Is OSBot using machine learning to improve antiban?

[Alek]

Bans in a nutshell:

Write a private script and only use it for yourself, write in all the antiban and bullshit you can dream of. Bot until you get banned. Come on the forums and tell other people that your antiban method worked . 

[HeyImJamie]

51 minutes ago, Alek said:

Bans in a nutshell:

Write a private script and only use it for yourself, write in all the antiban and bullshit you can dream of. Bot until you get banned. Come on the forums and tell other people that your antiban method worked . 

Where does your hate from 'Antiban' stem from when it's fairly common knowledge that bans are a result of the data analysts Jagex has employed? I can appreciate that 99% of 'Antiban' that scripters implement is worthless.. i.e. Computer-generated random ints, but still.. stating every suggestion is pointless seems a bit backwards to me. 

[Alek]

39 minutes ago, HeyImJamie said:

Where does your hate from 'Antiban' stem from when it's fairly common knowledge that bans are a result of the data analysts Jagex has employed? I can appreciate that 99% of 'Antiban' that scripters implement is worthless.. i.e. Computer-generated random ints, but still.. stating every suggestion is pointless seems a bit backwards to me. 

Name one suggestion that you believe hasn't been tried before.

[botnot]

16 hours ago, Alek said:

Name one suggestion that you believe hasn't been tried before.

Applying real user data to customize scripts? I mean, not that I think it hasn't been thought about, but I'm yet to see any clients to implement something like this (although this isn't a domain I'm overly familiar with). Would this not make bans far less likely if its mimicking the play-style of the actual account owner? 

[Criboo]

17 hours ago, HeyImJamie said:

Where does your hate from 'Antiban' stem from when it's fairly common knowledge that bans are a result of the data analysts Jagex has employed? I can appreciate that 99% of 'Antiban' that scripters implement is worthless.. i.e. Computer-generated random ints, but still.. stating every suggestion is pointless seems a bit backwards to me. 

How is it common knowledge?

It's pure speculation.

[botnot]

9 minutes ago, Criboo said:

How is it common knowledge?

It's pure speculation.

How do you think they find out who is a bot and who isn't? The masses of data they gather from their players is certainly used in their programs to figure out if an account is botted or not. Nobody is manually looking at the data and making these decisions

[spanishguy]

11 hours ago, botnot said:

How do you think they find out who is a bot and who isn't? The masses of data they gather from their players is certainly used in their programs to figure out if an account is botted or not. Nobody is manually looking at the data and making these decisions

Agree. They probably have bot data and real player data. If we can gather the same real player data as them that helps and simulate the playstyle, then how can they detect it. Its not an easy project though.

[botnot]

7 hours ago, spanishguy said:

Agree. They probably have bot data and real player data. If we can gather the same real player data as them that helps and simulate the playstyle, then how can they detect it. Its not an easy project though.

Absolutely a difficult project, although collecting real user data may not be as hard as initially thought. I imagine a Runelite plugin to track in-game interactions wouldn't be too difficult to implement, although the real challenge would be how those interactions change when doing specific activities in game. If such plug-in did exist, I'm sure the botting community would be happy to use it and send data back

[Alek]

On 7/5/2018 at 1:55 PM, botnot said:

Applying real user data to customize scripts? I mean, not that I think it hasn't been thought about, but I'm yet to see any clients to implement something like this (although this isn't a domain I'm overly familiar with). Would this not make bans far less likely if its mimicking the play-style of the actual account owner? 

This has already been done at a competitor bot when I used to be a scripter there (2013), again since I've left, and at least one time at OSBot. 

4 hours ago, botnot said:

Absolutely a difficult project, although collecting real user data may not be as hard as initially thought. I imagine a Runelite plugin to track in-game interactions wouldn't be too difficult to implement, although the real challenge would be how those interactions change when doing specific activities in game. If such plug-in did exist, I'm sure the botting community would be happy to use it and send data back

Not difficult, typically the person running it will give a script to players and have them run it for a few hours while they actually play the game. From that they can determine a bunch of things. Like I said though, it's already been done. There is nothing on these forums which hasn't been thought of and implemented - which hasn't been done in 15 years.

Edit: At least the majority of "new suggestions" and "new ideas", it's the same round of people who are new to botting re-suggesting the same stuff.  

[Eliot]

On 7/4/2018 at 4:37 PM, lizzzerd said:

all data collected is sent to the H I V E for analysis

SOON.

Edited July 6, 2018 by Eliot

[Logicfury]

The more a script is being used in the game (popular F2P or P2P scripts), the higher the chance it gets you banned. The less data points your account has accumulated outside of botting, the higher the chance it gets you banned. The longer you bot on your account, the more the data points you produce begin to fit a very clear and robotic pattern (RE: F2P woodcutters all synchronized going from one tree to the next). Combine all 3 (what a lot of greedy players do) and what you get is a 99.5% confidence interval that you are indeed scripting when you run any basic statistical analysis on the account. This has already been repeated ad nauseam but most people just don't understand basic statistics and how powerful of a tool it is. 

If you're going to bot a completely fresh account, I would recommend balancing your botting this way: 50% real play, 50% botting. In other words, if you cannot accumulate 1 hour of legit playtime, don't frickin knock the house down and bot 3-4 hours of woodcutting. Just take a look at how synchronized the F2P woodcutter suicide bots are. It doesn't matter if you just bot 3 hours a day with 6 breaks in-between. If ALL you do is bot on that account your data points are still going to put you in that 99.5% confidence interval once you accumulate enough hours.

Oh and IP flagging is 100% real. Once you're flagged (get banned a few times), you're definitely going to want to unplug / remove the battery from that modem.

Good luck.

Edited July 8, 2018 by logicfury