Parcel level bot detection

[Phil Deakins]

1 hour ago, Innula Zenovka said:

Have you written this up as a BUG jira?  

No. I'm looking at it and I've discovered what the difference is. Most people wouldn't follow this but I'll post it for you (and Qie et al).

llGetAgentInfo returns an integer, which is a string of binary flag bits for the agent.

The AGENT_AUTOMATED flag (16384 decimal, 100000000000000 binary) is ANDed against them to see if it's set (1).

That's fine so far. I returns 0 for non-scripted agents, and 1 for scripted agents. But the GOLD agent returns 16384.

Its info (llGetAgentInfo(agent)) returns 16438, which means it has a extra bit on the left, set to 1, that is not ANDed and so remains as a 1. It means that the ANDing operation in Signal Linden's script includes that left bit 1, and so the return is non-0, which is why that agent is seen as a scripted agent.

I'm thinking that the extra bit/flag on the left is a flag to indicate Speedlight Gold, or something to do with it. Whatever it's there for, it does need to taken into account when ascertaining whether or not an agent is registered as a scripted agent. The script in Signal Linden's post doesn't take it into account, and returns the wrong value for some avatars.

PS. How do you create a bug report?

 

Edited June 29, 2023 by Phil Deakins

[Love Zhaoying]

13 minutes ago, Phil Deakins said:

PS. How do you create a bug report?

I assumed it was a Speedlight bug, that's why I tagged Glaznah earlier. (They are the Speedlight expert / rep who starts Speedlight threads in the Forums sometimes.)

Otherwise, you don't see a way to file bugs on the Speedlight page?

(I'm too lazy to login and check. Websites, ewwww. Never mind that I'm a Gold member and never use it.)

[Phil Deakins]

3 minutes ago, Love Zhaoying said:

Otherwise, you don't see a way to file bugs on the Speedlight page?

It's not a Speedlight bug. It's an LSL bug.

[Love Zhaoying]

4 minutes ago, Phil Deakins said:

8 minutes ago, Love Zhaoying said:

Otherwise, you don't see a way to file bugs on the Speedlight page?

It's not a Speedlight bug. It's an LSL bug.

..but if it is only happening for Speedlight Gold logins..that certainly does not sound like an LSL problem to me.  It sounds like Speedlight is doing something DIFFERENT for Gold members.  But I see your point.

For an LSL bug: You file a JIRA. Like for any other type of bug.

[Phil Deakins]

@Love Zhaoying

You are mistaken. It's LSL that's doing it wrong.

It is me in the standard browser who is writing and running the LSL script. It's nothing to do with Speedlight. The only connection with Speedlight is that one of the agents that it's reporting on is logged in with Speedlight Gold.

Edited June 29, 2023 by Phil Deakins

[Love Zhaoying]

3 minutes ago, Phil Deakins said:

You are mistaken. It's LSL that's doing it wrong.

It is me in the standard browser who is writing and running the LSL script. It's nothing to do with Speedlight. The only connection with Speedlight is that one of the agents that it's reporting on is logged in with Speedlight Gold.

Fine, then. I'll just claim poor comprehension. (That is much easier than arguing with you about details that you yourself provided!) Plainly put: The LSL script does not know "who is logged on where".  But I can see, that if the new LSL Function is SUPPOSED to be querying the "Registered Agent status" - then it sure looks like an LSL Function issue.

Good luck filing your JIRA!  I'm sure it will help other people too.

Edited June 29, 2023 by Love Zhaoying

[Silent Mistwalker]

29 minutes ago, Love Zhaoying said:

Fine, then. I'll just claim poor comprehension. (That is much easier than arguing with you about details that you yourself provided!) Plainly put: The LSL script does not know "who is logged on where".  But I can see, that if the new LSL Function is SUPPOSED to be querying the "Registered Agent status" - then it sure looks like an LSL Function issue.

Good luck filing your JIRA!  I'm sure it will help other people too.

Remember when you could see what viewer people were using above the name tag (not sure if you were here back then)? That capability is still in the viewer code and LL could modify it. The point being LL still knows what viewer(s) people use even if residents can no longer see it.

[Love Zhaoying]

5 minutes ago, Silent Mistwalker said:

Remember when you could see what viewer people were using above the name tag (not sure if you were here back then)? That capability is still in the viewer code and LL could modify it. The point being LL still knows what viewer(s) people use even if residents can no longer see it.

I've only been around 16 years, but don't remember that.

Your comment DOES reinforce an idea I suggested in another thread (?) for detecting "bots", however! Thanks!

Edited June 29, 2023 by Love Zhaoying

[Innula Zenovka]

1 hour ago, Phil Deakins said:

PS. How do you create a bug report?

Go to https://jira.secondlife.com/secure/Dashboard.jspa?null and sign in.

Click on the Create button at the top right of the list of options at the top, and fill in the Create Issue form that appears.   Click the form's "Create" button (bottom right of the form).

[Innula Zenovka]

5 minutes ago, Love Zhaoying said:

I've only been around 16 years, but don't remember that.

