Bot detection TRIAL

[darkfhu 0]

After having a few bots banned recently, I decided I am going to try and find out what their current detection system they in place. Many people have their theories as to why bots are banned (age of account, areas farmed, mac address, ip address, certain settings, time of day, trading between accounts, etc etc.

 

How do I plan to accomplish this?

I my experiment will consist of 5 accounts. These accounts will run all with the primary setup:

 

Each account will have a completely random username (obviously not account1, account2, account3, etc).

Each account has already been made or will be made at a completely random time (sorry GGG if you plan on detecting all accounts made on 4-6-16 at 5:30pm)

Each account will be on it's own private VPN that is located at a different location on each of the 5 accounts.

Each account will run 24/7 (for the sake of speeding up the detection of the bot detection process (lol))

Each account will be running in a different area in the game while primarily mapping in hideout

Each account will not trade with each other or any mule account and will strictly trade only with other players for legitimate trades.

Each account will have it's own MAC address on each VM 

Each OS will have a different Username

Each account will level up the exact same way

Each account will use it's own build/class

Each account will have slightly adjusted memory settings

Each account will be running on normal mode (no hide process)

Each account will run a program to shut all applications down if the VPN drops.

 

Reporting:

I plan to report back every week with my findings of still up and running or down and what could have caused this. As bots get banned (which they will), I will begin to tweak certain things with the experiment to try and find links. I will make new accounts to replace the banned ones as they are banned (with different vpn/settings). Some of these accounts are already setup and some will be setup either now or at a future date (for obvious reasons). 

 

Things I have already figured out through being banned about 20 times previously before this experiment:

1. Trading level up gear from an existing bot to level up the account quicker will eventually lead to a multiple ban if either account ever get's banned so don't do this.

2. Level 1 characters and "The Coast" seem to be monitored quite frequantly, especially if you make a level 1 character, then straight trade or drop trade for obvious reasons, there is a chance that your account will get flagged, even if you make a new character and overwrite log (not going to go in detail about this as I will no longer be doing this). 

3. Without any pure evidence, I strongly believe that older accounts are less likely to draw flags as most of my older accounts are still up and running. 

 

Again, your experiences may be different and you may have success with other things that I have not, but hopefully this experience will provide SOME insight to certain links.

 

Suggestions/comments: 

Before I begin this experiment, please offer up any suggestions/comments below that would help with this. Thank you in advance.

Edited April 6, 2016 by darkfhu

[jps42 138]

Pretty cool undertaking, however, what are the specific criteria you will use to fine tune each subsequent bot as they get banned?

 

Unless you have an empirical way to track specific bot habits or past history (not sure the logs will suffice on their own to explain behaviour) then what's to say the bots will get flagged and banned purely by chance rather than something you can identify?  Then this could lead you to false positives or just the overall statement that running a bot 24/7 will eventually get you banned.

 

Pretty sure your still at a high level still on planning this out.  I would make these assumptions, maybe they can be incorporated eventually into your "ban conditions"?

 

Assumptions:

Mapping HO vs Lab - HO less risky

Mapping vs Zone farming - Mapping less risky, even less in HO

Single zone farming vs Rotating zones farming - Rotating less risky

Scheduled idle vs continuous runtime - Breaks less risky

Overleveled farming vs at/underleveled farming - Less risky to farm in zones where you are close to or slightly underleveled to zone level

 

Others not as easy to test out (bc money or exposing home pc/ip) -

MTX vs Free - Less risky if you paid them money

VM vs PC - Not sure at all about this, I would guess some would say non-VM is less risky

VPN vs Static home IP - Does GGG have blacklisted IP ranges

 

 

My 2 cents so far.

[darkfhu 0]

Pretty cool undertaking, however, what are the specific criteria you will use to fine tune each subsequent bot as they get banned?

 

Unless you have an empirical way to track specific bot habits or past history (not sure the logs will suffice on their own to explain behaviour) then what's to say the bots will get flagged and banned purely by chance rather than something you can identify?  Then this could lead you to false positives or just the overall statement that running a bot 24/7 will eventually get you banned.

 

Pretty sure your still at a high level still on planning this out.  I would make these assumptions, maybe they can be incorporated eventually into your "ban conditions"?

 

Assumptions:

Mapping HO vs Lab - HO less risky

Mapping vs Zone farming - Mapping less risky, even less in HO

Single zone farming vs Rotating zones farming - Rotating less risky

Scheduled idle vs continuous runtime - Breaks less risky

Overleveled farming vs at/underleveled farming - Less risky to farm in zones where you are close to or slightly underleveled to zone level

 

Others not as easy to test out (bc money or exposing home pc/ip) -

MTX vs Free - Less risky if you paid them money

VM vs PC - Not sure at all about this, I would guess some would say non-VM is less risky

VPN vs Static home IP - Does GGG have blacklisted IP ranges

 

 

My 2 cents so far.

