As I have heavily mod'ed my sites, I can't honestly remember if a clean install of SMF distinguishes between spiders and guests in the "Who's online" or not. All I know is that spiders are clearly identified as such and are not counted in page views on my sites. But that said, you do make a valid point.
- I would also encourage getting rid of guest counting in the stats -- *except* in the Who's online / Info center areas where you can easily sort between guests and online members, and determine that 90% of the 'guests' are actually bots, etc... This kind of thing is interesting, but recording numbers on the long run is pointless because you can't know whether you got a bot swarm or simply an actual surge in interest.
In other words, effectively a short-term "cookie" held on the server but as a database record? That actually might be far more accurate than the current system in which case I'm in favour.
- I think that even if we don't start a session for guests, we should still record whatever page they're on and store it somewhere for the next 15 minutes (by IP.) I really like being able to know what a guest is doing... Not only that, but some people are adamant on knowing it, for security purposes. I'd suggest maybe only showing this data to admins, and showing it in a way that's different from the Who's online section -- for instance, we could store the query string and $_SERVER and show that in a subsection of the Who's online page, or something... Because bots will often try to reach URLs that you never considered, it's nice to be able to spot the oddities in there.