Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
Buy OSRS Gold

Sell OSRS Gold
Jackthedon

[I] Human Anti-ban Response 'Cloak'

Recommended Posts

Warning: Half-baked Idea

Would it be at all possible to create a script that micromanages the scripts you want to run much like Fluffee's laniax's LAN Manager except the only aim of this is to implement additional anti-ban on top of ABC2?

For instance: ABC2 implements a purposeful misclick, and immediately after this the 'cloak' script effectively takes control and implements a human response to misclicking, for example Frustration. As a result of the misclick a typical player may occasionally show an emotional response in terms of their mouse control, a shake or spasm. The anti-ban script would do this then quickly relieve control back to the original script which would correct for it and carry on.

If this is possible, you could implement multiple emotional responses to instances in the game such as confusion, excitement etc. which could give a more life-like appearance heuristically. Also, much like ABC2, it would be universally applicable across all scripts.

I'm not a scripter so I've no idea how possible this is or if it would be worth the time invested but it popped into my head while revising this morning (revision time well spent) and I figured I'd throw it out there.

Edited by Jackthedon
Daydreaming about Fluffee
  • Like 3

Share this post


Link to post
Share on other sites

If there is i am willing to test it. I dont think you can do that without the script or cloak being 5000+ lines long. because of multiple different emotions, multiple responses. Because people act differently. Like you wont spasm everytime you misclick. Maybe when you do a long task or something you might do that.

 

Share this post


Link to post
Share on other sites
1 minute ago, gosubear said:

If there is i am willing to test it. I dont think you can do that without the script or cloak being 5000+ lines long. because of multiple different emotions, multiple responses. Because people act differently. Like you wont spasm everytime you misclick. Maybe when you do a long task or something you might do that.

 

Yeah it could get pretty huge as more and more emotions/responses get added.

In my mind if you had an account doing something where there is a potential rare drop/occurence (as an example), whip from abby demons or a pet drop etc, a bot wont heuristically appear different where other accounts will do (I don't think).

Think of it as a heart monitor, going up and down to account for the responses to positive/negative events in the game. For a bot it will typically just flatline right? Because it doesn't matter if the abbysal demon drops a whip or coins, it picks up the loot it carries on. Whereas a player will most likely have alternating responses to these events causing the sinusoidal appearance. A positive peak could be a pet drop, a negative peak could be a jar drop. That's how I imagined it in my head at least.

Share this post


Link to post
Share on other sites
15 minutes ago, Jackthedon said:

Yeah it could get pretty huge as more and more emotions/responses get added.

In my mind if you had an account doing something where there is a potential rare drop/occurence (as an example), whip from abby demons or a pet drop etc, a bot wont heuristically appear different where other accounts will do (I don't think).

Think of it as a heart monitor, going up and down to account for the responses to positive/negative events in the game. For a bot it will typically just flatline right? Because it doesn't matter if the abbysal demon drops a whip or coins, it picks up the loot it carries on. Whereas a player will most likely have alternating responses to these events causing the sinusoidal appearance. A positive peak could be a pet drop, a negative peak could be a jar drop. That's how I imagined it in my head at least.

I understand where you are coming from but they would have to do different spasm patterns and stuff is what i mean. Also they should really make a legit responder. Like how they recorded human mouse movements. Do the same thing for responses.

Share this post


Link to post
Share on other sites
25 minutes ago, solononforever said:

As an ex-jagex employee, I know that the most powerful part of our bot detection system is the masturbation detector. We scan players to see if they go afk every 4 hours of game play for 2 - 3 minutes followed by a short level of boosted efficiency. This is the only way to beat the system.

4 Hours? The friction burn is real.

Share this post


Link to post
Share on other sites
5 hours ago, Jackthedon said:

For instance: ABC2 implements a purposeful misclick, and immediately after this the 'cloak' script effectively takes control and implements a human response to misclicking, for example Frustration.

