[Sigia-l] Findability

[Jonathan Broad]

On Mon, 2003-01-27 at 04:06, Cunliffe D J (Comp) wrote:

> Whilst I
> accept Derek R's point that categories CAN be used to misdirect, mislead and
> market, I don't see that this HAS to be the case. I am also still not quite
> clear as to the practical nature of the alternative he is proposing. We also
> have the possibility in electronic environments to provide more than one way
> to access the underlying information - we can provide users with different
> categories representing different perspectives.
> 

Hear, hear!

I'm not sure why Derek R so absolutely refuses to acknowledge the simple
usefulness of the categorical.  I think it's a serious error on his
part--even on his own terms.  This has been bothering me for some time.

A category is nothing nothing more than a label that we attach to other
labels that we attach to things, in order to work with them (both labels
and things) at different levels of complexity.  They're just
tools--nothing more.  And they are a tool we have in common with the
users of our sites.

The "legitimacy" of a category is quite simple to assert: for a
particular user of the category, does the label clearly and distinctly
a) delimit a set of objects from the universe of objects and b) express
an attribute common to members of that set?

If categories conform to that definition for a relevant problem domain,
then they can be used to easily sort through a mass of information to
find appropriate results.  I.e. findability--full stop.  If they don't,
then *in that particular case* the categories have turned out to be a
bad tool--nothing more.  Bad tool, bad!  No biscuit!  

It does not follow that mass depression ensues, however.

The weakness of categorization isn't in it's supposed disjointedness
from the information in-itself, but in the phrases "particular user" and
"relevant problem domain" above.  

Like most language-systems, categories have speakers and listeners, and
if you lose the attention of the listeners you're just whistling dixie. 
This is also a weakness that has been acknowledged by the LIS and UI
worlds for at least twenty years. They call it "user-centered design".  

The purpose of this type of methodology is simple: to create systems
that better conform to the "mental model" of the users of said systems.
In other words, to maximize the usefulness of information retrieval by
presenting an interface that is as "natural" for the user as possible. 

So we must listen hard to users, and build a holistic connection between
their views and needs and whatever our clients are trying to offer. 
Which means we must also listen hard to what our own information is
trying to say.

Easier said than done, but possible.  What Derek proposes, when that can
be determined (in the case of Napster, for instance) is not only
impossible, but schizophrenic.

Derek R seems pretty fixated on Bordieu (that one quote in particular)
so I thought I'd be fair and make a challenge to him on his turf.

I assert: the notion of 'findability', which is really just an
aggregation of a lot of the insights of library workers over the last
century or two, is not contrary to Bordieu's theory about 'habitus'.  

For my interpretation of the "habitus" when this first arose on this
list, see: http://www.info-arch.org/lists/sigia-l/0202/0085.html

For discussion, Derek:

1) Categories are not 'obedience rules' as Bordieu thinks of them,
imposed on information (or a user) from without.  If they were, they
would fail to work at all.  

1a) A categorical label is a just node in a constellation of possible
'semantics', constructing a 'regularity' for the habitus of a given
population of people (users of a system).  It's neither good nor bad in
itself; it's effect depends on the way the regularity fits into the
dispositions of the habitus.

2) It is incoherent to base information design (speaking) solely on a
single document (information object) either, or on the basis of any
single user.  That is a schizophrenic project--finding attributes
'natural' to a thing, but idiosyncratic enough to resist 'totalizing' in
a category.  It is completely against the grain of the insight of
'habitus', as well, because the idea of an idiomatic habitus (without
regular flows and structuring structures between people) is
contradictory.  See also Wittgenstein's thorough refutation of the idea
of a completely 'private language'.

Derek, your fear of "systematization", of the "common" aspects of things
that allow categories to be useful is causing you to make the opposite
error--atomicity.  

No text, to shift to a Derridian key, can be fully isolated.  A
corollary is that no system of categories will ever completely abstract
their domain.  But that doesn't make categories useless, any more than
the vagaries of quantum mechanics makes us incapable of building
bridges.

So creating "navigation" on the sole basis of individuals' "encounters"
with information is incoherent--unless you pretend to know all users so
intimately that you can route around the need to talk in terms of a
common world.  

This is a much bigger mistake than saying you understand something
common about a network of text.

Jonathan Broad