In future, Blizzard should partner with anti-bot company

This is a good example of old tech and new tech where the new tech is very accurate and very fast: Google’s music ID system.

The older systems used hashing and you had to have an almost exact match between the copyrighted content and the content you were checking. It was super slow, data-intensive, the whole 9 yards.

But along come the Google folks with finite state transducers and so on. They develop a much faster system that’s also accurate but can deal with “variations in …conditions”:

Our acoustic modeling approach can tolerate variations in acoustic conditions more naturally than hashing, which is based on exact or almost exact matching of fingerprints.”

“.…identification accuracy of 99.5% on a database of over 15,000 songs running faster than real time.

With AI-based bot detection, they also have “flex” in their detection to allow for anomalies and some differences in sequence - they’re analyzing the overall patterns with AI and so on.

And you’re right - the patterns change when botters change their bots to account for changes in a game. But what doesn’t change is that non-AI bots use simple delays, often static and not RNG-based between inputs. They also use the same repetitive “sequencing” of inputs for various activities. For example, in WoW a tell-tale sign for some bots is that they do pure keyboard turns because it’s easier to code, a really clumsy way to turn that doesn’t use mouse/camera at all. Most WoW players will mix in mouse/camera to turn. But bot detection doesn’t rely on any one thing.

Another thing to think about with these systems is that they rule you in and they rule you out pretty quickly so they won’t impact performance all the time- in other words, they only need a short interval of heavy input behavior to make a detection determination with high confidence. So once the system decides you’re a person, they might disable collecting input data for a while and then crank it up again later every so often. That kind of thing improves performance because you’re only detecting every so often.

On posting links, I can post links but the example I used that had SSN was a fictitious URL so I didn’t want people clicking on it or 1-click copy/pasting it because it’s not a real link. EX, I already posted this link in above comment:

Imgur

Anyway - with false positives, think of someone randomly playing on a piano and coming up with Beethoven or the classic case of randomly typing on a typewriter and creating Shakespeare. Those are clear exaggerations to serve a point but the point is that long sequences of “symbols”, inputs in the case of bots, have a very clear pattern or signature when you analyze them in depth, especially when all of that input is time-referenced. The odds of you playing Diablo and winding up with the same input sequence patterns and the same delays between inputs in milliseconds, such that all that is the same as a bot are statistically very low.

Oh so the system would know when the game changes and would keep up to date with any allowed addons huh. Where when implemented it is like having a sentient being that says hey I read the forums and the news so I know the game has changed and the addons are different.

Remember after implementation you don’t need to do anything with it at all. Fine when you are using for industries that are not like video games that make a lot of changes over time. Some even drastic changes, think D3 vanilla to where it was considered a good game.

Say that when the amount of players that are banned are more than what you think possible since you believe it is a near flawless system.

Something similar happened in a game called Rift Online. During the early months of its life there was a problem where player’s accounts were getting hacked. At first players thought that the ones getting hacked did some stupid things that caused it. But when it got way out of hand, then the players knew there was something wrong. Then the white hat hacker found that the game was generating and saving some sort of tokens that when used would let a hacker to have access to your account without your account information. Sure it was quickly fixed by the devs. But it shows that while you are correct if the problem is small. But if it is large then forget it. Players then would complain in droves.

Otherwise we would have Real ID on the older and maybe new forums. Where everyone would know our real name. That is what Blizz was thinking about years ago and the players said no way.

Again I want to make sure that the system you and the OP wants is one that can handle the changes that video games goes through this days. This is not like the Pac Man, Street Fighter, or any other arcade or console games that don’t change at all. Sure with those types of games you can get away with using such a system.

But with games that change a lot and at times drastically I say no. More so if it is not used by a single video game developer.

An automated set and forget system. Where there is no further input from the devs to the system after implementation. Where you not even letting it know that the game has changed. Then there could be problems where a ton of players would get banned.

Appeals are there to be used by a ton of players getting wrongfully banned. That is a great idea, then why doesn’t Blizz do it right now on purpose just so that process can be used. That is how I feel about that sort of thing. To me that process should be used only when you have to. If you can prevent players from being forced to use it by having a system that is not automated where it doesn’t get anymore input from the devs or any other source other than the so called perfect data from the game.

If you think that this system couldn’t cause a lot of false positives. Then I wonder how in the hell (no pun intended) did that gold dupe bug every happen. All it took was just a bit off in the program that caused it

Oh so an automatic system won’t need updates huh. And you don’t think that if it does that Blizz wouldn’t be charged for those updates. I wonder what world you are living in.

