Parcel level bot detection

[Scylla Rhiadra]

5 minutes ago, Phil Deakins said:

If they are registered as scripted agents, they are allowed to be there, because they don't affect the traffic count  There isn't a rule about not logging in avatars simply for their green dots.

 

Ack, you're right. I keep forgetting this!

Thank you Phil.

[Scylla Rhiadra]

2 minutes ago, Love Zhaoying said:

Now I'm lost. 

If they are "Registered Scripted Agents", then what rules are they breaking, that you are reporting?

I may be misreading, but Prok is saying that he only reported the 4th bot, who did not get flagged as a registered agent.

[xDancingStarx]

1 hour ago, Scylla Rhiadra said:

Because I remember very clearly being told in one of the BB threads that it was possible to spoof a viewer in a way that made a bot undetectable.

An SL client technically can pass anything as viewer identification. But that doesn't matter. In the end you can just turn Firestorm itself into a scripted agent that teleports around. There's nothing that can stop it. And I dispute the claim that this is any of a big issue, only because some people fight an ideological war against "bots". What you want is that the remaining ones who register their bots as scripted agents (because by now a lot will not be registered as scripted agents anymore as they can be banned from private sims) continue to do so.

[Scylla Rhiadra]

4 minutes ago, xDancingStarx said:

An SL client technically can pass anything as viewer identification. But that doesn't matter. In the end you can just turn Firestorm itself into a scripted agent that teleports around. There's nothing that can stop it.

Thank you: if true, then that does mean there is no simple way to detect unregistered bots.

4 minutes ago, xDancingStarx said:

And I dispute the claim that this is any of a big issue, only because some people fight an ideological war against "bots". What you want is that the remaining ones who register their bots as scripted agents (because by now a lot will not be registered as scripted agents anymore as they can be banned from private sims) continue to do so.

Well, what we want really is to get all of those unregistered bots registered, if only so that they can't be used to game traffic.

And at the same time, I'll point out again that some people, quite understandably, don't want an endless stream of bots, registered or unregistered, popping on to their lands. It's not ideological: it's about privacy and general annoyance.

Edited June 29, 2023 by Scylla Rhiadra

[xDancingStarx]

4 minutes ago, Scylla Rhiadra said:

Well, what we want really is to get all of those unregistered bots registered, if only so that they can't be used to game traffic.

Well this new functionality has no impact at all on "traffic bots" since "traffic bots" are no bots but still controlled by a human. Unattended maybe. But then you'd need to change TOS and say "you can't leave your viewer unattended to gain traffic". And frankly this is impossible to implement.

As far as roaming bots are concerned, the more places that ban registered bots, the more bots will become unregistered. That's just facing reality.

And because it's a very recent event I wanna quote some lines that Blizzard posted very recently in regards of bots. Bots have a big impact on other games outside of the SL space. Maybe it's the case that a very few people in SL have the issue that bots chose their parcel as landing spot, but SL doesn't have a general issue with bots as those other games have. SL doesn't need to "fight" bots the way those other games have to fight them. Here's the quote in response to those advocating for "ban all bots now!"

"It’s an arms race, and it never, ever ends.[...]We will never completely beat “bots”[...]The ubiquitous nature of this type of thing in online games is an objective fact. It has always been a part of WoW, and every other popular online game for the past 25 years, and it will always be a part of online games going forward. It’s frustrating to fight this fight, but we will not stop fighting it.[...]"

Posted May 25th 2023 https://us.forums.blizzard.com/en/wow/t/thoughts-on-wow-token-in-wrath-classic/1598255

[Scylla Rhiadra]

8 minutes ago, xDancingStarx said:

Well this new functionality has no impact at all on "traffic bots" since "traffic bots" are no bots but still controlled by a human. Unattended maybe. But then you'd need to change TOS and say "you can't leave your viewer unattended to gain traffic". And frankly this is impossible to implement.

But it can, in fact, allow one to determine if those 30 avatars hanging out on a sky platform above a club are registered or not. If they are registered, they don't matter; if they are unregistered, they are likely gaming traffic, and can be ARed as such, whether they are automated bots or not.

10 minutes ago, xDancingStarx said:

As far as roaming bots are concerned, the more places that ban registered bots, the more bots will become unregistered. That's just facing reality.

Yep. Agreed.

And to reiterate something I've had to say over and over since the whole BB thing came up: I am not "anti-bot." I don't think most people are. Bots DO provide all sorts of useful services for the individuals employing them, and for those looking for information about the grid as a whole. I myself occasionally check out the BonnieBots site for information. The issue is, as Paul said above, about certain bot "behaviours"; about their misuse, in fact.

