Accessibility support can be improved but there
several fundamental problems that will make it difficult for screen reader
users to understand and deal with a figure element with the figcaption
instead of an alt text on the image.

I don't think we can assume that screen readers will begin to announce that
there's a graphic inside the figcaption. Assumptions about longdesc got us
into a lot of problems 15 years ago. I think reporting the figure with a
role of group could be a problem that will be hard to resolve for screen
reader users. Although semantically it is a group, it will take a while for
SR users to understand that it *may* be an image inside a group but maybe
not, the figure element won't tell them that.

I think this belongs in the larger discussion about alt text, and whether we
should allow options to it. Because currently figcaption is probably more
confusing than aria-labelledby, aria label etc.

I would like to see good unity with html5 and WCAG. My thinking is that we
should go one way or the other... allow the gambit of substitutes for alt,
or none... I don't think this one exception of figcaption makes sense.

Perhaps this is a Face to face CSUN topic?

Testing here
http://davidmacd.com/test/figure.html