complex-System Theories : In Economics, Evolutionary Biology, and Statistical Physics for a broader understanding of how to approach complex systems. It's also a good read just for fun.

Check out The IBM Glass Engine for an example of an excellent way to put an interface on categorical data.

victorFeb 09, 2002 at 7:33PM

if you really want to start at the top: check here

Erik KaiserFeb 09, 2002 at 7:45PM

This is off topic I know, The site looks great, I am just commenting on the whole layout and everyhting. How does this comment script work? did it come with your weblog software? and if so, is there a method of adding such a service to a weblog by blogger?

BlakeFeb 09, 2002 at 8:26PM

Ask a Librarian, it's what they do.

Morten BtbuktFeb 09, 2002 at 9:33PM

Your brain has allready organized your world. You memory puts everything in boxes that are more or less available to you due to the recentness of their use, and their general and emotional importance. It even stores away the little things that aren't really neccessary, but it's all there. How detailed would you want your system? Would you want it to have information about everyone that has written these comments, and everyone that has sent you an email? If you were to put your life in a system you'd use more time writing it down than acually living it. How can it be worth it? Or even possible?

mathowieFeb 09, 2002 at 11:22PM

Jason, if you elect to go with some crazy high-falootin' RDF framework, talk to Paul Ford. IIRC, ftrain is all xml stored data transformed with xsl on the fly, and it's all organized with some form of rdf. Paul's got little bloggy things, stories, and shorter posts.

JoshFeb 09, 2002 at 11:54PM

I think Moodstats is doing something like what you're thinking. Looking at their XML format might be useful.

Joe CrawfordFeb 10, 2002 at 12:09AM

Thought #1 is just to plop it all in a /life/ directory and hyperlink it all and sic the ol' reliable google at it. Heh.

Thought #2 is to think of using this "lifelines" concept as some mechanism to organize things. I tend to think graphically, so graphing location makes sense. You could easily do something similar by placing objects on axes - perhaps individual works could live on a personal vs. professional? As far as a storage retrival method chronological is a natural way to organize data.

Thought #3 is that metadata is wonderfully useful, but *adding* metadata to every work you've ever done seems like a recipe for madness. More than that, metadata can change meaning -- say you were to apply metadata to all the works of Picasso while he was at the midpoint of his career? The meanings of some of those categories would change as his career morphed again and again. If you want the catalog to have *meaning* beyond the dewey decimal system, you have to understand that the meanings of your work can change with time. Which is not to say that there are not immutable things you can say about a person's life output.

Thought #4 is that I'm rambling.

Actually, the nearest thing I can think of was the cataloging of Andy Warhols estate when he died. They put together reams of catalogs of nearly everything he collected and did. This included things like his art and photography to doodles and even just things he collected and junk he happened to own.

Here's an article on cnn that's quasi-related.

I'm done. Hope this helps you sort out organizing a life's work. Should be a piece of cake.

jkottkeFeb 10, 2002 at 1:00AM

You realize, of course, that you're peering into the abyss, right?

Oh yes, well aware of that. It's dark down there. I'm cold and frightened.

Jason, your blog is probably a very good taxonomy.

It's good chronologically, but there are no relationships between anything. Even with the search, it's hard for me to find anything, even when I know that it's there. Plus, a lot of the information that I come into contact with and want to "remember" somewhere doesn't get posted to my site. I need somewhere else to put that.

Your brain has allready organized your world.

That is a good way of putting what I'm looking for. I like Aaron's advice a lot...start very general and strengthen connections as you see them develop (which is very much how the brain works chemically...you remember things when pathways are laid down in the brain and reinforced through repeated use). Flexibility is a key here as well...if a relationship between two things (or two groups of things) isn't working, I'll need to be able to change it quickly and easily.

A quick shot at my schema (loosely hierarchical with no relationships):


  - kottke.org
    - comments system
    - Silkscreen
  - 0sil8

  - Typography
    - Fonts
    - Software
    - Cool signage
  - Flash
  - Color
  - Styles
  - Web
  - ???

