Complex framesets

During last week's call I volunteered to provide a critique of my own
proposal regarding frameset complexity, whereby I suggested that as a
priority 1 requirement, framesets that exceed a certain complexity
threshhold must be accompanied by a NOFRAMES element to provide an
alternative version of the content. This checkpoint would remain as
priority 2 in other cases. The main problems with this strategy are:

1. Would even a complex frameset meet the definition of a priority 1
requirement? Clearly, this is a cognitive rather than a perceptual issue:
can it be genuinely concluded that at a certain limit, framesets become so
awkward to navigate and the interrelations among the frames so difficult
to comprehend, especially in a serial medium, that the content becomes
impossible for certain groups of users to access? Is this still the case
if a LONGDESC attribute is provided to describe the frameset, or should
the checkpoint state that the author is required either to use NOFRAMES or
provide a description, via LONGDESC, of the function of the frames?

2. If the above issue were resolved in favour of my original proposal, in
either a modified or unmodified form, then there would still remain the
task of defining the complexity threshhold, taking into account: (1) the
number of frames in the frameset; (2) the proportion of these frames for
which the content changes in response to the user's activating hypertext
links and thus loading a new resource into the frame; (3) any additional
complexity introduced by contiguous framesets, where one frameset is
linked to another, with the latter possessing a different number of frames
etc.; (4) the visual layout of frames in the set; (5) any other relevant
criteria.

I still think that a relatively simple measure can be proposed, similar to
that described in my earlier message on this topic.

As a final question, if TITLE is used to label frames, as it should be,
and this information is exposed by user agents, to what extent does it
reduce the need for NOFRAMES by increasing the complexity threshhold at
which framesets become inaccessible, assuming that there is one (cf.
issue 1 above)?

Received on Wednesday, 10 February 1999 16:52:19 UTC