Is this more or less meaningful TO YOU than only an image element and req'd attributes? <figure> <img src="images/ennishouse.jpg alt="photo of Ennis House by architect Frank Lloyd Wright"> <figcaption>A photo of Ennis House by architect Frank Lloyd Wright</figcaption> </figure>

The problem with the structure is that it is too specific. The same underlying architectural form could be used for labeled equations, tables (which have a somewhat similar structure but not able to be adorned by common CSS), Scribd-like document inclusions, and other exhibits.

When a content architecture is not available, you get more elements by adding more elements, and the language accrues baggage through time (major XML languishes already suffer from such creep). Custom elements hold promise, but brittle implementation.