Let's say this is possible. In order for the cloak program to effectively take over, it needs to know when to take over. And in order to do that, it needs to have a clear idea what a misclick really is. And a misclick is highly situational, meaning a misclick in one situation might actually be the desired action to be performed in another situation. Not only that, but a script that implements an antiban system could immediately be seen as a misclick. If say TRiBot's default antiban decides to open the combat tab to check the player's level, the clicking of the tab might be seen as a misclick because it doesn't align itself with the goals of the script in its current context.

The best way to do this is for these emotional responses to be hardcoded alongside whatever is causing the misclicks to begin with. This prevents all of the issues I mentioned above.

Share this post


Link to post
Share on other sites

@IceKontroI So in essence you're saying that a purposeful misclick isn't identifiable when compared with something such as checking the combat tab?

I figured there would be a clear distinction between something like a purposeful misclick and idling mouse tasks, then again I'm not a scripter so forgive me :angel:

Kind of throws the idea out to some extent.

Share this post


Link to post
Share on other sites
Just now, Jackthedon said:

@IceKontroI So in essence you're saying that a purposeful misclick isn't identifiable when compared with something such as checking the combat tab?

I figured there would be a clear distinction between something like a purposeful misclick and idling mouse tasks, then again I'm not a scripter so forgive me :angel:

Kind of throws the idea out to some extent.

Well think about how you would describe to someone that a particular click was or was not a misclick. If I were say power mining iron ore and I misclicked while dropping the ore, I might have clicked the "examine" or "cancel" option instead of the "drop" one. The script would need to understand that the objective in that situation was to click the "drop" option, but instead that a different one was clicked instead.

Share this post


Link to post
Share on other sites
5 minutes ago, IceKontroI said:

Well think about how you would describe to someone that a particular click was or was not a misclick. If I were say power mining iron ore and I misclicked while dropping the ore, I might have clicked the "examine" or "cancel" option instead of the "drop" one. The script would need to understand that the objective in that situation was to click the "drop" option, but instead that a different one was clicked instead.

I'm guessing that there's no way to distinguish the between the two then otherwise it would be possible. Is it not possible to view it from the script side instead of the ABC2 side? In the way that an action is attempted but doesn't succeed and would require a retry/failsafe? Checking all bases :D

Share this post


Link to post
Share on other sites
1 minute ago, Jackthedon said:

I'm guessing that there's no way to distinguish the between the two then otherwise it would be possible. Is it not possible to view it from the script side instead of the ABC2 side? In the way that an action is attempted but doesn't succeed and would require a retry/failsafe? Checking all bases :D

You could add support for this type of feature from a script side. Most functions that perform an action in-game return true if they succeed, or false if they fail. If an action fails, you could perform an emotional response to the failure.

Share this post


Link to post
Share on other sites

I was actually thinking along the lines of this last night,

i.e, using your mouse to hover underneath words in chats as if someone were actually there reading it.

IMO there are much more breakthroughs with anti ban we can work to over time, I believe recording humans to be the best way.

Share this post


Link to post
Share on other sites
5 minutes ago, JoeDezzy1 said:

I was actually thinking along the lines of this last night,

i.e, using your mouse to hover underneath words in chats as if someone were actually there reading it.

IMO there are much more breakthroughs with anti ban we can work to over time, I believe recording humans to be the best way.

I think being able to add 'emotional' or habitual responses based off of in-game stimuli like you stated would be a great way forward, but yes it would most likely require a vast amount of data collection from players. I think it would definitely be worth it though in the long run.

Share this post


Link to post
Share on other sites

No, this is not feasible.

You could do this for "traditional anti-ban" such as moving the camera or mouse so often, or doing an action so often.

The problem with "traditional anti-ban" is that it usually has the opposite affect of what you're trying to achieve. Because it's super easily detectable.

 

Let's play a game to show you. Find the pattern:

  • 3,6,4

You probably don't see it, right? No idea what it is. Could be any number of things, but you can't know.