  - XML
  - Perl
  - MySQL

  - Books
  - Movies
  - TV
  - Magazines
  - Internet

That's a start, but it seems too obvious and not quite right somehow.

jkottkeFeb 10, 2002 at 1:15AM

I love what Art & Culture does with relationships. You can drill down into film hierarchically, but then you can hop across categories with adjectives like violent, which makes connections between Sophocles, John Woo, and the Marquis de Sade. I think if I can find connections like that to sew all my various ends together, I'll be on the right track.

plasticFeb 10, 2002 at 5:01AM

slightly-updated old-skool:

- people (analog)
- people (digital)
- places (analog)
- places (digital)
- things (analog)
- things (digital)

As you can probably tell, I'm a big fan of the KISS approach.

BrunoFeb 10, 2002 at 5:39AM

A schema is something that grows and evolves, i don't think you can define it upfront.

You said no tool recommendations, but i'm using vanilla because its a weblog and a wiki, which allows me to create new spaces, link them, and it puts them in a context automatically. For example, when i create a musician, i simple add ...releasing on *record label* (in that syntax)...
Now the artist is 'connected with that label' (example, notice the backlinks on the left side). You could also add 'violent' as a space and simple link it (*violent* movie), the backlinks are there without further editing. It also features direct Touchgraph support.

So i don't define categories before i need them, they grow over the time. My mind feels more comfortable with the thought of connecting things and putting them in a context, rather then storing it in predefined categories.

Cheers, hope it helped a little.

TimoFeb 10, 2002 at 6:47AM

Michael Fergusson and Dethe Elza have a go at defining what they call 'chaos tools' in this discussion

Juerg LehniFeb 10, 2002 at 8:29AM

Christian Langreiter, the creator of the vanilla log system (http://www.langreiter.com/space/Vanilla) uses touchgraph to create a graphical representation of his log.

vanilla is not only a chronological log but also has associative links between entries, so it may resemble the structure of the human brain more.

This combination of vanilla and touchgraph is called vanilla-vista.

IgorFeb 10, 2002 at 8:41AM

As a cultural anthropologist (doing fieldwork) and part-time journalist I have been struggling with this question for years. I've been inspired by how Pirsig describes the ordering of notes for writing a book in 'Lila', but most ideas I got from various writings on qualitative methodology (data collection, data analysis, data storage, theory development, etc.). I think a very good start to find a "systematic approach to 'soft' data" is provided in articles and books by Strauss and Corbin (Qualitative Analysis). To prevent inventing things that have been invented and thought over already: have a look at the site of Atlas Ti, a qualitative analysis tool that has been developed by people with over 30 years of experience thinking over this (your) question. Only reading the documentation would be a perfect introduction. The site: http://www.atlasti.de/atlasneu.html
When I saw this tool for the first time, I was surprised to find many of the answers to this (your) question that I have developed myself over the last ten years.
Igor (the Netherlands)

EditorFeb 10, 2002 at 10:50AM

I think that thinking XML overcomplicates everything and makes you say "crap! I want to move this from here to there! or copy it!"

The simplest approach seems to be keywords. With each entry you make, be it to your Web log or to your otherwise, enter a small number of keywords. Each new entry will pull up a list of these keywords with checkboxes alongside your input area so you can use old keywords or enter new ones.

The beauty lies in which keywords you choose and how you then organize them. You can give the keywords themselves a hierarchy, so that they can become more like broad titles ("fonts") or more like specific descriptors ("Din Mittelschrift").

This way, the parts can truly connect across subject and time and whatnot, without that directory-like structure of the XML stuff. Plus, you can always play with how you interpret the keywords, extending it indefinitely in form and function.

I have a simple example on a content management site I built if you wanna check it out. Good luck.

vanderwalFeb 10, 2002 at 11:18AM

I used something similar to Xplane's Xblog use of multi-metadata categories to classify and find similar items. Whump uses a similar approach. Items, like life, can have multiple classifications, which these approaches take into account. These classifications can also be tied back to a heirarchical approach should that be desired. I like the heir-classifications you have outlined and they can be remolded and modified as time passes and needs change. Building a system that has some elasticity in it will provide greater advantage than a set rigid system.

