How I was banned / The State of the Staff Team

To begin, I want to stress this as much as possible. Detecting ghost clients is NOT easy or simple. Anyone who tells you otherwise has never been involved in banning someone for a ghost client ever. There have been hundreds of thousands of dollars invested into ghost client hacks to make the job of detecting them as difficult as possible. By definition, if a client is easy or simple to detect, it is not a ghost client.

As well as this, there is no "standard" for detecting ghost clients. There have been decades of research and studies into stuff like advertisement, game design, staff management and other things Brawl does. There has been no research or studies into detecting ghost client hacks. There is no known "right" way to detect ghost client hacks. Yes, that includes the way you detect ghost client hacks, and it includes the way I detect ghost client hacks.

Unfortunately, we are winging it. I wish we weren't, but our only other option is to ignore ghost client hacks completely, which causes just as much backlash, as seen by the now-deleted "That's Crazy" thread from Dragoarg.

I'm going to go into this in more detail later in the thread.

Context

Spoiler: Context

This screenshot has been completely taken out of context. Here is everything:

Spoiler: Very Big Picture

I made a decision when promised, and got back to him when promised.



I sent him a list of some clips the day after, on the 4th of May.



I continued to answer some questions after this about his appeal, as you can see in the main screenshot in the spoiler above.

Why I Stopped Responding

Spoiler: Why I Stopped Responding

I stopped responding to him when he asked for updates on his appeal, despite there being no updates.

Unfortunately, staff are supposed to limit communication with permbanned players as much as possible. This is due to very serious threats that some permbanned players have made in the past. This is not written in protocol, but I have been directly told the following from a higher-up: "You have to stop talking to banned players" (context clearly implies permbanned players, not banned players in general).

As such, I try to limit speaking with permbanned players entirely except when doing so is necessary to resolve their appeal. I know that it is very unlikely that Cow would make any threats against me or the staff team, but I would have said the exact same thing about one of the people made the very serious threat that has caused this in the first place. Because of this, I will not be speaking with Cow directly on this thread.

I will answer questions from permbanned players once. If I have nothing to add, I won't be responding.

Why It Took So Long

Spoiler: Why It Took So Long

Whenever a staff member posts a video of someone potentially cheating, I spend about 30 minutes going over the clip to make an initial response. Over the next few days, I then analyse it in depth, to try and prove that the player is hacking. If I think I have done so, I then analyse the video in depth again, to try and prove that the player is legit.

This process takes roughly 3 hours per minute of video. If players are using a badly configured ghost client, it may take 30 minutes per minute of video. If players are using a very well-configured ghost client, it may take 10 hours per minute of video.

I do this before the ban is made. However, since I wasn't planning on being heavily involved in Cow's ban, I didn't. When Miskey resigned, I had to go through this process after the ban.

This is combined with the other work I do for Brawl, and the work I do in real life. Due to the coronavirus pandemic, I have had to put a large amount of overtime in. My team is made up of mostly high-risk people, and it is on a project that cannot be worked on from home yet. In order to get this project to the state where it can be worked from home, I have had to put several hours of overtime in a day to do this.

Not to mention, I was on a training course with an exam. Anyone who's taken a BCS exam before will know how tough they can be. That week, I had to focus on the training course and revision. When I was done with that revision, the last thing I wanted to do was analyse Minecraft videos.

In short: 2 weeks to handle an appeal is not the norm. Due to an unfortunate combination of circumstances, it did take this long for this appeal.

Ghost Client Detection

Spoiler: Ghost Client Detection

This is absolutely one of the most controversial things on Brawl, and it is one of my biggest priorities to fix. Overall, a lack of trust from the community is Brawl's biggest issue, and the controversy surrounding it is worsening this issue. Trust me - I would not have spent 2 hours writing this response if I didn't care about fixing this issue.

There are several reasons why ghost client detection is so difficult to do. Here they are:

Every Ghost Client is Different

People seem to think that "it is impossible to do x with aim assist", as if aim assist is a tangible thing that exists. Aim assist is the name given to hacks that assist aim in ways that is difficult to detect. It is possible to do everything with aim assist. Everything we have as evidence against Cow is possible to do with aim assist.

Ghost clients are not based on some sort of blueprint. People seem to think that every ghost client works the same, as if the only difference between different ghost clients is the name. This is not, at all, the case. Every single ghost client works differently.

Not to mention, ghost clients can be configured. It's not as simple as fly, where you enable it and double tap space to start flying. There are all sorts of settings that the user can configure to make it more blatant or less blatant. Aim speed, activation range, aim acceleration and deceleration and target prioritisation can usually be changed by the user.

There will never be a "one-size-fits-all" approach to ghost client detection.


They're Designed to be Very Difficult

Those behind making ghost clients aren't a group of friends messing about with some code in their free time. Popular ghost client makers are functional businesses, making far more money than Brawl is.

Brawl has a voluntary staff team. The only people who get paid are the devs. Our staff team have lives - often, very busy lives. Our staff team also have roles on Brawl outside ghost client detection. We cannot improve our processes for ghost client detection as if it were our job.