Now try this pattern:

  • 4,5,1,9,6,3,5,2,7,1,7,9,3,4,2,6,7,8,3,5,2,6,3,2,9,4,8

What's the first thing you notice? They are all single digit (IE: between 1 and 10).

 

From this, we know that whatever is creating these numbers is creating them from 1-10 specifically. What if I told you to do an action periodically, but you HAD to do it in intervals of 200,400 milliseconds? Could you do it? Probably not. 

Patterns with limits are not human.

 

Every time a script does something "periodically", it opens itself up to easy detection. The built in antiban at tribot does actions periodically, but they are produced randomly without actual limits and are based on environmental data.

 

See, humans don't "do things periodically". If a human does something every 4-8 seconds, it's not because we're programmed to, it's because of some outside stimuli. Remove or change the stimuli, and our actions change. This is how real antiban works. It needs to based on something specific (NOT time).

 

What you're suggesting is a bit different, I'll admit. However, first of all, I don't think humans try to show the computer his or her emotions through mouse movements. I really don't the computer (or RS) cares. 

The problem is still similar. How do we decide if the antiban should take over? Certainly not every time. And certainly not "periodically". What environmental factors affect how often humans show this behavior? If you can't answer that question, it shouldn't be implemented, because heuristic analysis will see right through it unless it perfectly matches human data.

 

So what do we do instead?

We make our bots play like humans instead of introducing gimmicks designed to workaround the detection system.

Share this post


Link to post
Share on other sites

@wastedbro I'm not sure you quite understood how I intended for this to come across or perhaps I explained it poorly.

My initial reasoning for this potentially not being feasible is how complex the code would be or how much effort it would require. If this is not feasible due to detectable patterns then it would stand to reason that the Human Mouse Movement for VIP-E is not feasible. The only difference is that this would be a Human Reaction Emulator or something of the sort. It would most likely require data collection of a similar standard and method.

I don't quite see why anything to do with the patterns would be relevant to this when it is based off of the in-game stimuli. The pattern would therefore be some essentially random percentage of the 'Events' that trigger a response from the player. I guess the percentage chance of someone responding that way to an individual stimuli would essentially become a pattern, but this is a pattern that would take a LOT longer to detect as it wouldn't occur all the time and may help lower earlier detection.

Imagine a pet drop. Do you realise how many people have a similar response? They spam "asred.tghsr.dlfihslidrhe.sizdhf pet" or something similar in the chat if others are nearby or in a cc or something. What's the pattern? It's that it happens when they get a pet drop (Or an incredibly rare drop) and it doesn't when they don't. In general, a person who responds this way for a single pet, will do so for EVERY single pet. It stands to reason that their mouse response may also differ in a similar fashion. With enough 'emotional' and 'personality' variables as mentioned in previous posts you could essentially create randomized profiles just as scripts often do themselves in terms of mouse speed and all that jargon, again the only difference being that this will be specifically for responses triggered by specific stimuli.

Now that's quite a specific case, I understand that, but if you could identify other stimuli on a smaller scale (not saying it's easy) then in my mind there could be something to this.

 

 

Share this post


Link to post
Share on other sites
4 hours ago, Jackthedon said:

@wastedbro I'm not sure you quite understood how I intended for this to come across or perhaps I explained it poorly.

My initial reasoning for this potentially not being feasible is how complex the code would be or how much effort it would require. If this is not feasible due to detectable patterns then it would stand to reason that the Human Mouse Movement for VIP-E is not feasible. The only difference is that this would be a Human Reaction Emulator or something of the sort. It would most likely require data collection of a similar standard and method.

I don't quite see why anything to do with the patterns would be relevant to this when it is based off of the in-game stimuli. The pattern would therefore be some essentially random percentage of the 'Events' that trigger a response from the player. I guess the percentage chance of someone responding that way to an individual stimuli would essentially become a pattern, but this is a pattern that would take a LOT longer to detect as it wouldn't occur all the time and may help lower earlier detection.

