Beekeeper,
Thanks for you insight, and the dialog. I am not asking you to apologize for whomever made the error, but the fact that you are willing to publicly acknowledge that I was not treated properly does mean a great deal.
Before I leave this poor horse alone, I will offer one final thought. I would gladly buy a beer for anyone who can prove that there was an exact IP match between me and a known "troll account". I post from two locations, my home, and my work. In both cases, I utilize NAT for converting an unroutable IP address into a routable public IP. For anyone to use the public IP of my work, they not only have to work for the same corporation, but utilize the same proxy (we have at least 12), and public firewall interface which is chosen dynamically based on internal network conditions. Odd's are not likely.
That pretty much leaves my home as the only viable option. Since I use a broadband provider, my public interface is a DHCP client, which means I lease a public IP address. Typically my public interface updates every 24 hours, and always renews the same IP from the DHCP server provided I haven't shut off my firewall. Effective lease of a single IP address usually lasts for about two months before I renew something completely new, or my ISP changes the DHCP scope/network ID. Then we also need to look at the chances of me obtaining a new DHCP lease that was previously used by another member. Well, for my DHCP scope, there is a possibility of 2048 IP addresses (subnet mask of 255.255.248.0 results in a binary subnet mask of 11111111.11111111.11111000.00000000 assignable host range is 2 to the 11th power, =2048). Remove one for the gateway, another for the network ID, and a third is lost for the broadcast address. That means that there is a possible 2045 possible addresses. The last subnet ping sweep I ran indicated that there was slightly more than 140 leased addresses out of the 2045 available. Since the DHCP has so many unused IP's within the scope, it is extremely rare for it to issue a client anything but an unused IP. Chances are not good that out of 140ish customers of my ISP, at least one had all of the following apply: a member of this site, a troll account, and leased the EXACT same IP address I had held previously (or would lease soon thereafter).
Since I wasn't the only member here who had the same lockout problem, I'd feel much more confident in surmizing that whomever was doing the log analysis only matched a couple of the octects (out of four), and not the entire IP address, before pulling the trigger. It is very easy to match the first couple of octects (which would denote the network ID's of the provider), than the entire thing, which would require matching both the network ID and the host ID on that particular network.
But, then again, I'm just a lowly security architect who has about a dozen years in the system engineering/network architecture field under my belt. I could be wrong