Your comment DOES reinforce an idea I suggested in another thread (?) for detecting "bots", however! Thanks!

It was a feature introduced by Emerald (remember them?) which the Phoenix Viewer (V1 predecessor of Firestorm) inherited.  

Linden Lab eventually banned it because of all the drama it caused.

Edited June 29, 2023 by Innula Zenovka

[Silent Mistwalker]

34 minutes ago, Love Zhaoying said:

I've only been around 16 years, but don't remember that.

Your comment DOES reinforce an idea I suggested in another thread (?) for detecting "bots", however! Thanks!

It was 2012.

What the heck both are relevant.

 

Edited June 29, 2023 by Silent Mistwalker
wrong url

[Coffee Pancake]

10 hours ago, Scylla Rhiadra said:

If LL had a means of identifying unregistered bots, they could simply sweep the grid clean of them.

They can. All clients must identify themselves and be honest about it, there are also easily identifiable differences in how they talk to SL that gives the game away.

I think they have no desire to start an arms race.

[Love Zhaoying]

5 minutes ago, Silent Mistwalker said:

It was 2012.

What the heck both are relevant.

 

Thanks for the info!

The point of my suggestion earlier was: If LL can tell "what viewer" people are using, and it looks like an "unknown viewer" (like an API or something), perhaps it's really a scripted agent.

That could be what is happening with Phil's problem! ..If LL does not recognize "Speedlight Gold", etc. and they are using this feature to help filter out any "apparent" Scripted Agents - not just using the account setting.

 

[Jennifer Boyle]

4 hours ago, Innula Zenovka said:

Meanwhile if a bot is collecting data, it can do that from anywhere on the region, or even from your parcel before your scanner has a chance to detect and remove it, unless the scanner is on ridiculously fast timer, in which case its impact on region resources is likely to become an issue. 

Exactly, so there is no value in detecting registered bots, and there is no practical way to detect unregistered ones.

[Silent Mistwalker]

3 minutes ago, Love Zhaoying said:

Thanks for the info!

The point of my suggestion earlier was: If LL can tell "what viewer" people are using, and it looks like an "unknown viewer" (like an API or something), perhaps it's really a scripted agent.

That could be what is happening with Phil's problem! ..If LL does not recognize "Speedlight Gold", etc. and they are using this feature to help filter out any "apparent" Scripted Agents - not just using the account setting.

 

It is a very real possibility. Scripted Agents are supposed to be using specific software that LL should be able to detect.

 

Quote

Any script or program that automates access to Second Life and its supporting web services through interfaces designed for the Second Life Viewer and its embedded browser is considered a Scripted Agent. Second Life allows Scripted Agents with legitimate uses, but inappropriate uses are violations of this policy.

https://wiki.secondlife.com/wiki/Linden_Lab_Official:Scripted_Agent_Policy#:~:text=Any script or program that,are violations of this policy.

[Love Zhaoying]

1 minute ago, Jennifer Boyle said:

Exactly, so there is no value in detecting registered bots, and there is no practical way to detect unregistered ones.

Well, to be honest..knowing if it is a "registered bot" would be nice for people who just want to know, "what the heck is up with all these avatars who visit my property?"  If these poor souls had a security system that told them "And by the way, these are all registered bots.." that would provide a clue and hopefully some relief from potential paranoia.

There ya go, a use-case!

[Innula Zenovka]

1 hour ago, Jennifer Boyle said:

Exactly, so there is no value in detecting registered bots, and there is no practical way to detect unregistered ones.

However, despite being told that by several people (including me) in earlier threads in these forums, people were still complaining that LL weren't giving them the same ability to prohibit registered scripted agents that estate owners have.

Regardless of what practical benefits this brings parcel owners (or estate owners), it gives people the sense of control for which they were asking, which was the point of my quoting Khrushchev's supposed remark to Nixon about the solution to complaints about imaginary lakes. 

(Come to think of it, I've seen something attributed to Aliester Crowley, which I rather prefer, about how the magician persuades someone afflicted by imaginary snakes that he has an imaginary mongoose to deal with them).

Edited June 29, 2023 by Innula Zenovka

[Phil Deakins]

1 hour ago, Love Zhaoying said:

That could be what is happening with Phil's problem! ..If LL does not recognize "Speedlight Gold", etc. and they are using this feature to help filter out any "apparent" Scripted Agents - not just using the account setting.

I don't have a problem. The new LSL function doesn't always work if the code is as it is in the example, and I've explained why it doesn't. I.e. there is an additional flag that isn't taken account of in the example, which can be a 1.

Don't forget that LL recommends Speedlight Gold (not Speedlight free), because it's a perk of Premium+ accounts. Since the additional flag only affects the result from my Speedlight Gold account, and no other account, I'm assuming that the flag concerns Speedlight Gold, and I have to assume that whoever wrote the example code either didn't know about the flag, or forgot about it.