BrunoFeb 10, 2002 at 11:25AM

Sorry to push vanilla again, but one example to explain how simple it is, compared to classic 'Content-' or 'Knowledge Management' Tools:

Day entry:

Yesterday i finished reading the *ADC Yearbook*.

- You created a space called "ADC Yearbook" and next you define it:

Its the annual *book* of the *ADC*. What i like most about it is the *creative* usage of *fonts*, especially *Silkscreen*.

Buy it now on *amazon:amazon*.
Official Website: *http://www.domain.com*.

- Lets say you already created the spaces "book", "creative", "fonts", "silkscreen" some time ago, vanilla connects it automatically. If ADC doesn't exist yet, the system tells you, and you create it. The *amazon:* is a macro you set up once.

That's all.

No more additional editing, no checkboxes, no meta-information. Done.

KaraFeb 10, 2002 at 11:56AM

Try Ranganathan's facet analysis, which traditionally organizes information into five main aspects or facets: PMEST (personality-matter-energy-space-time). You break each facet down into separate isolates and then can express more complex concepts by combining together different isolates from the entire scheme.

christinaFeb 10, 2002 at 12:24PM

The real question is why are you doing this, and what do you want to use it for? We can come up with better solutions if we know what it's for.

My recommendation (wihtout knowing the answer to that) is to buy a big pack of small post-its.

write one word for all your stuff on each post-it. "friends, fonts, matt haughey, bread recipe" and so on. no organizatin, no order, just pure spew. go as long as you can. if you have something that is kinda of a repeat but slightly different, write that down too.
now spread them out. I like to do this on a giant whiteboard (thus the post-its, since whiteboards are usually verticle and I want something sticky. you could use index cards or bits of paper if you want to do it on the living room floor).

Now start grouping like items. Sometimes you may find some of your words seem more like headers for groups ("friends" might be a group for "matt haughey")

Voila! You have the beginning of a taxonomy. When I do it on a whiteboard I like to draw circles around things, find overlaps (get all venn-like) add headers, and add new items. and so on.

card sorts rule for organizaing the metadata of your life.

pinkstarFeb 10, 2002 at 12:46PM

Everything in it's place. That's just creepy.

AdamFeb 10, 2002 at 2:00PM

If it's catagories you want, you might look to Thomas Jefferson as square one. He had a pretty decent structure for his personal library that could be modified to suit.

However, if you're really looking for a way "to organize all the stuff that a person comes into contact with," you're looking for nothing short of a philosophy, my friend.

It's about time for philosophy and the Internet to begin breeding some sort of hypermap for describing the new way in which an increasing number of us interact with the "world." Try to have something together by Wednesday.

rebecca campbellFeb 10, 2002 at 2:39PM

I'm sure you've heard of Everything2, but is that what you're trying to accomplish?

Actually, it sounds like you want something along the lines of the "Knowledge Web" that James Burke is working on. Have you heard of this? You can listen to him describe it at http://smithsonianassociates.si.edu/tapes/burke.htm, or just do a Google search for "james burke knowledge web".

I saw a demo of it at a lecture of his, and it's really amazing. His version seems to be primarily focused on famous people in history and the connections between them, but the structure itself could be used as a model for showing the stuff we come into contact with (and how they relate to one another).

