About the Bot Survey

How It Started

Did you know there are automated avatars that move around the Second Life grid, popping into places for a few moments and then maybe coming back again hours, weeks, or months later? Me either…until I made my own roaming bot for my Avatar Survey. Turns out there are many more roaming bots operated by other people—and I could make them land on my head! Countless hours of nerdy fun. So I started keeping track of them, and loosely grouped them into a series of "cohorts" based on their behaviours and other traits.

How It's Going

Since I got started in early 2019 I've identified about 300 bots that roam (or have roamed) around the SL grid. Some are on the move continuously—one group looks to have been operating over a decade—while others appear and disappear over a short span of time.

Methodology

How do I gather data? You know those "visitor count" scripts you see everywhere: "You're visitor 56 to my awesome store!" I'm doing the same thing except I'm only counting avatars I'm pretty sure are bots. Clients, friends, and landowners have been good enough to let me scatter bot detectors across an ever-changing range of mainland and private regions. The number of detectors fluctuates up and down, so if a particular month reports a higher-than-normal number of sightings, that probably means I'm just seeing the same bots more often because I have more detectors out or I've managed to park one where a roaming bot frequently appears. The monthly active bot totals are a better indicator of roaming bot activity.

I determine which avatars are roaming bots based on the nature and distribution of sightings, and also if they get pulled to little "bot traps." Bot traps are private parcels (not a store or anything with public traffic) with one of my detectors that has its landing point set to a weird location. Members of the group that own the parcel can TP anywhere they like, but bots are automatically pulled to the landing point. Occasionally a real person stumbles in, but its really unlikely someone is going to hit three or four of those "bot traps." As the total goes up, so does the probability I'll mark that avatar down as a bot. If that avatar also turns up all sorts of other places…well, then I'm pretty sure. If not—some mainland flyers and drivers occasionally trip my system—the avatars drop off the might-be-a-bot list automatically.

And, of course, on my own parcels that landing point is right above the place where I sit to work on scripts. That's how bots land on my head. I just do that because I think it's funny.