Bot Count, The Estates Strike Back: 50%

No you misunderstood me. I meant "if it's even true" that campers were around before slimmed down bot clients. That is what the person I quoted was saying. I don't know if anyone developed a program like that before camping was invented but the person I quoted said it like it was a fact.

only one question... (well actually a couple but they are related)

1. Why do you do these bot sweeps, then post the results here? What is your reason behind it, what do you wish to gain from it?

2. What, if anything, do you think posting your results on the forums, will do about bots, traffic and the such? (we all know LL rarely reads the forums, especially things like this)

That's what I understood, and it is true that camping existed before traffic bots and camping traffic bots. People-camping was all over the place before bots arrived to do the job.

Phil clean out your inbox

*returns you back to your regularly scheduled debate*

Yes, you're right, camping holes were the biggest ongoing problem in SL before traffic bots. They're both the same kind of problem, in fact, with the same cause, and the same effects... bots just turn the dial up to 11, that's all.
Of course the fact was that the vast majority of them weren't interacting. The only difference between camping holes and bot farms are that bot farms are cheaper to run.

But they do change the numbers of what you actually posted. You didn't post that your the numbers are a mixture of bots and people who aren't manned. You posted that they are bots, and you are wrong. Your bot numbers are inflated. *You* may understand what you mean, but people reading your thread titles and stats like "52% bots overall", for instance, will understand something quite different, and they would be misled by what you write. I'm inclined to think that the misleading is intentional, but that's just what I'm inclined to think.

As mentioned it's just as valid to examine random spheres with 96m radius as it is to examine random cubes of 256x256x4906. It's a fair criticism that my scanners mostly just examine avatars close to the ground leading to a bias towards avatars that spend their time on the floor. This should be easily solved by making the prims move up and down over time, scanning at each different height. With a system like that I can examine cylinders of 96m radius and 4096m height which would address most concerns I think. Only trouble is it takes me so long to put the boxes out every time I update the script. You might notice I've only done sims beginning with A and some with B that I have parcels in and unless I can get a fancy automated way to set out the prims then the whole thing is probably going to get quite tedious if I keep coming up with improvements.

Overall I think my figure, currently at 7%, is on the low side and I feel as if from bad luck I just haven't hit some of the really heavily botted sims. Also, it feels like most massive bot farms are on estates for some of the reasons mentioned elsewhere. Still, from the data I'm sharing I have strong evidence that the figure of 50% or whatever it is for bots on mainland is grossly inaccurate. I'll probably continue to work on providing proper independently verifiable data.

One interesting group that my system skips is avatars that are camping for a normal human amount of time, say overnight. If someone camps for 4 or 5 hours then leaves and gets replaced by some other camper then the system I'm using is not going to detect them.

I think Anya's approach is the correct one. Few campers these days are "live" avatars, with someone at the keyboard. They may be scripted bots, or they may simply be AFK avatars...but in either case, the result is the same: A drain on system resources, and an economic drain, with no return either in user enjoyment or in new created content.

Camping used to be a social activity. You could sit and make a few lindens while chatting with your friends and sorting your inventory. Now, it's a business, one that does SL no favors.

Anya's thought about using live observation to improve the ways in which automated tools can detect bots/AFK campers is a good one. Instead of arguing about it, maybe she and Elanthius could team up to build a better botfinder?

Bloody hell!. So not only are you calling me a Bot, but now you are calling me a duck just because I've got a fat arse and waddle a bit.

You are one mean mofo

It is splitting hairs to some, but it is not, not when you are trying to measure something specific. Although the end result is the same from one point of view (system resources), the phenomenon is not from another (social interactivity). Make up your minds what it is you want to measure and then the way this survey is interpreted changes.

Any avatar could be seemingly inactive to someone scanning, when they are just standing in their home talking on IM for an extended period of time, or may be building while logged in to receive IMs. That doesn't make a person a bot, it just means that they haven't moved. And because real people have become more insular in public spaces and less outgoing doesn't mean that there are more bots.

I assume that your "Few campers these days ..." is a guess rather than something you actually know. It depends on what you want to know whether or not Anya's counting is good. She says that she counts all real camping people as bots, when even with your (assumed) guess, they are not, and that makes her number of bots wrong. Also, Anya doesn't claim that it's anything to do with system resources - she says it's about discovering how 'social' SL is. Camping is still a social activity; i.e. there are still real people doing it. Working for hours on end in a workshop is not a social activity, so why aren't those avs counted as well? She says it's because they are not trying to game the traffic numbers, and so we appear to have a glimpse at the real reason why Anya does these sweeps.

Elanthius, I think we're calculating two different things. I think you're calculating the percentage of accounts that are bots, while I'm calculating the percentage of people signed in that are bots. Let me illustrate.

Say there are seven avatars: A, B, C, D, E, F and G. Say you find the following avatars at a scan location at three different times:

A, B, C
A, D, E
A, F, G

You would say 1 in 7 are bots. One in seven accounts are bots.

I would say on each scan 1 in 3 are bots. One in 3 people signed in are bots.

How many times does it have to be said? If you're in your home you aren't counted. You aren't even a candidate.

Listen. I did this by hand. Do you think I took note of where 900 avatars are? Do I look like Santa Claus?

Think. If you were going to look for bots, how would you do it? Are you going to record positions of 900 avatars, or are you going to narrow that down?

Which of these situations is more likely to be bots?

Three avatars sitting in a crappy, empty club.
Two avatars in a house.

Okay, how about this one...

Fourteen unrezzed avatars in an empty box over a store.
Fourteen avatars dancing in a club.

Okay, how about this one...

An avatar in an empty box over a crappy store.
An avatar in a box with partially-completed projects, over a nice store.

So where are you going to spend your time looking for bots? Snooping on everyone cybering on their 1/16 pancake islands?

anya were you going to reply to my questions?

I seriously am curious, I wonder why you posted these and why you took the time and effort to do the bot scans

what do you hope to achieve?

Do you think it will make a difference to LL?

Were you trying to educate the already educated about bots? (we all know bots are there)

Please, I really am curious, tell us why you did it.

hey!

three people discussing what to do to make a better club, to have the two people in the house spending their money in the club


sl is borked thats why 14 people are half rezzed, the other 14 have had luck.


both are in im chat the first person talks to person 2 what to do to make a nice store

-> sorry i couldnt resist

Maybe Anya will respond, but I'll tell you why I'd do it (and maybe I will, if I ever have some time on my hands): sheer curiosity. We all talk about bots, and argue about how many there are, or aren't. Even LL admits their own figure is only an "estimate". Anya decided to find out for herself. I would also imagine that she was startled by what she found, and posted her results because she wanted to share that discovery with others.

We can argue about her methodology. But this is basic science -- posing a question, and going out and measuring the universe, or some small part of it, in search of an answer.
Anya herself has said it -- the proper response to her posts isn't to diss her, or question her reasons or her honesty. The proper response is to go out and do your own survey. Do you get similar results, or different ones? In either case, why?

Elanthius has done this. Kudos to him! Now we can compare the results, and the methods used. We'll learn something.

There are at least three quite different questions one might try to answer with data like these, and they lead to different interpretations of campers and bots.

1. How much are LL's promoted numbers dependent on Traffic gaming? This isn't about beating up LL, it's about how much self-interest motivates LL's extreme reticence to get rid of Traffic as a Search metric. If they're really reliant on Traffic gamers to give them half-way respectable concurrency and user-hour counts, then we know why they just won't listen to reason on this subject. (This question is the one most pressing in my mind.)

For this question, Traffic-inflating bots and campers are exactly equivalent.

2. How much resources are consumed because of Traffic gaming? The recent OpenSpace fiasco focused a lot of attention on demand for shared resources, so this is a burning issue for some right now, but really it's a serious scalability question always.

There's no doubt that bots and campers place some load on those resources, but usually very different loads. A reasonably competent Traffic bot should access central services very little after login, and should stay logged in for very long intervals; if they're well out of sight so they don't interact with anybody or anything else in the sim, they should consume very little network bandwidth, too. In contrast, a camper is very likely to be running a full-fledged viewer, may be chatting with other campers, and may even be sorting inventory (!), whacking the central resources harder than even the average avatar, even though they're not moving around. Camper-generated Traffic is way more resource-intensive than bot-generated Traffic.

Hence, for this question, one might want to distinguish bots from campers in order to discount the bots (although only LL could really determine the correct discount factor, empirically).

3. How much do non-responsive avatars degrade the social Grid? This is a much tougher question, in my mind. I mean, while I'm building or scripting or testing stuff in-world, I'm sure not being very social, so I may as well count as a bot, for those purposes. In contrast, some campers may find themselves in the company of other human campers and can interact socially. Presumably, no matter how "sociable" a greeter bot is, we'd not grant those interactions any "social" standing (yet).

So I guess the ideal answer to this question would count bots and discount human campers.

But I have to confess that I can't claim I've had any less social an interaction with a greeter bot than with the typical club hostess doing her job with a canned greeting gesture; or the Waterhead creep who's just *anti*-social; or the content-creator-alt slaving away in the workroom. So for me, measuring "the social grid" seems a wholly different enterprise than counting bots. But that's just me.

Why not? If real people who have gone to bed or are watching TV while their avs camp are counted a bots, why not real people who are alone in their homes without moving for hours one end - perhaps also having gone to bed or watching TV.

It all leads towards the conclusion that you are not trying to find out "where SL is going", or how social it is, as you've claimed, but instead it all seems to be about traffic manipulation.

Because there's no positive feedback encouraging tens of thousands of people to park themselves in their homes that way.

It's like the difference between subscribing to a mailing list and getting spammed.

It leads me to the conclusion that you are feeling guilty. I really don't sense the conspiracy or hidden agenda you keep talking about.

I don't think anyone else in this forum would agree that I feel any guilt about my bots

I pointed out that Anya first said, in the first of these three threads, that her reason for the sweep is to find out which way SL is going. Then she said that her reason is because of the apparently dwindling social aspect of SL (she said that's how it started - there is no conflict). But in some of her posts, she's pointed to traffic manipulation as a criteria for counting some avs and not counting others. So it seems to me that, regardless of how it started, her objective is to do with traffic manipulation, and nothing to do with how social SL is or isn't. Her thread titles and data lines also adds to that conclusion.

Well she has pointed out numerous times that her method was to eliminate the no-brainers first. (traffic bots are bots, people in their homes are not) So that's how I think traffic got into the conversation.

Methinks thou dost protest too much.

It's a very different enterprise and ignores the fact that some genuine campers are camping to be social later on when their camping money reaches a figure they're looking for. They use it to pay rent, buy clothes, buy homes. This is what they're camping for, to socialise. Just discarding the activity as being the same as a traffic bot does not address the social aspect as a whole.

Methinks that one denial hardly falls in the realms of protesting too much