On the other hand, the devs making ghost client hacks can absolutely make ghost client detection more difficult as if it were their job. It is their job. They're getting paid to spend dozens of hours a week working on better ways to make it harder to detect their ghost clients.

The only way any Minecraft server can counter this is if we start paying people to spend just as much time going after them. For reference, Hypixel pays $70-100k a year for their anticheat team. We do not have the resources to fight against ghost client hackers - and even Hypixel, with their millions in profit a year, are struggling.


There's No Known Right Way

With all sorts of things Brawl does, you can look at studies, research and international standards on how to do it right. Staff management? Every company does this. There's a huge industry based around giving advice to companies about how to manage their staff properly. Advertisement? Game design? Public relations? Leadership? Web design? Same story.

This is not the case with ghost client detection. Minecraft servers are left in the dark on how to detect ghost client hacks. There's very little collaboration between Minecraft servers because most just don't bother with ghost client detection at all.

I'm currently redesigning Brawl News. There are companies dedicated to giving advice on how to make a good newsletter for your company, and I've utilized several of these companies to make sure the News is as best as it possibly can be. I know with fairly good confidence that when this redesign is finally done, it will be done right.

Again, this is not the case with ghost client detection. I can't hop on Google and find "47 ways to detect ghost client hacks efficiently in YOUR Minecraft server". I can't take a look at how Hypixel or other servers do it because they keep their methods secret (and are in the same boat as us, albeit with more resources to dedicate to it).

People with marketing degrees have made blog articles and studies and websites that directly give me advice on how to make a good newsletter. There isn't anyone with hack detection degrees giving me advice on how to detect ghost client hacks properly.


We Have No Idea How Good or Bad We're Doing

When the Party update is released, I will know exactly how good or bad it does. We track playercounts, and we can measure player engagement before and after the update. If it does badly, we can conduct a survey to find out why players don't like the update. We can get qualitative and quantitative data on how well it's doing.

We can't do that with ghost client detection. At all.

We have no idea for sure if someone's ghost client ban is correct. We can't get a percentage of ghost client hackers that get away with hacking. We can't get a percentage of banned players who weren't actually cheating. There is absolutely no way to know for sure.

We could be doing incredibly well at ghost client detection with a 100% success rate. We could be doing absolutely terribly at ghost client detection with a 0% success rate. We have no idea, and neither does anyone in the community.


Even If We Did it Right, Nobody Would Believe Us

Let's say we come up with a new method for detecting aim assist. This hypothetical method is 100% accurate - it will never cause anyone to be false banned, and nobody can get away with using aim assist.

How would you, personally, know if our method works? Even if we told you what this method was, how would you verify that it works? Would you even be able to make an informed decision about whether or not this method is more accurate than what we're doing now?

Cow would be denying that he hacks right now regardless of if he was hacking or not. We do get people who admit to hacking, and not just to get unbanned - we have some people who admit to hacking to their friends only, and their friends pass this information on to us. Nobody ever admits to hacking early on after their ban. Some people who admit to their friends will continue denying to us that they hack years after the ban.

I also don't at all blame Cow's friends for believing him. It's my word against his - and who are you going to believe: Cow, someone who you've been friends with for years, or me, someone who you may have never even spoken to before? Many people make up their minds before they even see the evidence.

We could be absolutely perfect at ghost client detection, and in this hypothetical world, we would still get just as much backlash for our ghost client bans. The only way to avoid that would be by building up a much better level of trust between the staff team and the community - and if we did that, there would be no backlash for our ghost client bans even if they were false.

What I'm Doing to Address All This

Spoiler: What I'm Doing to Address All This

First of all, I have asked xxMineSheepxx, Ritzy and yairster to look into how I handled Cow's appeal. xxMineSheepxx is the staff manager, Ritzy is the permban manager (who works frequently with appeals) and yairster is the substitute ban manager (who works closely with me to manage bans as a whole). In other words - I have reported myself to everyone so that my handling of Cow's appeal can be properly investigated.

I have also asked all Mods to do the same. I haven't asked anyone directly - I have sent a message in Mod chat, tagging all Mods, to give feedback. I have encouraged them to give this feedback to xxMineSheepxx if they would prefer to not send me the feedback directly.

When I'm done with the News, my next priority will be to write up a proposal based on all this feedback to change how ghost client detection and appeal handling is done. I will write up a protocol on how ban managers should handle these situations, so that we have robust instructions that will hopefully be more effective than what we're doing now. A proper framework like this will also be easier to criticise.

If I do not get a suitable amount of feedback, or if the feedback I do get is implemented and nothing improves, then I will make ghost client detection a far more public process to make it easier for community members to give specific feedback on how we can improve.

EDIT: I plan on responding directly to individual replies over the coming weeks. Unfortunately, I don't have the time to do it today.

 

While you get **** from a lot of people I just want to say I appreciate you and the time you invest(ed).