So you are saying that the system would only be for D4 and only implement it after D4 launches right. Because if you are wanting them to use it before that then it would be for D3 along with the rest of the games that exist in Blizz’s line up when it is implemented.

Oh so the automated system will know because it truly is a sentient being that even reads the forums and the news on Blizz’s website huh. Songs really don’t change like you think. Sure there are some songs that have different versions but still it is the same song and fingerprinting My Way by Frank Sinatra or Elvis Presley will have the same fingerprint when you on looking at only one of the two singers.

Silly me all wise, knowing and powerful Guru of antibot companies. It would be the absolutely perfect system. Where nothing could ever go wrong. Where the only false positive would be one every 10,000 years right.

Where I could say that any group of players that complain about getting wrongfully banned are lying. Where if Blizz itself has issued an apology for wrongfully banning tens of thousands of players. I have to look at both as ones that are blowing smoke up my nose. Even if it happened to me it would be just in my head right. Or better yet if it happens to you it would be just in your head.

False positive rates, depending on the solution and inputs, isn’t really “subjective” in terms of saying “false positive rates will be a problem”. It’s just math and it’s easy to show why false positives would not tend to be a problem with these systems, regardless of game changes in terms of adding content. We will focus exclusively on human reaction times.

Very simple example would be if you wanted to make all Diablo human players appear as bots and intentionally have similar input sequences. Everyone uses a 120 bpm metronome and is told to only press keys on the beat sound which happens once every half-second. All the metronomes are sync’d universally so they’re all in time. Everyone has the same key bindings, starts game at same point, same level etc. Everyone is doing their best to only click every half-second to that metronome beat. Problem is that human reaction times are, on average, per research, about 1-250ms off of their intended target time.

Again with an exaggerated example to make a point, assuming each input had an evenly randomized distrib of being 1-250ms off: even with just 100 inputs you would be looking at ~250100 possible sequences - more than the estimated number of atoms in the observable universe. The odds of 2 people doing 100 inputs with the same identical intervals in ms between key presses is astronomically low. But a bot, minus “noise”, with hardcoded delays would be roughly the same every time.

So my point was more that false positives will be a function of player numbers rather than an intrinsic issue with the detection. For example, if you have a billion players and detection error rate is 1-in-a-trillion then sure, you get an avg single false positive for every 1,000 detection iterations over those players. But we don’t have a billion players and the detection error rates for longer input sequences would be even lower for those theoretical examples.

I was getting at the “power of sequences” and pattern matching in terms of their statistical significance, i.e. false positives using long sequences of inputs occur at a very low rate with certain solutions. As explained, you could try your best to make the entire population of players act like bots but they would still all be easily discernible as humans because of the human reaction time variance. In contrast, the basic bots used w/ Diablo games use hardcoded timings and use repetitive keys/clicks - they have very little variance and are easily detected. And even with random timings and random coords for movement, bots would still be easily detected based on behavior pattern analysis.

2 Likes

Machine learning doesn’t stop learning. That’s kind of the point of it.

This is just fear mongering. Any well made anti-bot system will have checks and balances in place.

If they implement a badly designed anti-bot system it will have problems, yes. I am not suggesting they do that.

Your entire argument here relies on made up fantasy scenarios where the software is designed to be as terrible as possible and you keep hammering on this idea that the system would not be maintained once implemented which I never actually suggested(I even literally said anti-bot software has to be kept up to date).

If you only want to assume worst case scenarios then again literally every system should be unacceptable to you because we should assume it was made by a first year CS intern and every single line of code is bugged and doesn’t work properly.

I could make Diablo 4 itself look like a bad idea by using your logic if I assume the absolute worst in every single situation and assume Blizzard is a bunch of bumbling idiots who can’t code.

1 Like

I know you think that your idea is the best thing since sliced bread. I just look at seven of the best ones. None of them are used for detecting bots in video games.

That is where the problem is, if you can show me some video game developers that actually use them without incident then you have a point and I might be more on board with it. But I don’t think that you will find such a thing. Because the bot detectors I saw were for apps, websites, preventing DDoS attacks, etc…

Unless there are anti bot companies that make systems for video games. Then you would have to show a list of video game companies that actually use them.