redrickFeb 10, 2002 at 3:04PM

i agree with christina. the most important question is what are you doing this for. if you can answer that question, then you can begin thinking in schema/XML. that's just because schema/XML requires implying a structure. but i think if the core issue is to create a taxonomy, you shouldn't start from solving schema/XML issues. you should start from a relational database that allows for variations in the way data is related to each other, such as many to many, many to one, one to many. etc. because that is the kind of problem you will run into when you attempt to link anything to anything. you will realize that there is no one way to structure the world through language. why? because language is symbolic. a 'sign' to one person may not be the same 'sign' to another person. look at the problems you regularly encounter when using Yahoo categories. i've worked with Thinkmap applications and these are the issues that i've encountered whenever we attempted to "tweak" an existing schema. basically, you can't hardcode your schema if you are talking about mapping "everything". you'll need a dynamic schema, a something that is allowed to change as it grows. if you aren't able to create a dynamic schema, or if you choose not to, then you will have to make one decision that will rule how to structure "everything". what does that mean? you have to ask yourself, "what are you doing this for"? if you can't answer that question either, then you might just have to suck a dictionary into a database then add an abstraction layer on top if it called "kottke organization", which is not to be confused with "the organization of kottke". know what i mean? ;)

neuroproFeb 10, 2002 at 3:31PM

I did not have time to go through the site yet, but The Brain may be a good model after all :-)
You may try to imitate it (I don't mean the program alone, but your very own brain).

Sensory areas (people, books, objects, events that influenced you), motor areas (lifes of those whom you influence, products that you made), association areas, memory, limbic areas, and many many more. Not everything needs to be interconnected with everything, and although you will need to include loops, there should be an overall feedforward pattern. The connection patterns will need to change with learning as you don't necessarily want to keep everything forever. By the time you are done with the model, you will probably be ready to build an artificial human.

Rod WelchFeb 10, 2002 at 4:18PM

Working intelligently requires technology to create and maintain organizational memory with organic structures that convert information into knowledge. As you point out, a taxonomy, or ontology, of subjects is needed, which is more art than science, leading to the proposition that moving civilization beyond information technology requires a culture of knowledge that synthesizes art and

Many people report SDS is effective for organizing the record, as seen, for example, on 010916....


....noting "amazing memory," which is only possible with an organizing
methodology and technology to support it.

While organizing the record effectively and continuously to manage the garden of knowledge is indeed a big challenge, noted, for example, on 000221....


...none-the-less, given the events of 010911....


...and the collapse of the economy because too many people are having too many problems, as evidenced by Enron's bankruptcy related in the Powers' report reviewed on 020204....


..., there seems to be no real choice, but to tackle organic subject structure.

If we want to work our way toward national security and economic health, we need to work "intelligently," which requires organization, analysis, alignment,
summary connected to details and feedback, see POIMS....


This is disagreeable to many because organizational memory reduces deniability, and so seems to increase accountability. The nice thing about intelligence,
apart from being fun, is that, while intelligence does guarantee success, the chances of success and getting credit for improving productivity, earnings and stock prices, rather than being blamed for mistakes, is far greater than the common strategy of using bad management that relies on ignorance, fear and denial, as seen in the Enron case...


Thanks for timely reminder about the opportunity to rise above current difficulties by application of art and culture.



Derek of PenmachineFeb 10, 2002 at 5:09PM

I think the world is way too complicated for generalized categorization, and our brains just can't handle that.

JoshuaFeb 10, 2002 at 6:06PM

I agree with Christina, the card sort approach would work well to sort out a taxonomy.

The eventual problem is that with a system as complex as your life, new groups (ideas) will always emerge. From IAWiki:

When the desired changes cannot be reliably foreseen, and particularly when the target domain is computationally too complex for automation and thus relies on the understanding and development of the people involved, then top-down, preplanned approaches have intrinsic shortcomings and an emergent approach is required.

But that's getting into a whole other deck of cards that you might not want to play with.

redrickFeb 10, 2002 at 7:00PM

Joshua makes a great point about emergence. and that's also where Aaronland's point is related. one strategy is to establish a default structure --based on a limited number of nodes and edges-- and then allow the taxonomy to "emerge", based on your own feedback and/or other user feedback. the results can actually take an implicit form, where the most common groupings rise to the top. you can then set a cutoff point so it returns a smaller result set. hey, maybe you can write a simple sorting application, a-la christina's suggestion, where each post-it is actually a SQL insert. you insert or post-it all the items and then start ranking them in categories. the problem though, is that whenever you add more categories, you will need to re-rank all the previously inserted items to see how where they fall in the updated "bigger picture".

jedwardsFeb 11, 2002 at 4:39AM

You may wish to try The Literary Machine. From the site: organize random thoughts and snippets of information. Once in the database, words and concepts are linked together by the software. It's customizable enough to do what it sounds like you want.