Imagine a pet drop. Do you realise how many people have a similar response? They spam "asred.tghsr.dlfihslidrhe.sizdhf pet" or something similar in the chat if others are nearby or in a cc or something. What's the pattern? It's that it happens when they get a pet drop (Or an incredibly rare drop) and it doesn't when they don't. In general, a person who responds this way for a single pet, will do so for EVERY single pet. It stands to reason that their mouse response may also differ in a similar fashion. With enough 'emotional' and 'personality' variables as mentioned in previous posts you could essentially create randomized profiles just as scripts often do themselves in terms of mouse speed and all that jargon, again the only difference being that this will be specifically for responses triggered by specific stimuli.

Now that's quite a specific case, I understand that, but if you could identify other stimuli on a smaller scale (not saying it's easy) then in my mind there could be something to this.

 

 

You'll have to generalize it. It either works globally or not at all. 

Emotion is enormously complex. How would you measure what emotion to show and when?

When is the pattern part. It must be answered because if you just make it up, it forms an arbitrary pattern.

 

I suppose I am confused on what you're suggesting. I think you're way too vague. 

Share this post


Link to post
Share on other sites

@wastedbro Would, in your eyes, an emotional pattern be worse than no emotional pattern?

Because wouldn't that be in itself a detectable thing that every script could potentially get picked up on? Much like my flatline metaphor. People don't play any game stone cold emotionless, they get  hyped, excited, frustrated and I struggle to comprehend how this would not be different to the typical game play.

People do have emotional patterns based on stimuli, it's essentially what defines you as a person. A pattern, provided it isn't identical to many/any others isn't necessarily a bad thing. If someone randomly breaks my door down right now I'm gonna shit myself, if they do it tomorrow i'm gonna shit myself, it's not the case that if they do it on Sunday I'm gonna sit here, be chill and ignore them. The same applies to someones response to an unexpected event in-game. Yes the response will be much more discrete but I'd wager it's measurable and can be near-enough generated,  given enough real data.

With regards to the when, to me in a pseudo-code sort of way it's simple to visualise, as an identified misclick wouldn't trigger the ecstatic response that it would for a pet drop. So long as the stimuli can be identified I see no issue, if that makes sense? With regards to identifying them I agree it could be a mammoth of a task.

I mean you're a scripter so I can appreciate and respect that your knowledge and understanding will be greater than my own. I just don't quite see where it becomes unfeasible from a logical standpoint. I'm not trying to be vague at all I'm simply trying to open up a new viewpoint of how we could approach anti-ban and I'm far from perfect.

Share this post


Link to post
Share on other sites

In addition to everything I've suggested. IF we assume that the events reported by numerous people including myself about people's game denying any input for 30-60 seconds while still being connected enough to see everyone else acting as normal is actually a 'test' of your response to the given situation then my idea would combat this. A human response to this occurring is HUGELY different in comparison to a bot in the same situation, I say this with confidence but I've seen one of my own encounter this and be swiftly banned within 24 hours, AND i've encountered it myself while playing manually on a botted account which is still alive to this day. This could be pure coincidence but on the off-chance that it isn't, my idea of an emotional response to the given event would be critical.

Share this post


Link to post
Share on other sites

An emotional pattern is far worse than no emotional pattern.

Let's say you make all bots move their mouse in frustration to a certain event. 

Now let's say only 30% of real humans even do that. Now let's say those real humans only do it 40% of the time.

Compare that to thousands of bots who do it every time.

Cut and dry bot detection. You just added a way for Jagex to detect you. You made antiban worse. 

 

Jagexs detection uses data over time. Considering most players don't get a ton of pet drops, and 99% of bots don't even do anything with a chance at getting an pet drop, why add it? There's no way to tell a difference between a bot and a human when it comes to pet drops. What if 20% of humans don't say anything when a pet drops. That would make detecting it inaccurate.