Do you remember you earlier said that the bot detection system uses machine learning. You don’t need machine learning when you are looking for inputs that would be impossible for a human being to accomplish. Then you don’t need machine learning tech or software at all. Don’t need to worry about behavior or anything else for that matter.

I wasn’t basing my concerns on basing the system on input times that are impossible for a human being to do. I was focusing on content.

I will give an example, Path of Exile when ritual was the challenge league. Now let’s say that GGG partnered with such a company during that challenge league. Now it gets the behavior pattern from the players. Then when Ultimatium launches the players are doing something entirely different. That different behavior pattern could be seen as a bot when compared to the older pattern.

That is unless you are saying that the devs tell the bot detection system that they have a new league with all of the changes. Along with the changes to loot filters.

That is another area that could cause false positives. Loot filters, since they only show what you want to see and not show what you don’t want to see. Those are third party programs that GGG allows because it has coded the game where players can make their own loot filter. Although most use Filterblade.

Again I say that the system must be informed by the developers in some way. Even if it is only by having the system be able to tell when new content is added to the game. Or when balance changes are made.

Heck there could’ve been a lot of false positives when the Marauder’s set changed from players being able to hide behind their sentries. To where they have to be actively casting skills to get the sentries to cast skills.

Using a system that is not designed to handle video games, which the bot detection systems that I looked at. None of the seven that I looked at works for video game developers. They work for apps, websites, and DDoS prevention.

From the OP:

Also, because many of these systems have input/behavior profiles based on various bots in the ecommerce space, it’s likely Blizz would need to “train” the system and customize it. Some systems may also be better suited to Blizzard in terms of software stack compatibility.

From an earlier comment I made:

"I looked at the D2 Kolbot repo and it’s spaghetti code full of hardcodes - no RNG in behavior or any sophistication. So it’s obvious, at least with that bot, that it would be instantly detected. Another thing is that the repo itself warns people that D2BS is detectable lol.

So that would just lean into the theory that Blizz has more of an internal workflow problem or just doesn’t care because basic game bots like that are so easy to detect. [Thus implying a new system is NOT NEEDED.]"

I already commented that they would likely need customization and so on. I also already commented that a new system would be overkill based on the current Diablo bots being so basic. Additionally in OP:

Just a closing couple of points, I’ve read some comments here and there suggesting Blizzard already detects the vast majority of bots and simply isn’t efficient with manpower or their process is just too slow resulting in botters running for long periods of time. If that is the case, needless to say partnering with a leader in the field isn’t needed.

So what you’ve written has been covered in the past, just making sure you understand I looked at the issue as a devil’s advocate and went out of my way to paint both sides at various points.

But here’s the thing - new Diablo 4 bots with so much RMT potential and a giant new playerbase may be quite a bit more sophisticated. Much like WoW bots tend to be more sophisticated depending on the bot. I’d rather see Blizzard just use a solution that deals with with all bots in a uniform way - thus being more future-proof with AI-based bot detection.

Also - we know their current workflow or bot detection just doesn’t work well. Again: the current system doesn’t work well. Thus, this is just a different solution to change the status quo of bot-infested games.

Not sure why people would take issue with a potential solution that could benefit everyone by auto-banning bots.

It’ll be informed just not in the way you I’m getting the sense you think it works.

With machine learning you have a learning phase for it where you show it a bunch of gameplay or whatever data you’re trying to get it to learn. Since you’re in control of the gameplay it’s looking at, you can specifically tell it “this subset is what botting looks like” and “this subset is what legitimate gameplay looks like”.

It then takes all the data and analyzes it in a way that a Human could do but would take us a very, very long time to do to find the common pattern between just the botting gameplay.

You keep this up until the software has learned enough to have an extremely high degree of accuracy to tell bots from non-bots, but contrary to what you’ve been suggesting it doesn’t exactly stop there. If you encounter false positives you feed that information back into the software to tell it that it was wrong.

So your software is always getting more accurate, and if a new bot program shows up you can feed that data back into the system for it to analyze.

That’s why I originally said this wasn’t going to immediately and forever solve Blizzard’s botting problem. This will however cut down drastically on just how much manpower they need to commit to it in order to stay current.

Without machine learning they need to analyze the data from new bot programs and manually put that into the software. This would automate that process so they really just need to show it the gameplay to generate the data.

It also means significantly reduced time between suspecting an account is botting and being confident in banning the account for botting. Computers are simply faster than Humans at that sort of thing.

What are they gonna take what they are using to protect their servers from malicious bots and DDoS’ers to change it where it would be able to detect bots in their games. Great idea lets all use screwdrivers as hammers.