Greg RitterFeb 11, 2002 at 6:58AM

Great thread! Thanks, Jason. This is a problem that I've been gnawing at for years. It began simply as a need to better manage the thousands of bookmarks that I've collected since ~1996. Unfortunately, I've been unsuccessful in finding what I need. And since I'm no coder, creating it myself isn't an option, but at one point, I even began writing up the requirements for the software I was imagining. I'll see if I can dig them up (be easier to find the document if I had the tool I was imagining!)

Basically, what it boils down to is I want the Yahoo engine but expanded to handle files as well. I'm continually surprised that the DMOZ engine isn't available. I'm not interested in the data; I want the tool. The closest I've found is YIHAW, a Zope Yahoo-clone that I could never get functioning correctly.

MCFeb 11, 2002 at 8:50AM

I've found that most of the stuff I thought was worth saving really wasn't.

I can't find about a year's worth of syllabi and course descriptions, but I'm going to be a better teacher by recreating them than relying on the old crap.

I do want google for my hard drive, though, just for sloggin through old email.

Peter Van DijckFeb 11, 2002 at 11:48AM

Is the goal of this metadata to make it easier to find things? Or to help you think somehow? Can you tell me what the goals are, I'm interested since I'm contemplating building a tool...

Lukas BergstromFeb 11, 2002 at 12:12PM

I've been tackling exactly this problem at my site over the last week. Things became a lot easier when I decided to allow each entry to belong to multiple categories (the crucial "well, duh" step.) I can now easily create a new category/subcategory when making an entry. The resulting taxonomy may not be clean, but I'm mostly interested in finding things fast.

Jacob ShwirtzFeb 11, 2002 at 12:16PM

Maybe this is not what you're looking for but how about organizing everything according to emotions? Humorous, quirky, sad, romantic, indifferent, etc. You could have multiple levels or even dfferent amounts of each. Just trying to think of a way to give emotion and feelings to what would otherwise become just a cold collection of facts, memories, etc. The difference between "met Bob today" and "pissed off that Bob showed up late to our meetings because he was getting these funny hairplug things...."

JoshuaFeb 11, 2002 at 12:35PM

Getting back to the emergence idea, Tinderbox looks like it could be a great tool for this type of problem.

Tinderbox is a personal content management assistant. It stores your notes, ideas, plans, and dreams. It can help you organize and understand them. Tinderbox even helps you share ideas through Web journals and web logs.

It allows you to organize your notes by arranging them, coloring them and linking them. Agents can be created that automatically scan notes, looking for patterns and building relationships. If it really works the way they describe it, it's going to be one amazing tool. Victor Lombardi has mentioned that he might even combine Tinderbox with a Wiki. Now that would be cool.

Tinderbox will be available for Mac in February. Windows version is "coming soon".

Bill SeitzFeb 11, 2002 at 1:58PM

It might be helpful for you to define "requirements" by providing some sample questions you would ask your organizational system after everything was in there and "organized".

Also, how does this relate to any PIM (Pilot/Outlook) software you're using?

My bias is toward a small number of general categories, and using links (wiki backlinks) to move on from there...

I'm also moving a bit toward the attitude of "don't bother bookmarking, just look it up again on Google". Though I also blog for the purposes of bookmarking.

And I'm putting all my non-PIM stuff into wiki now. I've modified it enough to render weblog entries beside WikiName titles.

JohnFeb 11, 2002 at 2:10PM

Check out the following:


You can also check out the book by David Gelernter


and his web site which seems to have changed to something else!!!


Malcolm DeanFeb 11, 2002 at 3:56PM

Info Select (www.miclog.com)

Avoid ontologizing when the future need for information is unknown. IS's neural search is like Google for your personal yellow stick-ems.

sjaFeb 11, 2002 at 7:45PM

Jason, you're probably familiar with Pyra (the product). Well, I've been thinking for some time about creating a--er, "Personal Life Manager" (?) based on Pyra's model. How about expanding Pyra's tree, adding other types of objects (e.g. people, appointments, emails...)?

