Yes, this research tells us that we can model something, based on  
simplified determinations of the averaged decision of a collection of  
humans, that will allow us to change images of faces so that the  
better match the model built. That means precisely that all the  
beauty (if that's what it is) is in the eyes of the (collective of)  
beholders. It also means the outcome is not aesthetic but the outcome  
of a crude psychological sampling technique. Is this beauty? It's  
certainly not aesthetics.

Of course, tricks like this are fascinating and create possibilities  
we'd not previously had. But understanding? I don't think there's  
much there: it's sort of geeky and trivial, but with the usual  
inflation of the computer world, and the usual blindness of the  
mechanist to what designers actually do.

Reminds me a bit of a charming mathematician who told me that  
Hundertwasser was the only decent architect: he wasn't prepared to  
listen when I explained that what Hundertwasser does is about as far  
removed from architecture as anything I can imagine.

Goodnight, my beauties!