You just don’t want to grasp that the ones that I have seen are ones that are not designed to handle botting in a game. You have to have a special system for that purpose and that purpose alone. Then you would be doing just fine. This is what I would be looking for in such a system.

Sure RMT’ers are gonna make a ton of money off a trading system that isn’t completely free and unlimited. Where some items are traded without limit. Others are tradable only once then they are BoA. The rest of the items are BoA. Tell me how much money a RMT’er will make off of running a Dollar General type of business. I don’t think it will be much at all.

Trusting in a system that would do it for you automatically isn’t a wise move. Because that system has to be programmed properly. One wrong line in the programming code could cause many players to get banned for botting that are not botting, think of gold dupe bug here.

It is the behavior pattern that the system is looking at. If it is not constantly updated by the developers in some way. Then as soon as the new patch hits either live or PTR you could have a ton of players getting banned for botting when they weren’t botting at all.

Great idea find a way to cut the amount of people working for Blizz. You do realize that they need to feed themselves and their families like everyone else.

Here goes another problem that a machine cannot do. That is accurately detect if the original account holder is doing the botting or did that account get hacked and the hacker is doing the botting. People can do that better than a machine.

What do you think machine learning means?

That it doesn’t learn anything unless the developers code it?

Yeah clearly we should get rid of all machinery because somebody might have lost a job over it.

Nevermind that those people have transferrable skills to simply be moved to another position within Blizzard.

Wont somebody think of the strawman argument!?

Blizzard doesn’t care.

Even right now if your account is hacked and used for botting, your account will be banned. You can appeal the ban after you secure your account.

So again we have you criticizing the system for something that is already the case.

I guess the only winning move is for Blizzard to just let the bots run rampant and never ban people for it.

See, I can take your words to absurd extremes to!

What is it a sentient being that has some psychic power to learn that it is a new patch. Or it reads the forums and patch notes on its own. Look what I am saying is the way to keep that type of system up to date is to have it coded that when changes re made to the game the system is updated with those changes automatically.

What position would they move to?

Yes they do care enough to have such a system in place. Where the ban isn’t permanent and the account is terminated.

I was just saying that the system cannot figure out on its own that a account that is caught botting is a hacked account that the hacker was using to bot on. Only people can make that determination. You make it like that automated system is some sentient being (something that can think on its own). Figuring out the differences between the old content and new content.

Where I am saying that it is better if you have such a system to have it where as soon as the changes are made the anti bot system is told. All through how the system is set up.

It’s odd that you would say that when it’s literally the purpose of many enterprise software systems to automate manual business workflows.

Also:

  • These systems already work for large companies - they are fully automated. The automation is proven. The false positives are low.
  • The same system model used by these systems works in Blizzard’s case: input capture and SaaS service bot detection or in-house-server bot detection.
  • Game inputs are far more numerous and revealing than short-lived ecomm interactions thus game-bot detection is even more accurate as the input sequences are much longer.
  • We already know a manual system doesn’t work well. We’ve known that for years. It takes too long, they ban infrequently in waves. Players as customers are obviously not happy with the current bot solution.
  • A new automated solution is needed. With D4 and significant bot activity their manual bot workflow will be even more overwhelmed. Automation is the key, just as it is with a million other workflows that have been automated across the industry.
  • Using AI-based systems for input and behavior analysis of bot inputs is the same paradigm regardless of bot coding - it’s why these systems accurately detect thousands of different bots written by different coders in different countries and so on. Saying that this paradigm wouldn’t apply to game-bots is just not true.
  • Using pattern analysis and detecting the lack of input variance of today’s game-bots, is incredibly easy for these systems - they are even overkill as we’ve discussed. It’s obvious they would be up to the task, inputs are inputs.
  • A single AI-based system doing autobans would uniformly handle all of Blizzard’s games. Game bots across all of these different platforms, mobile, console, PC would all be handled in a consistent and centralized way. Think about how wonky and clumsy Warden has been on PC in the past. By analyzing inputs with AI, it’s a cleaner system that’s far more future-proof and works well with the various hardware platforms and games.

:joy: :+1:

I am more skeptical about the efficiency of machine learning than most here it seems, but it would not exactly be hard for it to see that the game had a new patch, for one the game version changes… and if not, the game code itself as changed which would be easy to detect.