Dave WinerFeb 12, 2002 at 12:42AM

User an outliner.



PeteFeb 12, 2002 at 7:20AM

A schema is a network, not an outline, so you may find that outliner-based approaches fall short.

The discussions at http://www.llamagraphics.com regarding how to use their product LifeBalance often touch on this sort of stuff, particularly as it pertains to time management, but because of LB's focus, also runs a little more broadly than just time management.

Sylvain CarleFeb 12, 2002 at 9:48AM

I would add from experience that the hardest part is not the base schema (altought it's a daunting task) but the possibility to start small, grow, cross-link, add categories and re-categorize...

If haven't found a system that does that well... Super categorized systems creates too much empty branches and are almost always conceived from a machine standpoint. What I am looking for is a system, that places the editor (or better, a group of editors) at the center of the process... Everything2 is close but is a bit chaotic.

Anyhow, this is a fascinating abyss to stare at. I will be watching this thread, but I think it could be worthwhile to create a mailing list on this topic.

Mr rlygssonFeb 12, 2002 at 10:16AM

Check out Peter Merholz's rants on "Faceted" I.A.

Also check out the Dublin Core meta-data project. They've been thinking about this sort of thing for a long, long time.

toddFeb 12, 2002 at 10:20AM

I think it's pretty much an equivalent problem of how to search the web.
What seems to work is to write pages
that have high content, link them together using URL/wiki as you
see fit, yet let the search engine also link stuff together based
on access patterns and content/relationship. Adding categories/meta tags
can be helpful, but doesn't scale over time.

I think what needs to be let go is the idea that a top-down organization can
be cast over your life.

kellanFeb 12, 2002 at 10:39AM

Wow, great comments, and lots, but then All Smart People Fancy Themselves Librarians. :)

My suggestion is don't try to come up with the One True taxonomy, but figure out 6 to 8 big areas, keep it perfectly flat, and observe.

In 6 weeks refine your model, throw out what you're not using, add some sub-topics to the remaining subjects, and repeat.

Think of it as refactoring.

MCFeb 12, 2002 at 12:48PM

You might enjoy and benefit from _Information Anxiety_ by Richard Saul Wurman, if you don't already have it.

Paul SnivelyFeb 12, 2002 at 1:15PM

You may also wish to take a look at the RDFWeb project. They have a distributed photo archive, came up with the FOAF (Friend of a Friend) RDF schema, and some other goodies that seem related to your "people, places, and things" ideas.

Bill SeitzFeb 12, 2002 at 2:40PM

If you haven't seen it, this recent Jon Udell article on Topic Mapping is a good fit. http://www.oreillynet.com/pub/a/webservices/2002/01/01/topic_map.html

James ArcherFeb 12, 2002 at 5:50PM

I think a possible solution to this dilemma would be to resolve the hierarchical vs. chronological schism that troubles so many webloggers. We want to put information in chronologically (i.e., what we're thinking, when we're thinking of it), but that results in a stream of jerky gibberish, since our brains jump from topic to topic.

This would require some tricky programming, but it could be resolved by blending the two systems. Have a hierarchical set of categories (a la Yahoo or Dmoz), and then associate a category (or subcat, or subsubcat...) with each blog entry.

The result will be a mapping of your brain over time. You'll start to see trends and patterns that you might not have been aware of. Certain nodes on the hierarchy will swell, while others will remain untouched.

Such a CMS/Blog-hybrid would allow users to view it either chronologically, or hierarchically.

Anyway, that's my 3 cents.

