as soon as they patch a fix, the hackers/botters can pull the source code and find change or just read a blizz blogpost update usually detailing what they changed, when and why. then they make work arounds. itll always be an issue
I agree with you about the data analysis and telemetry to detect bots, but what if the bots are trained to avoid detection? The bot would only need to act non-repetitive and not appear to be working optimally.
The problem would be if a pile of copies of a bot that does this, would show up more and more with it's own set of patterns, maybe on it's own it would avoid detection, but if 100K copies of it showed up, it may start the game the same way with same decisions and movements and clicks, that alone would be a red flag, it may attack the same way, that also would flag, so yes maybe alone it could avoid detection as it very well could be a unique player, but if a pile of players appear to start similar in batches then yeah it's bots, even if it has randomization, even in cryptography you can only do it for so long before the seeming randomness of it gets solved and becomes clear as day, it would have to be more like that 100K copies of the bot is an ai tied to a network, and every ban it sends it's play to the banned play, and updates all ai bots live at same time to alert, otherwise they all act the same rather than as their own. Ai agents are the same when released, but if they adapt then yes they will play like players more and more in time, however to counter it you just need 100k humans to submit 10 games each, then blizzards ai can see patterns, that way if 100K ai bot swarm is trying to not get banned, blizzard can have banned play as well and be where the bot ai's left off, so that way the company has a constant edge over this scenario.
I’ve been thinking about this myself.. I mean it doesn’t take AI to randomize things if you’re good, but I’m trying to understand how these things are even detected. I once was making a bot for fun, and the moment the bot clicked the mouse on the screen the game banned me.
But all you suggest is client side behaviour detection, which could easily be countered as well. Also the amount of data they would need to collect and process and well you showed it on a really simple bot, but you know as well, that you could easily implement more chaotic patterns and fuzzy logic in decision making. I think the issue is, they are game developers and probably not skilled enough or interested in hunting bots and dont have the budget for this task. But you could in general argue, if a game is plagued by bots, its mostly because if the game design. boring, repetitive, time wasting and unrewarding game loops. In a world were free time us getting less and less and people are more stressed and just want to play the enjoyable part if the game. This is what creates the market for RMT and Bots. They will have to fight the reasons, not the symptoms.
Indeed, there are lots of ways to make such bots more "humans" and therefore more difficult to be detected as you said. It's just speculation on my side but I believe this would force to build less effective bots on the long term, so this might still be beneficial to use this approach. And in any case nothing prevents Blizzard to combine behavior detection and more classical approaches together. I guess the design of the game does influence it, I've heard this point a few times. Some people find this type of game fun though, so should we change game design because of bots? Possibly, way but a bit unfortunate I would say. WoW is also plagued with bots, and you could potentially use bots in many of the mobile games asking you to do boring tasks on a daily basis. I suspect emerging Pay to Earn (crypto games) will suffer botting issues (and this could be much more lucrative than on D2R). I believe people can find reasons and interest in botting in many games that are both multiplayer and competitive, so this is a pretty wide range of game types. I haven't done a deep survey of areas where bots are more used, this could be an interesting analysis to do! Thanks a lot for this constructive feedback!
@@romainbardon6959 Well it is also extremely dependent on the game and its economy. Also the payment model of the game. Like D2R now, its pay to play but the only extra revenue for paying the servers are sales of new licenses to the game. So bots can be pretty decent here, you detect them, let them run long enough to be profitable, then ban them so they will buy a new license. I remember back in old WoW times, when a bot got detected, you got your ban email with something like you are a bad boy, you violated our eula, please next time be smarter. Here is the link to buy a new link. The ban mail asked you to purchase a new game key. If you look at WoW, the bots also pay subscriptions, even if they purchase it via a Token, then still someone else, paid for the game time with real money to exchange it for gold. Still they make a profit. Then you have games like Path of Exile in a free to play model. There you could say bots are harmful, as they dont give the developers any gains, but PoE is also a very complex and bloated game where you could say many players would stop playing out of frustration if there would not be an option to buy items via RMT (real money trading). Those items are most likely farmed by top players or bots. The game and its economy + drop chances evolved around a playerbase including bots. Banning all the bots, could have also a devastating effect. (maybe not). Then you have a game like Fifa with its Ultimate Team which has plenty of trading bots. They kinda regulate the market, make sure if you put up a player for auction, you usually get nearly the right price for it, as the bots want to flip it. I think banning bots is mostly about the Image of the company and the game. This is most likely why they always do it in waves with some blog post to show the people, they care. I think the bottom line is, those companies, want to make money. Bots make usually money, paying devs to ban them, cost money. Ofc they have to appear that they are actually doing something to not disappoint the players who are totally against it. I have myself some botting and bot development experience, but I would also work for a company who really cares to catch them as a consultant in the anti bot team. Good luck with your research, it is definitely an interesting and worth while topic.
