The general problem with this is you don't know if you're sampling fairly or not. Perhaps the easiest way to do this (although there may be a sampling bias) wo

The general problem with this is you don't know if you're sampling fairly or not. Perhaps the easiest way to do this (although there may be a sampling bias) would be to search twitter for 'gmail', 'gmail sucks' 'like gmail', etc. to get an idea of what people who tweet about it think.