Tim Bray continues his On Search series, and describes a hypothetical, Web-based interface.

If you take a look at the XML fragments, it seems to me that this is pure Web-based, but clearly not REST-based. I wonder whether that is intentional?