 

So far you have:

- move mouse randomly when missclick happens (very detectable)

- say random nonsense when a pet drops (does not define real human behavior)

Share this post


Link to post
Share on other sites

@wastedbro But no pattern IS a pattern? If anything the most detectable thing you could have a bot be is emotionless?

Yes I understand a pattern is detectable, but I don't see how it's more detectable than the pattern of no emotion?

Not to mention they aren't banning based on detecting ANY pattern. If that was true you'd have real people getting banned left right and centre.

I appreciate your input, but I figured you would be a bit more constructive given your ABCv3 Antiban Development thread and the whole aim of that is for the anti-ban to be more human, yet you wont adhere to emotion which is a key part of being human..

If me stating that is wrong, then you agree with the general direction of this thread and you're arguing semantics.

But I get it, you say it's not feasible, fine, I'll back up, take it on the chin and crawl back into my hole.

 

Share this post


Link to post
Share on other sites
27 minutes ago, Jackthedon said:

@wastedbro But no pattern IS a pattern? If anything the most detectable thing you could have a bot be is emotionless?

Yes I understand a pattern is detectable, but I don't see how it's more detectable than the pattern of no emotion?

Not to mention they aren't banning based on detecting ANY pattern. If that was true you'd have real people getting banned left right and centre.

I appreciate your input, but I figured you would be a bit more constructive given your ABCv3 Antiban Development thread and the whole aim of that is for the anti-ban to be more human, yet you wont adhere to emotion which is a key part of being human..

If me stating that is wrong, then you agree with the general direction of this thread and you're arguing semantics.

But I get it, you say it's not feasible, fine, I'll back up, take it on the chin and crawl back into my hole.

 

That's highly dependent on what the human pattern is. Until we specifically have thousands of empirical data to back up your assertion, it's not really possible to say that your pattern is better than no pattern. The measure is how close we are to human.

So if only 30% of humans do that action, then yes, not doing the action every time is actually closer to human than doing it every time. That's not considering the actual action and variations of the action, which can be immense.

 

The second part of this, is that you can't code emotion. You can code tacky gimmicks that look like certain emotions, but that's not really human. Emotion itself is far too complex. Not even Google an Microsoft have the technology for that (although they are working on it). We're talking super computers here.

Emotion is simply not feasible because it can't be emulated. It can be imitated to an extremely shallow extent, but it's kind of like watching a bad actor on TV or in a play.

 

Let me end this by saying I'm not saying your ideas are invalid. But they aren't really ideas yet. They are vague concepts. 

 

Let's take for example reaction times:

  • We have, through statistical analysis, correlation and regression, have heavy evidence to suggest that human reaction times are affected by Idle Time, Expected Idle Time, Combat Status, Hovering Objects, and whether or not the mouse is on screen.

Therefore, we can use these "predictors" to determine reaction times that we know look human over thousands of predictions.

 

Let's take for example emotions:

  • We have [insert evidence here] that suggests pseudo-random mouse movements change in frequency after a given event (IE: missclick).

Until that evidence is blank, we have no idea how missclicks affect such actions. Therefore, if we just implement it without that key piece of data, it's basically a 50/50 chance of either moving towards being human or away. 

 

I am seriously asking, do you see why?

Share this post


Link to post
Share on other sites

@wastedbro I understand what you're saying.

The only part I disagree with is "30% of humans do that action"

Lets say bot#1 performs the 'emotion' actions throughout it's gameplay and therefore looks like the 30% real of players that perform similar responses.

Lets say bot#2 doesn't perform the 'emotion' action, then this looks like 0% of real players because no real players display no emotion at all. (you may disagree)

That's my perspective on how adding it to some extent is better than not at all, because at least it's looking somewhat human. I know it's not as black and white as that, and yes over time it'll be detectable I'm sure, but surely that time is longer than being detected for general mouse movement? It would also delay being detected for general mouse movement because you're adding an aspect that isn't general.