I suspect a better solution to the mainland issue of bots swarming particular parcels might have been better handled by restricting where bots can spawn or land. But of course even that would only impact on registered scripted agents.

And your quote from Blizzard about endless war more or less echoes something I've actually already said: there is no ultimate solution to any of this. Code is not a god-like force that can change human behaviours with the stroke of a pen (or refresh of a script).

[Love Zhaoying]

12 minutes ago, Scylla Rhiadra said:

And to reiterate something I've had to say over and over since the whole BB thing came up: I am not "anti-bot."

You should be! 

Mark my words, "Bots" will cause more trouble than good before this is all over.

(Seriously!)

 

[Paul Hexem]

1 hour ago, Coffee Pancake said:

The reason this isn't done .. is it's actually harder than using open source SL libs and a real programming language that aren't so limited or quirky.

It could be done .. but anyone doing it with RLV is nuts.

Right. A dedicated viewer is absolutely the best way. Just not the only way.

46 minutes ago, Scylla Rhiadra said:

And at the same time, I'll point out again that some people, quite understandably, don't want an endless stream of bots, registered or unregistered, popping on to their lands. It's not ideological: it's about privacy and general annoyance.

The thing is... This doesn't really stop that. They can still show up. At best this can only ban them after they arrive. Immediately after, if you want to get really laggy. And considering the length of ban lists and that it's a quick, free process to register a new account, that's not sustainable.

13 minutes ago, Scylla Rhiadra said:

But it can, in fact, allow one to determine if those 30 avatars hanging out on a sky platform above a club are registered or not. If they are registered, they don't matter; if they are unregistered, they are likely gaming traffic, and can be ARed as such, whether they are automated bots or not.

This, unfortunately, is the only real practical use for this LSL function. To make vigilantes. Which brings us back around to my very first statement- it causes new problems even as it solves others.

I maintain that this needs to be solved by LL enforcing policy, not giving residents placebos and vigilante tools.

[Scylla Rhiadra]

1 minute ago, Paul Hexem said:

The thing is... This doesn't really stop that. They can still show up. At best this can only ban them after they arrive. Immediately after, if you want to get really laggy. And considering the length of ban lists and that it's a quick, free process to register a new account, that's not sustainable.

Interesting point. Where the Estate-level bot banning tool can prevent them from arriving, is this function only able to boot them after they land? Surely there are ways to script it to prevent their arrival? (And by "they," I of course mean only registered agents.)

If you're running a bot operation, and you discover that your bots keep getting booted from a particular parcel, you're surely going to respond by finding a way to make them spawn elsewhere, rather than going through the process of creating endless numbers of new accounts, just so that you override that particular parcel owner's obvious objections to bots?

The scenario you evoke here, of a bot farmer responding to his bots being booted by overwhelming the landowner with a small army of bots is surely a little . . . silly?

6 minutes ago, Paul Hexem said:

I maintain that this needs to be solved by LL enforcing policy, not giving residents placebos and vigilante tools.

I don't disagree. We wouldn't be here now if LL had 1) worked to actually enforce its policies, esp. around traffic bots, and 2) foreseen the sorts of issues that were raised by BB and the proliferation of bots, and put in place rules and mechanisms for dealing with it in the first place.

I am skeptical that this is going to produce vigilantes, as you put it. But prevention is always better than a cure.

[Gabriele Graves]

22 minutes ago, Scylla Rhiadra said:

Interesting point. Where the Estate-level bot banning tool can prevent them from arriving, is this function only able to boot them after they land? Surely there are ways to script it to prevent their arrival? (And by "they," I of course mean only registered agents.)

Yes initially.  However it would be possible to put them on the ban list after detection which will block them from entering at all in future.  Of course the ban list is limited but with the script managing the entries, the most recently/commonly seen bots could be retained and ones that don't come back would drop out.

Edited June 29, 2023 by Gabriele Graves

[Scylla Rhiadra]

9 minutes ago, Gabriele Graves said:

Yes initially.  However it would be possible to put them on the ban list after detection which will block them from entering at all in future.  Of course the ban list is limited but with the script managing the entries, the most recently/commonly seen bots could be retained and ones that don't come back would drop out.

Yeah. I don't really buy the "my ban list is too small argument." Unless you are being maliciously targeted by someone, you're unlikely to find your ban list overwhelmed by individual bots, I think. There are lots of them, but not that many different ones always visiting the same place.