Well I wish that if games creators become better at detecting and banning bot users very easyly that there's still gonna be some games in which they'll allow people to use bots becose while it's cheating it's still another kind of fun to try to make a bot work and do what we want in a game. Me I think that instead of trying to fight the bot users, companies should make realms appart in which bots are allowed for the people that prefere this way of playing. At this point I thinl that if AI keep continue to evolve in direction of a certain consciousness that it is also some kind of racism against them.
True! My crystal ball tells me that in 5-10 years every major gaming company will have to have those skills (or use external tool/expertise), simply because their customers will have serious expectations about it. Look at how the players are now more and more demanding, vocal about problems and prone to switch games, on top of the business models that are evolving. It's not like 20 years ago where it was just "cheating in a game, who cares?". We'll see if my crystal ball is right :p
@@romainbardon6959 Dont jump so far ahead. There will be a big shock around 2029-2030, that will affect all human biengs on Earth. Your logic is correct, but only if you know what will happen :P
wow crazy this is your only video and only 14 subs? was a good video.
as soon as they patch a fix, the hackers/botters can pull the source code and find change or just read a blizz blogpost update usually detailing what they changed, when and why. then they make work arounds. itll always be an issue
I agree with you about the data analysis and telemetry to detect bots, but what if the bots are trained to avoid detection? The bot would only need to act non-repetitive and not appear to be working optimally.
The problem would be if a pile of copies of a bot that does this, would show up more and more with it's own set of patterns, maybe on it's own it would avoid detection, but if 100K copies of it showed up, it may start the game the same way with same decisions and movements and clicks, that alone would be a red flag, it may attack the same way, that also would flag, so yes maybe alone it could avoid detection as it very well could be a unique player, but if a pile of players appear to start similar in batches then yeah it's bots, even if it has randomization, even in cryptography you can only do it for so long before the seeming randomness of it gets solved and becomes clear as day, it would have to be more like that 100K copies of the bot is an ai tied to a network, and every ban it sends it's play to the banned play, and updates all ai bots live at same time to alert, otherwise they all act the same rather than as their own. Ai agents are the same when released, but if they adapt then yes they will play like players more and more in time, however to counter it you just need 100k humans to submit 10 games each, then blizzards ai can see patterns, that way if 100K ai bot swarm is trying to not get banned, blizzard can have banned play as well and be where the bot ai's left off, so that way the company has a constant edge over this scenario.
Super travail ! Surtout sur le pattern des clicks. C'est vraiment pas bête, bon boulot 👍
the first step when doing a bot is to add a random delay to every action.. so the clicks/actions/delays would not have a dettectable pattern..
I’ve been thinking about this myself.. I mean it doesn’t take AI to randomize things if you’re good, but I’m trying to understand how these things are even detected. I once was making a bot for fun, and the moment the bot clicked the mouse on the screen the game banned me.
Very nice video with great explanation.
Thanks, much appreciated!
i know that its a bit old... but how did you make the "pathing" (how does the character move)
if you can't figure that out don't bother :P
Спасибо за интересное видео! Анализ поведения игрока и правда самое необходимое для борьбы с ботами в играх.
But all you suggest is client side behaviour detection, which could easily be countered as well. Also the amount of data they would need to collect and process and well you showed it on a really simple bot, but you know as well, that you could easily implement more chaotic patterns and fuzzy logic in decision making. I think the issue is, they are game developers and probably not skilled enough or interested in hunting bots and dont have the budget for this task. But you could in general argue, if a game is plagued by bots, its mostly because if the game design. boring, repetitive, time wasting and unrewarding game loops. In a world were free time us getting less and less and people are more stressed and just want to play the enjoyable part if the game. This is what creates the market for RMT and Bots. They will have to fight the reasons, not the symptoms.
