Last summer, we looked at spamming behaviors in social tagging networks such as del.icio.us. Recently, some interesting ancedote of spamming in music social tagging data have surfaced: http://blogs.sun.com/plamere/entry/the_1_brutal_death_metal
The blog post above discuss how vandalism is affecting a particular tag at last.fm. It seems that basic human desires to work in anti-social ways occurs in many social Web2.0 systems. Of course, data mining experts and others have worked tirelessly to come up with algorithms that filter out these 'noise', but I can't help but wonder if these 'noises' are just as valuable as 'real data' in understanding human behavior. Moreover, these outliers seems to point to real data that we could extract and potentially use.