Now that is interesting. I wonder how much the prevalence of bad honeypots skews e.g. research statistics on remote attacks.