Welcome!

By registering with us, you'll be able to discuss, share and private message with other members of our community.

Sign up now!

How Jagex's detection could work, in theory?

Ya dig?

  • You need to stop smoking so much weed

    Votes: 8 53.3%
  • You actually make some sense you brilliant bastard

    Votes: 7 46.7%

  • Total voters
    15
Joined
Mar 24, 2017
Messages
341
Jagex has algorithms to compare how you play to a human player. Like the RuneMate client gathers data

So, every time a player clicks a chicken, or buries a bone, the data is logged. The time he did it, the movements of his mouse, his keyboard, his camera, if he right clicked or left clicked, if he moves the mouse off screen, if he checks his quests, stats, talks to a friend, etc etc etc. Everything is logged.

So after developing algorithms, and back-testing them on millions of players they can end up with some solid numbers for how an actual player acts, roughly. Every player is different, so putting down in numbers what being a human means is rough, which is why they probably have many different levels of flags to decide if a player is a bot or a human.

Here's my idea on how some of these flags could look like, and their levels

High indicators:

  • Plays over 12 hours a day (Or some other high number)
  • Is F2P
  • Is level 3
  • Uses unrecognized third party client
  • Makes over X amount of gold a day (They really don't like gold-farmers)
  • Over X amount of clicks a day (Imagine clicking an average of 2 clicks a second a day, no human does this. A flawed bot could)
I could probably think of 20 more things, but we have no way of knowing which things they look for.

Medium indicators:

  • Has no/few friends on the account. (Say an average player has 1.6 friends, and the average bot has 0.4)
  • Doesn't use chat (Average player 20 words a day, bot 1 word a day)
  • Only levels select skills (Average player gets xp in 6 different skills a day, bot gets xp in 2)
  • Mouse click timing (Player clicks within 2 second when hovering an object/npc etc, bot clicks within 0.4 sec or too long)
  • Over x amount of reports

I'm sure you can think of a load of different other indicators that would set a player and a bot apart. But you get the idea.

My theory is that when a player receives over X amount of flags, or when the flag "points" reach a certain value, certain actions happen on their end. Maybe it's a scale. Like from 0-10 how "bottish" a player is. At 1-3 they're low priority to investigate. At 4-9 they're medium and might require Mod review. At 10 they're guaranteed to be bots and are banned automatically.

I'm aware that many of these flags could be something a legit player would do as well. That's why they need more than one flag, or two, or three perhaps, to be even considered.


Am I making any sense? :p
 
RuneMate Staff
Joined
Oct 2, 2015
Messages
3,225
Copying a previous post of mine below - Don't get me wrong, discussion on this is interesting but there are too many wannabes out there who think they know the answers to everything. At the end of the day, we know very little about Jagex's true methods.



I've been around the botting scene for the better part of 6 years now. I've been through Powerbot and the infamous "hack", through RSBuddy's legendary days and more recently as a proud user of Runemate. I've also tried a plethora of other bot clients out there during this time with varying success.

During the years, I've observed 10s if not 100s of theories about Jagex's bot detection systems and how we can best stay 'safe' from being banned. Some of these theories do in fact sound reasonable and practical, while others sound out of this world.

My thoughts have always been the following:

1) Never bot on an account you do not want to lose
2) Be prepared to have all of your accounts banned, if you try out macros.
3) Bot sensibly and you will be okay

I've only ever had one account banned in my entire botting "career" and it was last week on OSRS - I made a new account and was "suicide" training it - It got banned after about a week of play time and only after training fishing for 18 hours straight.

At the end of the day, the theories bot users (and even to a lesser extent, the developers) put out there about Jagex's detection systems mean little to me, as they are just theories.

Only Jagex themselves know what their real detection methods are, and I'm sure they are pretty keen on keeping those methods a secret for as long as possible.
 
Superior All Knowing Person!
Joined
Oct 9, 2016
Messages
642
I've had a few 80-90 hour sessions, which was only interrupted by runemate forced updates.. or runescape maintenance updates.
there's a pic somewhere of my 114 hour session, without breaks. ( @Party can confirm it. )
I've been running 1 account for 5 months now i believe, 24 hours a day without breaks :) (only down for the 1-2 min a week for either runemate/runescape updates)..

imo. it has nothing to do with hours, otherwise i'd obviously be banned ages ago.
imo, it also has nothing to do with gp/h made, i've made 963k gp/h for 5 months straight..
imo, it has nothing to do with your clicks, otherwise i'd be banned too.
imo, it has nothing to do with friends on the account - i have 0 on all my botting accs.
imo, it has nothing to do with not chatting - since mine doesn't.

From what i've tested;
imo, it has EVERYTHING to do with player reports.

and as @kazemanie says, in general.. don't bot on a account you don't want banned, that's the safest way to keep your account ;)

- also, if jagex really wanted bots gone.. they would be gone, but it would be bad for business.
/imo.
 
Java Warlord
Joined
Nov 17, 2014
Messages
4,906
More than half of your example flags are perfectly normal for legit players.
 
Joined
Dec 28, 2013
Messages
190
Afaik the code to upload mouse movements hasn't been active for years. Analyzing mouse movements would be pretty unreliable (and a waste of resources) when there's plenty of other ways to reliably detect bots.

I really do think it's primarily a cross reference of behavior among all the players in a certain region. I think this is why private and premium bots tend to have a dramatically reduced ban rate, since only a few accounts are using the same bot at the same time.

Although if anyone remembers InuBot that was 100% action based (do/processAction bot, literally didn't use mouse/key events) and received no bans until someone leaked the method to Weath. If bans were based off of behavioral analysis I would think this would be detected regardless. Maybe there's some intermediary that's the key to discovering what data Jagex really does use?
 
Only Jagex themselves know what their real detection methods are, and I'm sure they are pretty keen on keeping those methods a secret for as long as possible.
Exactly. I wish developers would stop trying to beat a system based on speculation and instead concentrate on improving their bots quality. Pseudo antiban only looks good to the uneducated consumer.
 
Joined
Nov 3, 2013
Messages
2,387
if you wanna know how to dodge, check my guide. been botting with you lads since 2009 and trust me, i know the game and the scene
 
Joined
Feb 18, 2017
Messages
176
There's no way jagex has the machine power to compute such data on each of the players. Assuming a player action (mouse movement, etc like you say) every 2 seconds which generates 5 data, and the average player plays 20 minutes, you have 3000 data per player. Even if such advanced algorithms only took O(n^2) and had good constant factor jagex would be dying under the load of computation for some 400000 total players online daily and even more on the weekends. And these are conservative numbers.
 
Top