Who cares? It doesn’t matter.
(I mean, it does matter in a broad sense, as the effects of automation is one of the two largest threats to human civilization over the next century. It just doesn’t matter for this context)

@Solo

The big difference between having an automated system that is designed to handle the job. And one you are trying to make it do something that it isn’t designed to do. Having a third party doing something for you is what I was commenting on. A party that has programs that are used for keeping servers protected from malicious bots, DDoS attacks along with other things that have nothing to do with bots in video games.

Great idea to use a screwdriver as a hammer. I guess according to you screwdrivers make the best hammers right.

Do you really believe that there is very little traffic on the internet. Where the bots that are on the internet are small compared to in video games. You might be surprised to find out how wrong you are on that subject.

Look the reason why Blizz has been slow to ban botters in this game is due to the priority of the workforce that they do have. There is only one team of workers that handle botting for all of Blizz’s titles. With their priority on the titles that bring in the most money. Now do you see the problem.

The solution is simple instead of firing the team expand the team where you would have more than one team to handle botting for all of their games. Where each game has its own team that is dedicated to handle botting for that game specifically.

A new automated system is not needed if Blizz has enough money to hire a team of workers that would handle botting for D4 and only D4.

The automation only works when it is designed to work that way. It isn’t gonna be as easy as you think. Where you are taking a piece of software that isn’t designed to handle video game botting and making it do that. Then putting your full trust in the system. All it takes is just a few mistakes in the code to cause a ton of players to get banned for botting that wasn’t botting at all.

If you think that the software is so powerful that it a Swiss army knife with an infinite amount of possible uses you might be sadly mistaken.

Again you won’t show me proof that some of the big dogs in the gaming industry are using this system. Then you would have a point. But that is not what any would be using such a system for. They are using it to protect their servers not to find bots in video games.

@Shadout, what is a person that is specialized into handling botting for video games gonna go to in Blizz. A person that has no other skills. If that happens then this person has to be trained by Blizz. What is next will that person be put on a design team learning about game design as he goes along. I don’t think that is the best way to learn about video game design.

He will be laid off. Which is fine. You are trying to make a problem out of something that is not.

1 Like

@Shadout I guess that it would be okay as long as it is not your job that is in jeopardy.

How long would it take two 5-person teams to manually deal with 1 million bots (20% of D3’s first week of sales)? 100,000 bot cases per person, we’ll generously presume they’re super-fast with entering tickets in their ticketing system, and verifying things you deem critical with the account and ban, and so on - 5 minutes per bot case.

5 mins per bot, 12 per hour. 96 bot cases in an 8 hour day per person. 10 people gets through total of ~1,000 bot cases in a day at a very generous speedy manual clip.

It would take them 1,000 days at this 5m manual process clip. Several years.

A highly concurrent automated system could do it in minutes given the scale and response time which as shown before is as low as milliseconds. So we’re just rounding way up to minutes to make sure we’re grabbing nice long input sequences of playing.

Manual system = Several years to manually slog through it with generous per-case speed.
Automated system = done in minutes with highly concurrent automated system.

This is why we automate things. Manual processes don’t work especially when we need to make botting cost-prohibitive with frequent bans.

So we’ve laid to rest your argument to just increase the size of a team - a manual team is too slow.

Next with the “client has to be video-game maker” argument. That’s a logical fallacy in a general sense as I’ve already noted. Video-game makers use all kinds of software solutions that other enterprises use - from using the exact same persistent store / database, to event-driven behavior trees, to cutscene video compression / playback, to Autodesk products like Maya, to Atlassian stack with Jira/confluence for collab, to cloud computing with AWS / Azure, to various agile dev tools, build systems and so on.

Would it make sense to say that Blizzard couldn’t adapt and use a certain high-performance highly-concurrent persistent store just because a “video-game maker” isn’t currently using it? Of course not.

Just even look at their work with DeepMind and their AI project collabs - they wound up creating in-game RL Agents that could compete with the best players. They created AIs that were about as good as the best human players with some of their games like Starcraft. Before that collab happened, was it logical to say that this kind of AI could not be applied to Blizzard? Of course not. Tech is tech and should be analyzed and evaluated in a formal way, that’s all there is to it.

The best of these systems, and I’ve looked at docs for several of them, analyze bot input sequences and use AI-based machine learning to do pattern analysis on that input. That paradigm applies directly to Blizzard’s bot case: collecting input on client and analyzing on server-side. This overall paradigm would clearly work with customization as noted several times and included in OP.

