Theory: One-layer web-design
Jorn Barger
frustration of clicking thru page after page, links-list after
links-list, without ever finding any actual _content_...
So I spent Sunday prototyping a new solution to this problem, using the
domain of James Joyce studies. Previously, I'd created a popular page
of 'unabridged' Joyce-links [1] that tried to sort out the best sites,
and describe what's good about them.
But I sensed that there was still a lot of messy redundancy there, and I
finally got the notion of doing a 'Joyce portal' that ignores the
concept of _sites_ altogether, linking instead to all and only the
'leaf' nodes-- the real-content pages-- and sorting these into a dozen
categories that are pretty obvious: general, biographical, timelines,
short bios, images, multimedia, interactive, shopping, magazines,
publishers, centers, libraries, and links [2].
Because they're so finely sorted, these hardly need more than a word or
two of description, though I think I'll add a companion page that covers
the same links in the same order with more detailed critiques.
This notion of creating ONE page that links directly to every relevant
webpage, bypassing all the navigation trees inbetween, seems to me like
a big winner for the Web in general-- imagine that for any topic that
crossed your mind you could find an 'x portal' that linked to
everything, making it trivially easy to zero in on the likeliest pages
of interest, AND THEN to compare them to find exactly the one that best
meets your needs.
[1] Old unabridged Joyce-links page:
<URL:http://www.robotwisdom.com/jaj/jajweb.html>
[2] New-style one-layer Joyce 'portal':
<URL:http://www.robotwisdom.com/jaj/portal.html>
--
XML for webpages is like plastic bags for comic books.
I edit the Net: <URL:http://www.robotwisdom.com/>
Ehud Lamm
organize infomration - for others.
The wbe apporach, is by its very nature distributed. This means that if I
think I can trust someone, I can link to his page - and not have to manage
all the info in his site myself. He will do it. The same of course applies
to linking to my site etc.
Building good sites, for example shallow sites that don't require too much
clicking is of course a worth while goal. But keep in mind that the
tangled web is a result of the way people work with the hypertext tools
they have. It is much easier convincing someone to put put a page of links
(which he already has), plus some material of his own, than to conince him
to find the time to deisn a "perfect" information portal.
Had we strived for perfect we would have arrived at very little content...
Ehud Lamm msl...@pluto.mscc.huji.ac.il
http://purl.oclc.org/NET/ehudlamm
Jorn Barger
> [...] It is much easier convincing someone to put put a page of links
> (which he already has), plus some material of his own, than to convince him
> to find the time to design a "perfect" information portal.
If people like the one-layer design, ther won't be any convincing
necessary. The point is not to get rid of inept sites, it's to bypass
their inept parts while accessing their real content directly.
Stuart Yeates <sye...@cs.waikato.ac.nz> wrote:
> this would, of course, give incredible power to whomever created that
> page---the power to define the ways we read, think and act about that
> topic.
> the argument ``in a competitive market like the net abuse of that
> power would lead to alternative sites being created'' doesn't hold,
> because as soon you have more than one site, you quickly revert to
> the current state of affairs.
You're not getting it-- sites that pick and choose don't qualify, and
will be ignored. By not filtering, the page-author gains no personal
power whatsoever.
Jorn Barger
> as soon as you start creating pages of links to all ``X'' you need a
> system for choosing your X's (topics within your system). this implies
> an ontology, a worldview and your system becomes culturally biased
> towards the culture(s) that worldview is founded on.
this is pure FUD... so what's _your_ hidden agenda?
Nick Kew
> page---the power to define the ways we read, think and act about that
> topic.
That already happens with Yahoo. Their power is not based on anything
inherent in their directory, but in the fact that all the journalists
off-the-web have heard of them, and regularly cite them.
This is already a dangerous monopoly, that can make or break a company.
> what is needed (IMHO) is _librarianship_ on the web. librarianship
> not using the web to enable access to paper documents, but using
> the web to enable access to electronic documents.
Agreed. Again, we have that already, but it seems to be a lot better
developed in "intranet" situations than on the web. HTML's decidedly
impoverished hypertext model (dangling GOTO links) doesn't lend itself
to catalogueing material that is not under the librarian's control.
--
Nick Kew
Jorn Barger
> *) not use cuttering -- currently sites such as yahoo tend to
> list groups of sites alphabetically. there have got to be
> better ways of presenting lists of sites, i'm not sure what
> they are, but i'm sure there are some.
this i agree totally on-- an alphabetical list is almost always a wasted
opportunity.
best-first is the optimal approach, and one that works fine even within
the one-layer-portal paradigm.
Nick Kew
> name (think Xerox).
Not at all.
Xerox is (or rather was) identified with photocopying, which is presumably
tha analogy you have in mind. But the difference - and this is absolutely
crucial - is that the suppliers of xerox kit (i.e. office equipment dealers)
are also the suppliers of other brands.
Yahoo *is* the main supplier to a lot of websites. That is, as I said,
by virtue of the fact that it is referenced in all the off-net media.
Every journalist has heard of it and - in their usual sheeplike manner -
they always cite it.
That makes Yahoo a monopoly supplier for firms seeking to do business
over the web. They are the channel for - how shall we say - meta-hypertext
links from non-Internet media.
> as the fact that they are months behind in adding submissions) which
> flow mainly from the fact that it relies on human maintainence.
I was deliberately avoiding any comment on the merits (or otherwise)
of Yahoo. They are indeed an important issue, but not the one I was
addressing.
> if someone such as the library of congress were willing to throw
> a hundred librarians at the task, i have no doubt that a far better
> internet directory could be built. some of the things i'd like to
> see in a web directory
But they'd need to throw billions into PR to achieve public awareness
before they could break Yahoo's monopoly. As I keep saying, the
quality of the directory is not the issue: it's the public profile
(think Microsoft for an analogy to that ...).
--
Nick Kew
Grimly Fiendish
wrote:
>
>what is needed (IMHO) is _librarianship_ on the web. librarianship
>not using the web to enable access to paper documents, but using
>the web to enable access to electronic documents.
>
Have a look at www.eevl.ac.uk for an example of this in practice -
librarians with an engineering knowledge providing specialist gateways
A little searching should turn up SOSIG for social sciences and also
MARS and ADAM - all based on the same concept.
GF - (Librarian turned webmaster)
Jorn Barger
> | best-first is the optimal approach, and one that works fine even within
> | the one-layer-portal paradigm.
>
theoretically, maybe, but practically all it implies is a human observer
who feels favorable and unfavorable reactions.
Jorn Barger
> my understanding is that as soon as you have a human classifying sites
> you'd run into the kinds of problems yahoo has---too many sites, too
> few human observers to classify them and too fast a rate of change
> in existing sites.
My Joyce-portal has 250 links. I could verify them all with a day's
work, and I always have bots watching for new pages.
I'm not trying to create all portals on all topics.
> having an automatic classifier removes these costs, but forces an
> explicit ontology (which may or may not require more work to build
> than the sites would take to classify, depending on the number of
> sites).
This is called 'solving the AI problem'. Tell me when you're done...