Indeed, there are lots of ways to make such bots more "humans" and therefore more difficult to be detected as you said. It's just speculation on my side but I believe this would force to build less effective bots on the long term, so this might still be beneficial to use this approach. And in any case nothing prevents Blizzard to combine behavior detection and more classical approaches together.
I guess the design of the game does influence it, I've heard this point a few times. Some people find this type of game fun though, so should we change game design because of bots? Possibly, way but a bit unfortunate I would say. WoW is also plagued with bots, and you could potentially use bots in many of the mobile games asking you to do boring tasks on a daily basis. I suspect emerging Pay to Earn (crypto games) will suffer botting issues (and this could be much more lucrative than on D2R). I believe people can find reasons and interest in botting in many games that are both multiplayer and competitive, so this is a pretty wide range of game types. I haven't done a deep survey of areas where bots are more used, this could be an interesting analysis to do!
Thanks a lot for this constructive feedback!
@@romainbardon6959 Well it is also extremely dependent on the game and its economy. Also the payment model of the game. Like D2R now, its pay to play but the only extra revenue for paying the servers are sales of new licenses to the game. So bots can be pretty decent here, you detect them, let them run long enough to be profitable, then ban them so they will buy a new license. I remember back in old WoW times, when a bot got detected, you got your ban email with something like you are a bad boy, you violated our eula, please next time be smarter. Here is the link to buy a new link. The ban mail asked you to purchase a new game key. If you look at WoW, the bots also pay subscriptions, even if they purchase it via a Token, then still someone else, paid for the game time with real money to exchange it for gold. Still they make a profit. Then you have games like Path of Exile in a free to play model. There you could say bots are harmful, as they dont give the developers any gains, but PoE is also a very complex and bloated game where you could say many players would stop playing out of frustration if there would not be an option to buy items via RMT (real money trading). Those items are most likely farmed by top players or bots. The game and its economy + drop chances evolved around a playerbase including bots. Banning all the bots, could have also a devastating effect. (maybe not).
Then you have a game like Fifa with its Ultimate Team which has plenty of trading bots. They kinda regulate the market, make sure if you put up a player for auction, you usually get nearly the right price for it, as the bots want to flip it. I think banning bots is mostly about the Image of the company and the game. This is most likely why they always do it in waves with some blog post to show the people, they care.
I think the bottom line is, those companies, want to make money. Bots make usually money, paying devs to ban them, cost money. Ofc they have to appear that they are actually doing something to not disappoint the players who are totally against it. I have myself some botting and bot development experience, but I would also work for a company who really cares to catch them as a consultant in the anti bot team. Good luck with your research, it is definitely an interesting and worth while topic.
Well I wish that if games creators become better at detecting and banning bot users very easyly that there's still gonna be some games in which they'll allow people to use bots becose while it's cheating it's still another kind of fun to try to make a bot work and do what we want in a game. Me I think that instead of trying to fight the bot users, companies should make realms appart in which bots are allowed for the people that prefere this way of playing. At this point I thinl that if AI keep continue to evolve in direction of a certain consciousness that it is also some kind of racism against them.
9:11 I'm pretty sure they used this on d3 bots
what's the price ? :P
Haha, I would only accept money to solve the problem... yeah you can say I'm not a good business man :p
@@romainbardon6959 but you didnt solve the problem.... and noone will
Human can't sit near pc 20hours non stop, so Blizzard should introduce AI to fight AI.
True! My crystal ball tells me that in 5-10 years every major gaming company will have to have those skills (or use external tool/expertise), simply because their customers will have serious expectations about it. Look at how the players are now more and more demanding, vocal about problems and prone to switch games, on top of the business models that are evolving. It's not like 20 years ago where it was just "cheating in a game, who cares?". We'll see if my crystal ball is right :p
@@romainbardon6959 Dont jump so far ahead. There will be a big shock around 2029-2030, that will affect all human biengs on Earth. Your logic is correct, but only if you know what will happen :P
Then you just train your bot using GAN.
you are funny if you think bot detection does not involve data science