 

It is currently tough to pinpoint the exact criteria to change in the beginning. This will be decided on how/when/why assumptions that each bot(s) get banned. Hopefully there will be situations where 2/5 3/5 bots get banned, that way it would be a little easier to try and find/guess variances between the accounts still alive and the dead ones based on the logs of each. The important thing is to try and get the accounts banned asap without "trying" that is why I had to implement the 24/7, that way it gives me a basis to start from. I would be pretty upset if somehow all five survived for months as this study would be going nowhere LOL.

 

As you said, some of the accounts could be flagged purely by chance or the fact they are running 24/7. Even more, it is going to be harder to determine with only a pool of 5. There is so many ban conditions. The assumptions you pointed out are great. As I mentioned, I want to try primarily HO mapping first on them. I really do believe account age is a prime factor, but another variable would be, how old does the account have to be before it reduces the risk. Unfortunately, since we are going for quick bans with newer accounts, this is something that won't fit into the conditions. 

 

Another find I just came up with (which at this time is also a theory), but the city as which the VPN is in could have a huge factor. I was running 7 on my last run recently and 4 died. The 4 I was running, where all on different VPNs, but they were in the same city. I looked up the city estimate and it had an approximate 50,000 population. The other 3 were running in larger cities. Of that approximate 50,000, how many people under that city (whether passerby/vpn/etc) actually play Path of Exile? I would say not enough that would put me under the radar. So if they flagged one for something and they have a system to type in a city and see active connections, maybe that could lead to something. Obviously using VPNs with a city estimated population of 3 million+ would be impossible to do this with. I assume they have a way of tracking the city since when you receive the authentication code e-mail when logging from a different location, it specifically has the city listed plain as day there. 

 

It is really hard to do anything empirical with this study. It will be mostly guess work/theory from me and everyone else who has suggestions on this thread for each update. But if we get situations where 2/5 3/5 and even 4/5 die, then we might be able to come up with something. If 1/5 die, well, it could be almost anything, maybe something could be helpful from the logs.

 

It will probably be about two weeks before everything is setup for this as I want to hand level the first set without any boosts, gearups, etc other then legit trading or finding myself.

[ghbdtls 30]

nice one dude....i hoppe u will find the best way to bot 4 you  

[Bart 0]

Jps basically covered it. 

 

What I do suggest is that you act as much as a human as possible, which has been said so much before. Talk in chat every once in awhile, trade, and more importantly, at least to me, don't run for a long time. Just remember that greed will be your downfall. 

[RaZe 148]

Gl with this, hope it gives more sense then the 10+ bans i triggered on purpose while botting or the 10+ times i ran 24/7 and started to cry or the 5+ times i got banned from pure randomness. Hope it gives u clarification, it did not for me and no matter what result you come up with you wont make me happy until u get some inside information from GGG.

[RajtNajder 0]

darkfhu, do you download new POE completely on each VM? I'm pretty confident each content.ggpk has it's own code, should be called hash, so if u only clone VM and don't download new POE with new content, they can link all accounts, as they  have the same number. I had several dozens of accounts banned by now, usually all or almost all of them during one day, and after consulting with programmers and admins this seems to be one of main reasons for chain banning (back then I didn't download new POE on each VM)

[legitplayer 1]

I guess most bans occure from beeing greddy and using cheap? detected vpn providers + not playing by hand at all. Just get a dedicated ip on your own, and not a shared vpn ip. It cost some bucks more, but it will provide alot more saftey since u cant know how many other people are botting on the same ip or were botting on this ip in the past.

Even better would be an own server where you setup your own vpn.

 

I mean i just started botting in poe, but i botted in a couple of other games, and never got caught. It´s all about balance, and i´m quite sure i can beat poe too. I dont plan rerolling my chars every  week.

Just started with 2 bots now, but i plan to expand to 3-4 more when i´m sure those accounts last atleast 1month+. I do alot of normal stuff on those accounts like talking in global, some trades here and there, master rotations. and so on. It´s all about staying under the radar. If u dont invest any time in your accounts, and u just set them farming every day for 8-12hours, never interacting with any players except for doing empty or fake trades to your mules it´s no wonder you get banned after a week.

However i understand that u cant handle those measurements when u are running 10+ bots.The work is getting out of hand there.So you need to think about some other counter measurements.

Just my 2 cents, and i guess this should be common knowledge.

Really curious how far i can come, since i read alot about bans in poe. This seems actually to be a challenge, unlike in most other games.

 

@rajtnajder. interresting point, guess i´will start downloading poe too than.