joey said:
>   I see no distinction between your model and mine,
>   other than what they're called.

i can't really say that there _is_ a difference between them,
joey, not until your model is fleshed out with a real live app.

in my version, a book is represented by alias files that live
in various folders.  the names of those folders _could_be_
considered as "tags".  but that's not how "tagging systems"
are generally architected, not if i understand 'em correctly.

what i'm looking to create is a simple system that people can
understand implicitly, and operate easily on their machines...

how that system is labeled is nothing but a semantic matter,
just as long as everyone understands exactly how it works.

i'll give people a "starter-set" of folders, but after that, they
can develop things from there according to their own aims.

if they want a category called "phat books", they tell my app
to create a "phat books" folder, then they start checking off
the books that they want to have represented in that folder.

i see this as more disciplined and restrained than tagging,
where idiosyncratic tags are more-or-less routinely applied.
but, you know, i guess there's nothing to stop a person from
generating many folders with just one or two books in each.

at any rate, it is this _personalization_ of the categorization
that i see as the main difference between folders and tags.
tagging systems usually operate in a social network arena.

and this is perhaps an important distinction as well, in that
i see my system running as an app on a person's machine.
although i think people haven't been too clear on it thus far,
it seems that most of you see this operating on a webserver.
as an aside, y'all might want to look at ning.com for a means
by which you can easily create a social-networking web-app.

now, it may well be that the best start-set of folders is created
via a tagging system, perhaps one that is generated on a wiki.
as i said yesterday, though, i'm more interested in doing this
_by_myself_, because it seems too difficult to get any helpers,
and too unwieldy to build a system that captures all their help.

(it's far easier to just write the program for the end-user and
leverage the work that's already done in regard to cataloging.
for instance, one of the first cuts will be on the _language_,
and i can determine that by simply checking that in the file.)

but, speaking of a wiki, i think what you would get from that
would be more amenable to my "folder" structure than tags,
because each "page" on the wiki would represent a "folder".

at least that's how _i_ would organize the wiki.  for instance,
i'd have a "beatrice potter" wikipage that listed all her e-texts.
and i'd have an "esperanto" wikipage listing all those e-texts.

(of course, you could also organize the wiki with each page
representing an e-text, and then apply the tags on the page.
but i think you would find that approach to be cumbersome.
again, until i can see an actual working example on your end,
it's difficult for me to comment positively or negatively on it.)

but since i'm doing my thing by myself, the architecture of my
catalog depends on being able to collect almost all of the data
_programmatically_, via computerized analysis of the e-texts.

the other source of information i will use in generating the
starter-set of folders is the catalog-structure richard seltzer
has set up over at samizdat.com.  (and, if i could recover it,
i'd add the one that david moynihan had at blackmask.com.)

to sum up, i don't want to spend a lot of time generating
the initial catalog structure, and i don't want to spend a lot
of time assigning e-texts within that catalog structure.  ok?
my third concern is that people can modify to their desire.

there are other concerns, too, such as being able to capture
any additional information that people might contribute in
the long run while modifying their catalog (the arena where
tags really shine), but my 3 main concerns are the ones listed.

also, a main goal of the starter-set is to give end-users a way
to quickly eliminate the parts of the library they do not want
to have downloaded to their machine, and give the rest of it
a basic structure that can be navigated easily and intuitively,
and i think the "folder" model qualifies well in that regard...

so yes, you might be right that the same program that can
administer the folder-structure might be an equivalent of
one that can administer a tagging model.  or it might not.
i know what my app will look like.  i need to see the other.

-bowerbird