When I built a site that existed for a very long time, was very popular, and involved monetary transactions, I had to track nearly everything. IP addresses, address changes, email changes. Everything I could think of. This was then utilized when I suspected someone of fraudulent behavior. I could pull up an administrative screen that compared data in an archive copy (where I dumped the older information for just this purpose and to specifically keep it inaccessible to the outside world for user security purposes). With that, I could see whether several users were actually the SAME user. I even tracked things like user-agent string and detected screen resolution.
A lot of pieces of data can come together to provide more than circumstantial evidence that someone is shilling, trying to feedback-bomb another user, and so on. Enough correlated points of data can confirm suspicions like this. You'd be surprised how many people use an email address for one account, change that address, then create a second account with the email address they used to have on the first account and then use the second address to drive up the value of their stuff by shill-bidding against another user on their own item.