Evidence for spatial variation of the fine structure constant

I think this is probably non-sense... but it's perhaps interesting if true.
http://arxiv.org/abs/1008.3907