Shawn YeagerFeb 12, 2002 at 9:07PM

I've also been in search of the ultimate brain-storming/organizing tool. I'd been very happy with TheBrain, but recently returned to my roots and bought a PowerBook. Regrettably, TheBrain isn't available for OS X. :)

So, I went searching. What I found, Shared Space 2.0a6, is still in Alpha, but looks very promising:


Joseph RuvelFeb 13, 2002 at 1:39AM

I have looked into this some. Keeping it high level, I think what your looking for is a map. A detailed map yes but a map. Nodes connected to like nodes the connections being the most important part. Start seeing connections in things you didn't really know were there. I haven't looked at it depth but Six Degrees might be the type of product you are looking for

MichaelFeb 13, 2002 at 7:13AM

The taxonomy you came up with seems servicable. Taxonomies in general, however, suffer from the rigidness of the relationships you create. They also make you suffer when you want to make changes.

Other indexing methodologies are more useful and flexible. Thesauri are interesting alternatives. You come up with a set of "Authorized" terms that you will use to classify your stuff. Define relationships (broader/narrower/related). It gives you a flexible way to index. Creating linkage to related terms could be done programmatically. We are working on something similar to this (a loose set of requirements for blogging classification based on indexing concepts) over at the Drupal project.

jkottkeFeb 13, 2002 at 7:48PM

Thanks for the comments and the links everyone. The approach outlined in Jon Udell's article is exactly what I was thinking of doing before I brought all this up (thanks Bill. Each chunk of data gets some keywords, and then you can search/sort on those keywords and even build relationships among them (for example, the keyword "emergence" would be related to "ants", "games", and "city planning", much like the Wiki-ish Everything2 does it).

I like everything about it: it's quick, it's dirty, it's bottom-up, and it avoids the "taxonomy lock-in" that several people brought up.

asellkeFeb 15, 2002 at 12:08PM

Six Degrees might prove useful? Check it out.

Mary KindredFeb 16, 2002 at 9:02PM

I opened this comment box in a new window so I could search it with Edit>Find in IE. No Find if you just right click in the pop up. The minds of library types work this way. ;-) Anyways... I was surprised no one mentioned mind mapping . Along those lines, TreePad sounds interesting. Lockergnome says "It looks (and works) like Windows Explorer, in the sense that you can create an infinite amount of cascading topics and subtopics." Sounds like the learning curve isn't steep. Nope... I don't work for either one. I ran across the link for this program an hour or so ago and then read your blog entry a while later. Synchronicity.

Janne AukiaFeb 18, 2002 at 10:16AM

Gzigzag, which is based on ted nelsons ideas, is also worth loking at. It's a semi-structural system, which should allow complex data structures.

Janne AukiaFeb 18, 2002 at 10:20AM

I'd like to have a huge mind-map, with subfolders and list-elements and automatically updating elements, that get their material from the web. And generic property-list elements with easily definable fields (kinda like filemaker or hypercard - cards). When I'd have all possible ways of looking at hierarchical information, i'm sure I'd fnotice what are the best ways of displaying different kinds of information.

Christophe KotowskiFeb 19, 2002 at 8:31AM

I have been using The Brain (Personal Brain to be exact, their desktop application) for more than a year. I use it to point to internet pages, documents stored on my hard disk (html, PDF, doc, pictures, movies,...), diverse files and folders from my hard disk and applications I have on my PC (to start them). I tried hard from day one to create an intuitive and easy-to-use structure of my thoughts. I have used special characters to introduce a hierarchy between thoughts.

The problem is that one tends to amass an amount of links that tends to get so big that memorizing and maintaining the whole thing is getting difficult after a while. Finding past thought is getting difficult, the whole web of thoughts is so extended that navigating around is slow and cumbersome. The multidimensional visualization does not allow having a sense of "place" contrary to simple logical structures we are used to (like a drawer, a folder). Also getting an overview is difficult as it is never possible to see thoughts that are further than two links away. A 2-D visualization doesn't enable to get a grasp of a multidimensional model.

Conclusion: I think it's a nice tool, but it's only useful for limited amount of information. Our own brain is more apt to handle the huge number of associations it generates. Just think of it: how many thinks do you relate with the word "dog"? I can think of a dozen and more immediately and I'm sure that if I tried long enough I could relate hundreds of thinks to that word. No tool will ever compete with this ability in terms of versatility and performance. I'm now compiling a classic link-list in html, it's more efficient.

