Peering Inside: How many bots?

Nobody, not even Linden Lab, is able to determine how many bots there might be operating on the Second Life grid. Not without intensive and potentially disruptive per-account data analysis, and even then the results are likely to be flawed. A bot can indicate to Linden Lab at login that it is a bot, but it doesn't have to. It might choose to appear to be a more normal user.

We've seen bots with first-life profiles, and Second Life profiles, and picks (people and places), groups, group titles, partner, payment information, prim wigs, custom clothes and skins, and in every way indistinguishable from the avatar of an actual user. While we may not be able to say with any certainty which accounts are definitely bots, which are campers, and which are merely idle, we think we can give you a pretty good estimate of how many bots and campers there actually are.

The short answer is around 10,000 of them.

You see, when the Second Life grid subsystems wholly or partially fail, as they have done so routinely lately, two things happen. One is that users have troubles logging on, or are unable to do so at all. With potentially hundreds of logins and logouts per minute, that causes sudden levelings or sharp declines in overall concurrency.

If grid subsystems other than login systems are giving trouble, users begin to log out in large numbers, being unable to travel, send or receive messages, access inventory or perform other ordinary tasks.

When both of these conditions hold, the overall user concurrency plummets towards the minimum. You would expect the minimum to consist of a few hundred die hard users, Linden Lab staff, campers (who are unaware of the troubles) and bots (who are incapable of awareness of the troubles).

Unless Linden Lab starts administrative logouts of users in bulk, that minimum is relatively stable between 10,000 and 11,000 (though it is rising slightly). This happens consistently, and has been visible in the figures on a number of occasions over the last two months.

Assuming roughly 500 of those are still active users who are still willing to hold on without being able to send or receive IMs, teleport or cross simulator boundaries, or access objects or inventory, that leaves us with as many as 10,000 or so bots and campers (some campers potentially are bots themselves). That's the closest we can come from statistical observations.

While a bot can (and generally should) consume less resources and contribute less to lag and system loads than an ordinary account, some few of them are likely not written with any such considerations in mind.

A bot may be a useful tool (indeed they can be faster, more efficient, cheaper and less impactful on grid performance than any manual operator) but ... ten thousand of them?

What are they all doing?

Actually, in the main -- nothing at all. There are islands populated almost solely by unmoving bots. One corporate installation does this. We don't know if the corporation (an energy company promoting green technologies) is responsible for these energy and resource wasting bots, or if they were introduced by the marketing company or developer to make the place look more successful and popular than is actually the case. Approximately sixty of them stand around the corporate island in poses of conversation or of rapt attention to the promotional videos.

And remember, what can possibly be more effective to boost your traffic than a set of camping stations? It's a set you don't have to pay for. This seems to be the most common use we can see.

Set up a camping station with good pay rates, park your own bots on the camping chairs, and rig the bots to return the money to you as they get it. Then advertise your great camping rates, and watch people turn up and stand around waiting for their turn at a chair that may never become free for them to use.

That's maybe 50% more traffic than you'd otherwise get with a camping installation, and you're not even paying for it anymore.

From the point of view of someone who is keen to artificially boost their traffic numbers or otherwise want to make a site in Second Life look more popular than it is, this is a major win.

For the rest of you, ordinary users of Second Life, it appears to be nothing more than a waste of resources. Or maybe it constitutes just a wide-scale con -- the difference between 'good business practice' and outright deception can be tricky to pin down.