Over at Urban Demographics, they link to The Atlantic's photo essay on Landscape Absurdism: Las Vegas. Quite a few planners (and many others) love this stuff. It plays to the idea that there are "chaotic" and "wasteful" land use patterns, especially in sunbelt suburbs and perhaps most especially in Las Vegas.
We know now that there was suburban over-building, not just in Vegas, but in most places and this had very little to do with lax lands use planning practice. In fact, there is evidence that the most controlled places experienced the biggest house price bubbles.
How do we know "waste" when we see it? From the air? There is no way that visual inspection can tell us about "better" or "worse" land use arrangements. Richard Peiser showed some years ago that it is possible to build inter-temporal models which indicate that "leapfrog" development can be economically efficient, leaving important space for later infill development.
Besides "waste" is always a complex idea -- and almost never what it is suggested to indicate. Is washing my shirt the 25th time and wearing it again more wasteful than replacing it?
When it comes to land use patterns, the acid test is is simply whether the urbanized area grows or not. Does it attract labor and capital in ways that they can make a go of it? Las Vegas did very well in the years 1960-2000 (years for which we have data), its urbanized area exceeded large-urbanized area growth by a factor of almost twenty.
Labor and capital go where they expect to be productive. This includes opportunities to interact with a variety of other nearby activities -- at a price that makes it all possible. This involves the spatial arrangement of activities and is much too big a problem to be solved by bird's-eye visual inspections of land use patterns, followed by prescriptions for more compact development. Rather, it takes flexible land use markets. That's the best we have.
Ed Glaeser seems to be saying "let a thousand flowers bloom." There is no one density or arrangement that works best.