[Wikipedia-l] Limits to the non-paperiness of Wikipedia?

Fri May 30 14:25:00 UTC 2003

Oliver-
> On 29 May 2003, Erik Moeller wrote:

>> Long articles have several advantages:
>> * A topic is kept in context, eliminating the need to write a separate
>> intro for each individual article

> The context depends on the reader. What is the context of Chap's views on
> Stuff? For someone interested in the life of Chap, the relevant context is
> his life, and so they would want that section to be at [[Chap]]. For
> someone interested in Stuff, the relevant context is that subject, and so
> they would want the section to be at [[Stuff]]. Of course, there should be
> mentions of his views in both articles, but to duplicate the whole lot
> would cause problems. I think it would be better to have a separate
> article on [[Chap's views on Stuff]], and have links to it from both
> articles.

I would prefer to have

=Chap==

'''Chap''' is ...

== Chap's views on [[Stuff]] ==

> Okay, maybe that's an unfair characterisation of what you said, but I think
> that Web-based things should be tailored to the medium, and that means
> that information should be arranged in a web-like
> structure, not artificially forced into a one-dimensional structure like
> traditional writing.

It is. Wikipedia articles have more links than any other encyclopedia.  
Compare Wikipedia to Encarta, an electronic encyclopedia, and you will  
notice that we will have much more cross-references, often to completely  
off-topical articles. I think that's a good thing and allows readers to  
explore freely (don't get me started about link underlining again,  
though). It separates us from Encarta, though, who have obviously done a  
great deal of usability research.

The question is, do we want relevant information on one subject to be  
grouped together, or do we want to *unnecessarily* use the "hypertext  
medium" just for kicks? The latter seems like a gimmicky thing to do, and  
reminds me more of a project like Everything2. By doing so, we do not  
accomodate but actually encourage short attention spans, with lots of  
disjointed small articles. Remember that for each link the reader has to  
follow, he needs to develop an idea what that link is about, and whether  
he wants to read it, whereas it is much easier to skim a well-structured  
text in front of you based on section titles.

(BTW, it may be nice to have a small table of contents auto-generated for  
articles with more than 4-5 sections.)

I do believe writing a well structured coherent long article is more  
difficult than writing many small ones. I still think we should try.

> It would be good to have a way to save related
> articles together, but what is related depends on the reader, again. A
> reader interested in Chap would want [[Chap]], [[Chap's views on Stuff]],
> and [[Chap's views on Things]] saved together, whereas a reader interested
> in Stuff would want [[Stuff]], [[Chap's views on Stuff]], and [[Guy's
> views on Stuff]] all saved together.

There will always be differences in what is perceived as belonging  
together. However, where coherence is reasonably clear, it should be  
reflected by having relevant information merged into a single document.  
For example, I left some fictional characters alone because these appeared  
in several different fictional realms. So it would not have been useful to  
have them just in one article, because that would complicate linkability,  
saving etc. -- but for those characters which just appear in one realm,  
having them in the article about that realm is preferable.

>> * We do not require the reader to click around unnecessarily, which can  
be
>> confusing to many people

> I'm not sure I follow this argument. Isn't that what people do on the Web
> all the time?

One of the key problems of empirical web design is link predictability. If  
the reader does not know what to expect, he will get confused, frustrated  
and stop navigating. Obviously, the smaller your information pieces are,  
the more frequently the reader will have to find the proper link to  
navigate to the next one, and the more likely it is that he will get  
frustrated in the process. With too long documents, you risk having too  
much uninteresting information in the same file, making the text harder to  
skim etc. So finding the right balance is important. With lots of highly  
interlinked short articles, you get a structure like Everything2, which  
only appeals to very strange persons.

> I don't think so. Merging small articles into big ones increases the
> number of redirects floating around.

Replacing all redirects to FOO with redirects to BAR is a job for a  
machine. Replacing all links to FOO with links to BAR is a job for a  
human, because you don't know in which verbal context these links appear.

>> * A short average article length does not reflect well on our article
>> count, which is one of the key instruments used for size comparisons

> It makes the count bigger, which is a *good* thing. :)

No, it's not a good thing, if the next time someone does a random page  
sample for a review, they get 10 one-pargraph articles about fictional  
characters. This makes us look unprofessional.

And the more such articles we have, the more difficult it becomes to  
enforce standards, to systematically copyedit articles about certain  
subjects etc. -- like Everything2, the whole database becomes increasingly  
messsy and unappealing to deal with. It's like the subpages mess we just  
barely avoided.

>  As for comparisons,
> if you check other popular encyclopaedias, you'll find that many of them
> have articles a lot shorter than 20 Kb.

Certainly not about "Gnipper the dog". The point is: If you actually check  
out these articles, you notice that they are about subjects that can quite  
plausibly grow into reasonably long articles -- an author, a politician, a  
company spokesperson; a mechanical device, a philosophical concept, a  
published work .. We currently have plenty of small articles that can  
never be anything but small articles because their subjects are so  
insignificant.

>> Hard to edit:
>> One of the features on my personal wishlist is the ability to edit an
>> individual article section.

> Doesn't that give the reader more unnecessary things to click on, and
> wouldn't that be confusing to many people? ;)

Not for the reader, but for the editor. Editors can be expected to click  
more, but even for them, it should be a user preference to avoid  
confusion.

> The whole idea of linking to individual sections is just so evil that I'm
> going to have to write another whole e-mail on this subject...

Actually, I tend to agree with you on that one.

> The solution *is* structure. We agree on that. I say use the natural
> web-like structure of the Web; you say force things into a traditional
> book-like structure.

No, I don't say that. I say use both: traditional structural elements and  
associative web structure, to allow getting relevant coherent information  
quickly. Wikipedia is not a hypertext experiment.

I don't know how much we disagree in practice. As long as you don't move  
Chip 'n Dale and all their fictional supporting characters back to their  
individual articles, I'm happy.

Regards,

Erik