[Sigia-l] integrated catalogues?

[Alexander Johannesen]

Hi,

On 10/25/05, Skot Nelson <skot at penguinstorm.com> wrote:
> Is this a failing of technology, or of the sub-30 second attention
> span young kids (and I can't believe I just used that phrase, btw)
> have today?

Google has shown that with low hardware requirements and high
smartness you can acomplish amazing results in a fraction of a second.
They do this because they've thought about the problem in new ways.

The traditional search is going to be dead, I'm pretty sure of it. The
reason it persist is because it is easier to implement while smarter
is usually harder, but not *that* hard. :) An example is a search
service I recently created that had a lot of clever ranking algorithms
and the use of thesaurii for magic lookups and so forth, but the
actual search was more traditional, with search response times ranging
from 1 to 10 seconds.

I've now taken the exact same service and done a clustered search
across static XML data (that can change at will, just like any other
DB) instead of a RDBMS, and now I see times raning from 1 to 10
milliseconds! Application and interface identical, but going from
"indexed RDBMS search" to "indexed file clusters" - which pretty much
is what Google did - boost performance a tenfold.

It's not about attention-span of our kids; it's the smarts we adults
apply. If someone has shown you that what used to take a minute could
take seconds, are you going to ignore that fact and go on your merry
1-minute-per-search-is-just-dandy way? I thought not. :)

> I think that a world in which if what you want isn't ready
> immediately is not realistic. Academic research lives in a
> drastically different timeframe than google based trivia.

Only to the people creating the service; the users, the academics if
you like, are also affected by the Google factor. In fact, my service
described above *is* an academic search, and 10 seconds for a search
was borderline enough for us to try and be smarter and faster about
it. And seriously, it isn't that hard.

> > The fact that each is a
> > different database is indeed an excuse for it not working
> > particularly well.
>
> Indeed. It is not, however, an excuse for presenting results poorly.
> This is but one aspect of "not working particularly well."

Libraries have a long tradition of designing things that are hard to
use, cryptic to understand and generally produce sub-par results. But
the good news is that they're coming around. Death to Z39.50 and
functional specifications!!


Alex
--
"Ultimately, all things are known because you want to believe you know."
                                                         - Frank Herbert
__ http://shelter.nu/ __________________________________________________