Agenda

Thursday, 29 November, 2100 UTC (4 PM US Eastern, 10 PM France, 8 AM
Eastern Australia) on the W3CMIT Longfellow bridge: +1-617-252-1038

The purpose of this meeting is to discuss success criteria,
specifically:

1. The working group has agreed that only checkpoints for which we can
   develop adequate success criteria will be included in the normative
   document (i.e., count for purposes of conformance to the
   specification).

2. A set of success criteria for a checkpoint will be regarded as
   adequate if and only if it is possible to determine objectively
   whether or not they have been satisfied. Objectivity is not
   equivalent to being "machine testable", but items which are
   amenable to automatic testing will be treated as objective.

Questions:

a. What conditions must a set of success criteria for a checkpoint
meet in order to be regarded as objective?

Proposal (discussed at previous teleconferences): the success criteria must
be sufficiently clear and testable that 8 out of 10 human observers
evaluating the same content would agree on whether or not it met the
success criteria, and hence on whether the checkpoint had been
implemented.

b. In the working group's deliverables (guidelines and techniques),
how should we handle those aspects of accessibility for which we can't
develop good success criteria? Some have suggested that they should be
either included in techniques documents only, or moved into a separate
section of the guidelines, or included in the guidelines document
itself but flagged as merely "advisory" or "informative".

c. We need to work through the guidelines document systematically to
decide which checkpoints have adequate success criteria already, which
do not, and where further work needs to be carried out. It may well be
possible to provide adequate success criteria for checkpoints that
presently lack them.

This last item may be expected to occupy much of our time during the
meeting, and realistically probably won't be completed this week. It
involves a meticulous examination of every checkpoint and its
associated success criteria.

Received on Wednesday, 28 November 2001 18:48:01 UTC