I agree I have been too cavalier with saying "add emotion", by that mean trying to recreate it from collected data much like the Human Mouse Movement we use.

Share this post


Link to post
Share on other sites
5 minutes ago, Jackthedon said:

@wastedbro I understand what you're saying.

The only part I disagree with is "30% of humans do that action"

Lets say bot#1 performs the 'emotion' actions throughout it's gameplay and therefore looks like the 30% real of players that perform similar responses.

Lets say bot#2 doesn't perform the 'emotion' action, then this looks like 0% of real players because no real players display no emotion at all. (you may disagree)

That's my perspective on how adding it to some extent is better than not at all, because at least it's looking somewhat human. I know it's not as black and white as that, and yes over time it'll be detectable I'm sure, but surely that time is longer than being detected for general mouse movement? It would also delay being detected for general mouse movement because you're adding an aspect that isn't general.

I agree I have been too cavalier with saying "add emotion", by that mean trying to recreate it from collected data much like the Human Mouse Movement we use.

0% is closer to 30% than 100% is, so.....

Not performing the action would be closer to human in this case.

 

But this all hypothetical. Means nothing until data is collected. I have no idea how you will possibly measure emotion, though. You're very much welcome to try. I'm always open to actual data.

Edited by wastedbro

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Our picks

    • Over the past few months, I’ve been working diligently on a new project - TRiBot X. Everything has been written from the ground up, with all of the best practices of software engineering. Every aspect of TRiBot has been re-imagined to support three main goals: flexibility, useability, and reliability.
        • Like
      • 35 replies
    • Come give us feedback on the next version of TRiBot!
        • Thanks
        • Like
      • 75 replies
    • TRiBot is looking to improve a lot of its customer relationship management, customer on boarding process, customer experience, design elements, community engagement and pretty much everything else you can imagine when it comes to marketing.

      Our goal: To ensure that the marketing done TRULY reflects the experience and does not shine an inaccurate light on what TRiBot is lacking in.

      So I ask, what do you love about TRiBot and what do you hate about TRiBot? What does O S Bot, Rune M8, PowR Bot and Dre amBot do better? (yes I purposely didn't spell it right 😂).

      Love, 

      RileyZ
        • Like
      • 23 replies
    • Over the last three weeks, I've been working on upgrading our server infrastructure. It's finally ready and is now live!

      Why?

      Increased reliability - less server errors


      Increased availability - less downtime


      Increased security - keeping us and you secure


      Increased capacity - ability to serve you better


      Increased speed - less waiting for things to load


      Faster development - server and service updates will come faster


      What are the changes?

      Move from a single AWS EC2 instance to AWS ECS (Elastic Container Service)


      Distributed computing


      Load balancing


      Git management of server files and filesystem


      Redis caching


      How?

      AWS ECS (with 10 EC2 instances)


      AWS ElastiCache (Redis)


      AWS Load Balancing


      AWS EFS (Elastic file system)


      Please bare with us as I continue to tune the server for maximum performance. Slow loading speeds may occur temporarily. I thank everyone for their patience.

      Please post on this thread if you experience any issues other than slow loading times.
        • Like
      • 51 replies
    • This release will:

      Fix prayers and world hopper API (Thanks @JoeDezzy1 and @erickho123)


      Improve banking API (Thanks @Encoded)


      Adds methods for returning and using Java Lists, rather than arrays


      Slightly randomizes some hardcoded behaviour


      Removes sleeps from waitConditions; the efficiency saving potential is negligible in these use-cases, therefore cleaner code is preferable


      Other back-end improvements





      Note: If you are using LG, please restart both the RS client and TRiBot.
        • Sad
        • Haha
        • Thanks
        • Like
      • 90 replies
  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...