Referrer spamming is significantly else effort than this, and the validation of the target link in the source does provide a higher barrier. Parsing for microformats from the source also gives more structure to validate heuristically. Eventually a whitelist model will make more sense; the 'vouch' extension gives a way to construct one which puts more burden on the spammer than the receiver.