[Prokofy Neva]

The use case I have for this is managing areas next to infohubs where bots frequently land when their home sim crashes or when there is a rolling re-start. So you get a pile up of a dozen or more avatars at the Moth Temple, for example. Let's say you're new and you land there -- or maybe returning to SL. How do you know which of these avatars are a bot? I occasionally come across people trying to talk to them, especially if they seem to be advertising some club or venue.

My problem with this script is that when you deploy it in world near a likely gaggle of bots, it just keeps chatting endlessly which avatars are bots, so that it becomes spam in the chat.

I asked Signal Linden to put an on/off touch switch on it so that you can turn it on and off. You land in a sim, you turn it on to see who is a bot. If you control that territory, you can then eject them, especially as they take up scarce avatar slots.

I'd make a hand-held version as well.

I'm not 100% certain the script is picking up ALL bots. Evidently it can only detect REGISTERED scripted agents whose owners are conscientious. But unregistered? I see some avatars with names that sound like they're bots, yet the script isn't saying anything about them.

 

Edited June 29, 2023 by Prokofy Neva

[Phil Deakins]

1 hour ago, Innula Zenovka said:

Go to https://jira.secondlife.com/secure/Dashboard.jspa?null and sign in.

Click on the Create button at the top right of the list of options at the top, and fill in the Create Issue form that appears.   Click the form's "Create" button (bottom right of the form).

Thank you, Innula. I had a look at the form and decided that I don't want spend the time filling in all the boxes, so I'm not bothering about it. It's not exactly a big deal, and it's not really a bug because it can be handled by taking account of it in flag in the script. Of course, that flag isn't listed so, unlike the other flags, we don't really know what it is. I know how it affects the result with my Speedlight Gold account, but that's all I know about it. I assume it will be the same with all Speedlight Gold accounts.

[Phil Deakins]

11 minutes ago, Prokofy Neva said:

I'm not 100% certain the script is picking up ALL bots. Evidently it can only detect REGISTERED scripted agents whose owners are conscientious. But unregistered? I see some avatars with names that sound like they're bots, yet the script isn't saying anything about them.

It does NOT detect bots. We've already established that. It detects registered scripted agents. That's all. Sometimes it will detect scripted agents that aren't actually registered as such if the example code published by Signal Linden is used at face value.

Edited June 29, 2023 by Phil Deakins

[Innula Zenovka]

/me registers a scripted agent that runs around the region asking people to check all the squares containing traffic lights in the picture it's showing them.

Edited June 29, 2023 by Innula Zenovka

[Prokofy Neva]

3 minutes ago, Phil Deakins said:

It does NOT detect bots. We've already established that. It detects registered scripted agents. That's all. Sometimes it will detect scripted agents that aren't actually registered as such if the example code published by Signal Linden is used at face value.

Scripted agents ARE bots. Bots ARE scripted agents. This is word salad. There may be unregistered agents serving as bots; there may be bot-like objects that scan, etc. but aren't avatars. Who cares??? The main concerns are about AVATARS that COME ON YOUR LAND, sometimes 30 an hour from one notorious company, I've counted them. THEY are registered. We now can ban them on islands, but not Mainland. This script helps to ban them. The end. Stop the edge-casing. If it detects scripted agents that aren't registered, good!

 

[Scylla Rhiadra]

1 hour ago, Coffee Pancake said:

They can. All clients must identify themselves and be honest about it, there are also easily identifiable differences in how they talk to SL that gives the game away.

I think they have no desire to start an arms race.

So, I don't know if this is possible, but I'd love someone to provide something close to a definitive statement on this.

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.

Separately, I think you're right about LL not wanting an arms race -- although, to some degree, it may be too late for that: they've created one. But honestly, I'm surprised they ever actually published this new coding function; it would be more like LL to notice that the public furor had died down, and quietly shelve it.

[Scylla Rhiadra]

3 minutes ago, Prokofy Neva said:

The main concerns are about AVATARS that COME ON YOUR LAND, sometimes 30 an hour from one notorious company, I've counted them. THEY are registered. We now can ban them on islands, but not Mainland. This script helps to ban them.

I want to highlight this because it has a lot to do with the outcry against bots -- probably much more than privacy concerns, to be honest.

The enormous and very visible upsurge in bots -- registered or unregistered -- has demonstrably impacted on some people's SL in negative ways. Here's Lou Netizen's figures on (registered) bots on the grid (which includes only those belonging to "cohorts" he follows). I was in error when I said May was the busiest month, but it's still up there. Interesting to note that the numbers are trending downwards, but they are still much higher than they used to be:

If you are unfortunate enough to live on a mainland parcel where these bots tend to spawn, you're going to be flooded with them. Personally, I don't much care how many appear in my parcel, but I don't get that many, and my parcel is public anyway. But given that we do in SL tend to value "privacy" and property ownership, it would be odd argue that landowners should have control over entry onto their property but not give them tools to prevent them from being swarmed by random bots.

Qie has argued that this new script is simply going to mean that more bots go unregistered, and he's undoubtedly correct. I wonder if a better solution, in some regards, might have been to restrict spawning points to public land -- highways, or maybe abandoned land.