[Silent Mistwalker]

4 hours ago, Love Zhaoying said:

Words!! All of the above is words!!

Circular argument.

 

Not your post, his. 

Edited June 29, 2023 by Silent Mistwalker

[Silent Mistwalker]

4 hours ago, Rowan Amore said:

Yes, I know and the knowledge base article has it written quite differently.  It says scripted agents used to inflate traffic are not allowed.  Yet, scripted agents don't effect traffic.  There is nothing there about people creating 30+ avatars and using them to inflate traffic.

That's because the section you're referring to is in another location but it's still there, somewhere.

[Istelathis]

Bots do not bother me as much as they do others, but then I always found furbies to be adorable and am fascinated by AI and chatbots.  I do remember back in the 90s when people were freaking out over furbies, and thinking they were spying on them which at the time I found hilarious.  I might try to write out a simple script at some point just to count how many bots enter the region my avatar is on, out of curiosity,  but won't ban them outright because the ones that are registered are usually used for practical services and offer a tool to gather data not otherwise obtainable on such a wide scale by LL.  Things such as security orb location, in the case of BB popular attachments, avatar choices, and so on.  

It is nice that LL is providing a tool for people who do not want bots on their land though, even though it ultimately is useless, people can at the very least feel safer.  Unfortunately though, as others have noted, this is only going to spur on more bot wars, as people find more creative ways around this.  

The furbies will eventually take over, and have their revenge.  I for one welcome our new furry robotic overlords

but in the meantime, I will leave this video behind.

At some point though, we will not even be able to tell the difference between a bot and a human in our virtual spaces.  AI will make conversation with them indistinguishable from what we have with people, they will be able to see and interact with the same virtual objects we do, and there really is not much any of us can do about it.

If you want safety from the bots, for those who are afraid of them, then the best you can do is just be careful with the information you share with others online.  For those who are just annoyed with them, this is just going to be one more thing online we are going to have to put up with.

Edited June 30, 2023 by Istelathis

[Rowan Amore]

26 minutes ago, Silent Mistwalker said:

That's because the section you're referring to is in another location but it's still there, somewhere.

It used to be in the scripted agents section before the rewrite, I believe.  It also.said we shouldn't bother caring since LL monitors search rankings and investigates.  😂

The only thing I could find was in a knowledge base article about creating search listings...

Search policies

Any attempt to inflate your rank artificially in search results, including the inappropriate use of traffic bots or Picks, could result in penalties to your rank, de-listing from search, and disciplinary action against your Second Life account.

[Paul Hexem]

3 hours ago, Scylla Rhiadra said:

Surely there are ways to script it to prevent their arrival? (And by "they," I of course mean only registered agents.)

No, only estate/land tools can do that, not script.

3 hours ago, Scylla Rhiadra said:

Yeah. I don't really buy the "my ban list is too small argument." Unless you are being maliciously targeted by someone, you're unlikely to find your ban list overwhelmed by individual bots, I think. 

There's not too many other people besides those being maliciously targeted that need to ban bots, to be fair. 

Edited June 30, 2023 by Paul Hexem

[Love Zhaoying]

4 hours ago, Silent Mistwalker said:

Circular argument.

 

Not your post, his. 

That's how Phil wins arguments! Can't trap him in being "wrong" since he only quotes himself. 🙂 I need to learn from that!

[Love Zhaoying]

3 hours ago, Paul Hexem said:

There's not too many other people besides those being maliciously targeted that need to ban bots, to be fair. 

But..but..Second Life isn't fair!!1!!1!

[Love Zhaoying]

4 hours ago, Istelathis said:

Bots do not bother me as much as they do others, but then I always found furbies to be adorable and am fascinated by AI and chatbots.  I do remember back in the 90s when people were freaking out over furbies, and thinking they were spying on them which at the time I found hilarious.  I might try to write out a simple script at some point just to count how many bots enter the region my avatar is on, out of curiosity,  but won't ban them outright because the ones that are registered are usually used for practical services and offer a tool to gather data not otherwise obtainable on such a wide scale by LL.  Things such as security orb location, in the case of BB popular attachments, avatar choices, and so on.  

It is nice that LL is providing a tool for people who do not want bots on their land though, even though it ultimately is useless, people can at the very least feel safer.  Unfortunately though, as others have noted, this is only going to spur on more bot wars, as people find more creative ways around this.  

The furbies will eventually take over, and have their revenge.  I for one welcome our new furry robotic overlords

but in the meantime, I will leave this video behind.

