W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > July to September 2006

Re: The benefits of footnotes in webpages

From: David Woolley <david@djwhome.demon.co.uk>
Date: Tue, 29 Aug 2006 18:36:46 +0100 (BST)
Message-Id: <200608291736.k7THall32377@djwhome.demon.co.uk>
To: w3c-wai-ig@w3.org

> I have studied the way Wikipedia has implemented footnotes, and it is mostly
> extremely wrong.

There are various styles.
> Take a look at two Wikipedia articles:
> Footnote, http://en.wikipedia.org/wiki/Footnotes
> OpenDocument, http://en.wikipedia.org/wiki/OpenDocument

You should look at the meta articles, a point I tried to make clear.  Your
examples are main text articles and will not show the discussion and
will probably use specific styles.  For various reasons, I cannot check
these references quickly, at the moment.

Another key point to realise is that Wikipedia uses HTML/CSS as a 
presentational output format and uses its own language as the user
facing markup language (an example of how standards get re-invented
when they get too bloated).

Being a Wiki, you can, of course, make your own comments, or even try 
to change the macros yourself.  Note that changing the footnoting or
citation style in an individual article can be unwise, and is likely 
to result in an edit war, as people can be very possessive about their
particular styles, even though the articles are really supposed to
be owned by the community.  Doing that for macros, without first getting
consensus on their dicsussion pages, is likely to have a similar effect.)

> full id for footnote 1 looks like this: "_note-0". This is much better:
> "footnote-1". And the full id for the footnote 1 reference looks like this:
> "_ref-0". This is much better: "footnote-1-referrer".

I imagine this is a consequence of their only being intended for machine
consumption.  One of the problems they have with numbered footnotes is
that the numbers aren't stable.

Note that there are other styles in which, for example, Harvard style
references are used in the main text.  My preference is for such 
references, but many contributors aren't really used to that convention.

> 5) The footnote section has a misleading h2 header: "References". It is
> better to call it: "Footnotes".

The most common use of footnotes in Wikipedia is to provide references to
primary sources.  There is a convention that such source references go
in a section called References.  This does cause problems, because many
authors don't understand the verifiability requirements for Wikipedia and
the need for such primary source references, so sometimes misuse them.
However, because this is the primary use of footnotes, References is
the going to have to be default section.

Another part of the history, is that many people have provided primary
sources using a short form link which just shows a machine generated 
link number.  That results in an incomplete citation that cannot be
repaired when the original link rots.  Having links to the References
section, helps to promote the inclusion of proper citations in the 

(People have difficulty understanding and distinguishing between, 
"See Also", "References" and "External Links"; the first is intended
for within Wikipedia references which are related, but don't appear in the
text, the second for primary source citations, and the last, which are 
discouraged, for non-Wikipedia references that are not primary sources.)

> 6) The footnote section is made as an ordered list. This is bad since we can
> not use the index numbers to link back to the referring footnote. Is is
> better to use a definition list as I do.

Normally, references are given as an unordered list, amongst other things,
because the entry numbers are not stable.   Expecting editors to 
maintain empty slots for broken references and ones removed as link spam
is unreasonable, so the HTML generating engine would have to renumber
the links in the text every time - I think the process is one pass, so
that's not possible.

The problem with definition lists is that the early GUI browsers abandoned
the option of showing the dt in line for short terms, and the CSS constructs
that would allow this to be re-instated are not well supported.  Taken 
in conjunction with the fact that they have taken the view that HTML is
too complicated and use it only as a presentational output format (although
generally with much higher quality than, say, email programs), such a
borderline use of dl would be unattractive because it would, unnecessarily,
have tags on separate lines.

> 7) In the footnote section all the links have the same stupid link text:
> "^".

It's a while since I looked, but I think the aim is to keep the online
display as uncluttered as possible.  There is also an issue that, for
proper citations, there may be multiple valid references in the same
document, and I'm not sure if they ever really solved the problem of 
back referencing these.

> 8) In the second Wikipedia article I mention above two footnote systems are
> used in the same page. Superscript for footnotes at the end of the page and
> "[3]" (the text mode way) together with an arrow symbol for external links.
> The last type of footnotes takes you to another page and at the same time we
> also have an "external links" section at the end of the document. Using
> "external" footnotes just being ordinary external links is in my opinion so
> far out that it is all bad. Too use "[3]" notation for such external
> "footnotes" I find extremely confusing.

[3] is a short form link that is older than any of the footnoting
mechanisms.  It is used for links that the author doesn't want to clutter
the main text, but does want to be present.  As noted above a lot of
authors don't like citing sources, and they have typically been used for
that purpose.  Although it is better to have full citations linked from
the text, having sources linked directly, or citations with no indication
of which point they support, are better than not having sources at all,
which is the risk involved in removing mechanisms.

At least the initial implementation of footnotes actually generated such
links to the same page rather than there being a specific, low level,
footnote mechanism.

The Wiki mechanism does allow one to have full text links, although ones
that are not recognized as within Wikipedia will get tagged with a symbol
to show they are external, but the policy is to discourage external links
except as primary source citations.  (They often get used as vanity links
or advertising links to sites that are related to the subject, but are
not valid primary sources, and that is not the intended purpose of an

> 9) The footnote references also differ from instance to instance in where
> the footnote reference is placed. Sometimes right after a word, a sentence
> or paragraph, sometimes we have a space in between. It is not done in a
> logical or consistent way.

Different editors.  But, again, you need to look at the meta articles to
see the policy.

> 10) In the "footnote" article of Wikipedia, I also find it incredible that
> year and date even for a resource in a footnote, link to articles about the
> year and date. Not what most users would expect. 

This is a global function of Wikipedia.  The advantage of marking dates as
special is that they are then treated as microformats and are rendered in
the reader's date locale, rather than in the locale in which they were entered.
The disadvantage, as you note, is that they are treated as special links.
Wikipedia users soon come to understand the nature of such links, even if they
are not terribly useful most of time and the Wiki rendering  code could have
supressed their linkness (it already suppresses self references).

The important thing about Wikipedia in this context is that it does make use
of footnotes and the level of semantic markup, albeit in the Wiki language,
is exceptionally high by the standards of typical web pages.  People have
thought about how to implement citation and other footnotes, even if there
is no one consensus, accessibility probably hasn't been made an explicit
consideration, and many authors aren't fully informed.
Received on Tuesday, 29 August 2006 17:37:01 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 July 2011 18:14:25 GMT