I can't check the guy's methodology, but Robert Shiller claims in the NY Times to have built a better, more accurate measure of housing prices. You might ask, don't we already have that - I always see things like "median home sales price" in the paper? The problem with existing metrics is that they don't correct for mix. If a lot of large houses in the pricey part of town sell, median home prices will rise just given the mix shift of the sample. What you really want is a price index for equivalent home sales, something that corrects for things like square feet, inflation, and perhaps zip code. This is what Shiller claims to have done, and the results are dramatic. He shows that real housing prices have been flat for most of the century, right up until the last decade, where they have increased dramatically.
I can't think of any structural change that would explain this (except maybe a change in relationship between mortgage rates and inflation) so it certainly creates a flashing red light saying "bubble".
By the way, isn't it interesting that people can see the graph above and immediately think "prices are due to crash" but when they see this very similar chart:
...and think that prices will keep going up and up and up.