At some point though, we will not even be able to tell the difference between a bot and a human in our virtual spaces.  AI will make conversation with them indistinguishable from what we have with people, they will be able to see and interact with the same virtual objects we do, and there really is not much any of us can do about it.

If you want safety from the bots, for those who are afraid of them, then the best you can do is just be careful with the information you share with others online.  For those who are just annoyed with them, this is just going to be one more thing online we are going to have to put up with.

Yeah, I follow the "Look Mum, No Computer!" guy on YouTube.  Mostly because of my own synthesizer stuff.  I usually don't find his videos interesting enough to actually watch.

[Scylla Rhiadra]

7 hours ago, Paul Hexem said:

There's not too many other people besides those being maliciously targeted that need to ban bots, to be fair.

I suspect not, but I'm not going to tell someone who has them popping in regularly on her property that she should just be tolerant.

[Phil Deakins]

7 hours ago, Rowan Amore said:

It also.said we shouldn't bother caring since LL monitors search rankings and investigates. 

That's exactly what LL said at the time they banned the gaming of traffic. But, more recently, I'm sure they've said to report traffic bots.

[Phil Deakins]

@Innula Zenovka @Qie Niangao and anyone else who may be interested. After a good night's sleep, I'm thinking more clearly, and this is what's happening.

llGetAgentInfo() tests 15 bits (flags), the 15th bit being the AGENT_AUTOMATED flag. Normally it works fine but, in the case of my Speedlight Gold account, bit 15 is set (1), even though, according to its LL account, it is not set as a scripted agent.

Here is the result of checking the automated agent status of 2 avatars:-
[02:01] Bots Control: ----------------------------------------------------------
[02:01] Bots Control: info:   100000000110110
[02:01] Bots Control: auto:  100000000000000
[02:01] Bots Control: AND : 100000000000000
[02:01] Bots Control: Wicked Leigh is a scripted agent
[02:01] Bots Control: ----------------------------------------------------------
[02:01] Bots Control: info:   000000000000110
[02:01] Bots Control: auto:  100000000000000
[02:01] Bots Control: AND : 0
[02:01] Bots Control: Phil Deakins is not a scripted agent
[02:01] Bots Control: ----------------------------------------------------------
info  = the agent's flags, returned by llGetAgentInfo(agent).
auto = binary of AGENT_AUTOMATED constant to AND with info.
AND is the result of the AND operation.

It looks like LL had already allocated the 15th bit for something else, probably concerning Speedlight Gold accounts, before they decided to use it as the AGENT_AUTOMATED flag.

It's a bug, so I'll report it.

ETA: I've reported it (https://jira.secondlife.com/browse/BUG-234064).

Edited June 30, 2023 by Phil Deakins

[Love Zhaoying]

That's interesting if LL repurposed the "previously not used" bit and forgot. Like a left-hand / right-hand thing.

[Phil Deakins]

8 minutes ago, Love Zhaoying said:

Like a left-hand / right-hand thing.

Yep

[Qie Niangao]

44 minutes ago, Phil Deakins said:

It looks like LL had already allocated the 15th bit for something else, probably concerning Speedlight Gold accounts, before they decided to use it as the AGENT_AUTOMATED flag.

Interesting. The whole Speedlight Gold arrangement seems weird to me, so I can't say I'm surprised there are lurking bugs under that particular rock. I kinda suspect that whatever flag the buggy code was supposed to be setting was in a whole different bit vector, but somehow got pointed to the one llGetAgentInfo is using. Just in passing, and probably unrelated, before AGENT_AUTOMATED was introduced, there was some speculation the info would instead be made available asynchronously with llRequestAgentInfo, or even the list-passing llGetObjectDetails.

On other points: the reason an RLV "bot" is significant is to highlight the fact that however easy it is to use dedicated bot software, it's ultimately only a matter of convenience and inertia that unregistered bots might be detectable based on which viewer they profess to be running. If whatever the bot is doing is valuable enough, it can be made indetectable for any level of inquiry: the more valuable the bot, the more can be invested in disguising "botness".

FWIW, I don't think RLV is even necessary. llTeleportAgentGlobalCoords just needs to be in a script owned by the target agent—an attachment, say—so the Linden viewer should be just fine for this kind of script-navigated bot.

The trick is, though, what can the bot do when it arrives,? That's where the dedicated bot client has a big advantage: it can gobble any data exposed to the viewer, not merely the stuff in the script API. This is the entire root of the bot problem: The Lab persists in hiding things from scripts on the (naive) notion that making that same information available to viewers protects that access in some way unavailable to a script interface. It doesn't protect jack, it only creates the entire market for the bot industry.