TubesFeb 19, 2002 at 1:44PM

Greg Ritter:
Hi - found your post at Kottke's place re: bookmark organization. Try PowerMarks (powermarks.com). I'm about to register my 30-day demo. It rocks. Rather than a rigid heirarchy of folders, you assign relevant keywords to bookmarks as you save them. The process is so quick it's no hassle at all. And importing from IE or NS is slick - it turns your folder names into keywords for the appropriate bookmarks. One of the most useful programs I've ever tried.

James Archer: for a CMS/blog hybrid see Kevin Fox's fury.com

SoulhuntreFeb 22, 2002 at 8:06AM

There was a program a while ago (years) called "Ecoo" that was a lot of what youw anted - it was perfect for my use but the company died out and the product died.

It was an outliner with user definable keywords and an item could exist in multiple places int he outline (automatically updating in each place when edited).

This was a great way to combine the outline and network organizations.

Mark CraneMar 06, 2002 at 9:44AM

David Pogue, in a recent Macslash interview, was talking about "EZnote" for the mac as a tool that helped him organize incoming bits.

There's a piece of software for professional ethnographers, Bad interface but very, very powerful. It's called Nvivo. Not as web enabled as it could be. I will trust you to use google to find both of these pieces of software.

Sebastien PaquetJul 23, 2002 at 7:44PM

Here's how I do it in my OpenMind. I use a wiki with one page per concept or thing that interests me. I go bottom-up and at first I don't worry at all about where a given item will end up relative to other items. Most items start out as islands, unconnected to anything else.

Then, later on, if I happen to notice that something is related to an item that is already present in the Wiki, I create a new page that represents what the two have in common. The new page links into the items that got me thinking up that page. (This makes the scheme much clearer than directly linking "sibling" pages, by making the relationship between them explicit.) My wiki supports backlinks, so it is easy to use it to reach something whose name I forget by starting from something similar that I recall and following (back)links. For instance, I have begun making pages for people I find interesting. At some point it became apparent that a few of those people shared a particular interest, so I made a new page for listing people with that interest, and put pointers to those people's pages in it. (I'm not sure all these people know one another, so I might actually be seeding a new community of interest...?)

When I'm pretty sure that a kind of page comes up consistently, make up a category and mark pages in traditional wiki fashion when they clearly belong in the category. I'm pretty conservative in making categories. I only use categories that I'm pretty sure won't change quickly as my worldview evolves: Person, Document, Quote...

Finally, in some cases I'll link from particular concepts to more general and more specific concepts. Again this requires clear thinking - I won't do it with ideas that are still fuzzy. It's a multiple inheritance hierarchy: each concept may have several generalizations. I think Know-how Wiki exemplifies that idea pretty well.

Just as I do with my physical brain, I am inclined to link an item to related items as densely as possible, so that there are many ways to reach it. Think network, not tree. Links may be added at any time during the lifetime of the page. I think the central challenge lies in avoiding linking sloppily, as in the long term I believe it'll get me confused. Overall the thing is pretty anarchic, but order emerges over the long term, at least for the pages that I end up visiting repeatedly. Actually I'd be afraid if it were to become very well ordered; I would take this to mean that I'm no longer learning anything...

I call this a personal hyperlinked informal ontology. It pretty much feels like an organic, living thing. The description I have just given is pretty abstract. Navigate it a little and you'll probably going to get a feel for it.

Ben TremblayFeb 07, 2004 at 9:00PM

My "gnodal" LiveJournal page gives some sense of what I'm working on ... something like ToughGraph but with more meaningful relationships ... not just "connected to" as though "see also", but "conditions" and "is conditioned by", or entailed, or caused ...
... you can see why I'm working to bring Bayesian theory to play here.

If we could mediate discourse with something that make explicit the tension between _thymos_ and _psyche_, I suspect we'd have a method that not only exchanges data and information but that would also give rise to solidarity ... something like empathy.

This thread is closed to new comments. Thanks to everyone who responded.