In other words, could you work out that good, average, and poor are
one disjoint group whose union is equivalent to essay, and that long,
medium, and short constitute the same kind of pattern? For simple use
cases it should be relatively easy, but there are edge cases even
there. Consider the case, for example, where short and poor contain
the same essays, but the other tags have a more random distribution.
It would then be difficult to tell whether good, average, and short is
one subclassification scheme, or whether good, average, and poor is.

Since tags are basically sets, as you can tell by the fact that I've
been talking about disjoint and union functions on them, I also
thought about using set operations to browse tags. You could add a new
tag in by unioning it or by intersecting it. You could put in
differences, and so on. The only problem is that since most people
aren't all that well acquainted with set theory, making the interface
intuitive would probably be quite difficult. I was surprised that
there wasn't any better tag browsing software off the shelf, as far as
I could find, anyway.

Some things I found:

http://indev.ca/MailTags.html

http://www.bronsonbeta.com/mailappetizer/beta/

http://www.queuesoft.jp/product/MailUnreadStatusBar-e.html

> I am interested in how you're going to continue tagging your
> essays after a period of sending them via email.

Disclosure: the substantial part of this post was written in circa
January 2009, and I just updated it a little and posted it so that I
could catch you up in the weblogomachia. The point about tag modelling
and browsing is still valid, but no longer relevant to the context in
which it arose initially.

The -machia suffix means war in Greek.

Aha.

Do you plan to continue using it with your email posts?

The problem with tagging, and any classification system really, is that they are temporally rooted. At any point in time when you're thinking about creating one, you have a back catalogue or things that potentially need (re-)classifying, and a potential mount of unrealised material that will need classifying. Having to go back and forth changing things sounds like a lot of busy work to me - and yet, I have been wondering how I might search through and browse my old emails if I'm sending so many instead of keeping notes or working on formal prose.

My thoughts always come back to wabi sabi, and an embrace of full-text search and "smart folders" or some similar feature. However, I also figure that it can't hurt things to tag them, or categorise them, anyway - as long as you don't get distracted by it, take it too seriously,  and as long as you realise that it's temporary and you may only use that system for a single email. All it does is adds a bit of data to email, and as long as you remember GIGO - you're probably not making things worse. I particularly like Apple's smart folder idea because it lets you just stuff everything into one location, and then organise it add-hoc post-hoc. So maybe I will add that to my list of things you should try to do if you're going to do it.

Steps to painless organisation:

        * It's never permanent

        * It shouldn't be a distraction

        * It shouldn't be taken very seriously

        * Remember GIGO if you're going to alter the originals

        * Favour ad-hoc post-hoc sytstems

Can you think of anything else?

tags: wabi sabi, categorisation, gigo, technology, email

> Do you plan to continue using it with your email posts?

Nope. I never used it even before email, it was just hypothetical.

Only date based classification schemes work for me, and I don't even
use those on the web except as embedded metadata. Check out the
Vanilla Web Design essay that I sent you for why.