Edit - if you’re worried about job displacement, anti-cheat folks per job openings I looked at for both Valve and Blizz have highly sought-after skillsets. They’d be fine.

Also if it was my job.
Keeping people around in “fake jobs”, just to pretend they are doing something useful, is no way to treat anyone. You are not doing them a favor.

As Solo said, the people affected in this little scenario would likely be more than capable to find new jobs, but if not, that is the purpose of offering training for new and more valuable skills.

1 Like

Clearly a sentient being is the only thing capable of taking data and analyzing it.

People working on anti-bot software would have to be software developers.

Now I wonder what kind of positions they might have available at a video game development company with half a dozen game development teams.

It’s permanent in the sense that it will never expire unless you appeal and they over turn it, and the account is not immediately terminated.

Which is not a system that needs to go away just because you’re using machine learning.

and you make it like any system is bad and Blizzard should just let the bots run rampant.

Or maybe neither of us are saying these things and you should stop taking things to idiotic extremes and that was my point by saying that all along.

@Solo I see I have to get technical with you right. Okay then here goes. What if they have Warden programmed to do more than just flag potential botting accounts. Where it also bans them for an undetermined amount of time. Meaning until either Blizz calls the account owner or the account owner calls them.

Now if they have laid off a lot of their customer service and it can’t handle all of the calls then there will be a problem handling the cases. Again it is manpower based on those two teams. Both customer service and the team that handles potential botters in games.

If Blizz could handle this setup then what they could do is wait till near the end of the season around one to two weeks before the season ends like they normally do when they do ban waves. Then do their ban wave along with cleaning up the leader boards when they have obviously found botters.

What am I wrong for wanting a special system that is designed to handle bots in video games? One that isn’t designed for any other task. Or are you saying that your idea is so perfect it isn’t needed because those systems that protect servers, apps, and do a decent job at preventing DDoS attacks is the perfect Swiss army knife that can handle all of that and handle botting in 11 of Blizz’s titles. There has to be limits to what is capable of any system.

I think that would be putting to much of a workload on a system that isn’t designed for it. Wait you will tell me that Blizz could customize what they do have to protect their servers from those things along with handling all of their games (11 games). I think that is too much for any one program to handle. Doing that could weaken the protection to the servers and cause more frequent DDoS attacks. Or allow hackers to break in and steal account info.

Now wait you will tell me that they will just get a second program like the one that protects their servers and customize it. Remember, if Blizz is off on customizing it that could cause players to get ban that weren’t botting, remember the gold dupe bug when the RMAH existed. If not look it up it is something interesting to learn about.

@Shadout then I guess you would say that those jobs were always fake jobs from the time they were created. Long before any automation of any kind existed. Which would include your job as well if it could be automated. Which means you whole life has been doing a fake job that is useless. That is depressing to think about.

@Cyonan, I see you miss the point again. I am saying that the way the system is built it has to be able to detect the changes in the game. If it is not built to do that and gets no input from the devs then there is a likelihood that a lot of false positives would happen. Otherwise you are saying that I can take a firewall program and have it work as an antivirus program without programming to do so.

You like Solo make me feel as if by asking for a system that is similar to the one that Solo wants that is used to protect servers from malicious bots, protect apps, and prevent DDoS attacks. One that is solely designed to handle botting is a wrong thing to ask. No instead you would probably want Blizz to either get another program that is similar to the one they use to protect their servers and customize it to handle botting in games.

But the danger is if they make any mistakes they could cause a lot of players to be banned for botting when they didn’t bot. Plus some bugs in programming cannot be fixed. That is what I don’t want to happen.

I want a system that would be designed by the ones that make the systems that Solo seems to think are so perfect that they can handle trillions of things without even breaking a sweat. To handle botting in games as well.

They are fake jobs from the moment where it would make more sense to automate them, but you choose to keep them as some sort of job support initiative.
And yeah, automation is going to replace most of our jobs at some point.
It will be kinda depressing. There is a reason it is one of the biggest threats we face in the future. How do you keep a society together where a major part of the population can’t get any work. Solve that question and you might get a few Nobel prizes in various fields. But I think that is slightly beyond the scope of this thread…
Still doesn’t change the fact that the solution is not to behave like a ostrich and keep going as if nothing has changed.
Which means, to bring the point back to earth/the topic; people losing their job is not a meaningful argument against automating more of Blizzards bot detection system (which probably already is heavily automated).

:woman_facepalming: