Radical Aimbot Solution

[beerbitch]

Ok I'm going to try the master ban list idea.

1) Have a master ban list (master admin.dat with bans in it)
2) Have a script that can fetch the list and synchronize your admin.dat file to include these global bans.
3) Have a forum where registered server admins can go and add new bans to this list.

I'll host the forum in our clan server for now, url pending.

If I can make the script work so it won't bork the admin.dat file then I bet I can get some admins to test with me.

[Nux]

So how does this system determine who is global-ban worthy and who isn't?

Is it fair for one server-admins opinion to mean a ban from all servers?

If proof needs to be given to the other server admins that the guy was ban-worthy before a global-ban is made, surely the current system works just as well. The admins can only be sure that they are ban-worthy if they see it themselves.

[beerbitch]

So how does this system determine who is global-ban worthy and who isn't?

Is it fair for one server-admins opinion to mean a ban from all servers?

If proof needs to be given to the other server admins that the guy was ban-worthy before a global-ban is made, surely the current system works just as well. The admins can only be sure that they are ban-worthy if they see it themselves.

A discussion forum for this is not that new except now we have a list that we can all use. Now when a hacker is caught, he is less likely to be able to play on other servers.

I can implement a voting system for the admins to add a ban to the list or something, so it won't be just one admin. I can make a web app that can accept a ban request, then the admins would discuss it, then vote / approve it.

[Overdose]

Ok I'm going to try the master ban list idea.
--snip--
3) Have a forum where registered server admins can go and add new bans to this list.
--snip--

Maybe add a separate public forum, where people who were banned can argue their case. Thats all I wanted to say

[beerbitch]

Ok I'm going to try the master ban list idea.
--snip--
3) Have a forum where registered server admins can go and add new bans to this list.
--snip--

Maybe add a separate public forum, where people who were banned can argue their case. Thats all I wanted to say

Thats a good idea.

Current feature list:

1) Global admin.dat ban list
2) Private forum for admins to vote on submitted bans (add/remove)
3) Public forum for ban rebuttal or public submission

[BeerBastard]

Also I was thinking a warning system for admins. If someone enters the server and their subnet matchs a subnet of a ban on the local server bans or the globar bans. It could pm all the admins about it. That way you don't have to subnet ban but you still get a heads up when someone who might be ban dodging comes in and trys to play. You can watch him.


Also this system can work for deconners too.
This would be more of a tjw feature.

[David]

I'm all for the web of trust idea.  when my server downloads your list, it can process it however I want.  So say it takes 5 points to get someone banned, all bans are for 2 weeks, and ignore bans under 1 day.  Then server that has banned someone earns them a point.  I could then weight servers differently, so some I trust, and some that I think are trigger-happy.

Do you have a dev forum up yet?  I am keen to contribute.

[Eeeew Spiders]

Something i posted a while ago.

The main idea was that this forums user database could be used as the master server. Authentification is easily done by adding some php scripts on this forums server and using simple http requests from either the tremulous client and tremulous servers.
Features:
- It proposes forum user group concepts to be used for player management on tremulous server
- Checking up on IP and management of players can be done at one place, on this site by this site's admins, or alternatively a special user group can be created here for that job
- The ideas go a bit further than just sharing ban lists, but it includes sharing ban lists.
- Names as registered on this forum are globally protected on servers that use this forum as their master server
- How the servers make use of the authentication system and banning system is highly flexible and can be easily customized

[Odin]

This should be a disableable feature. If a server does not want to participate in the global banlist, then allow them to disable it.

[beerbitch]

This should be a disableable feature. If a server does not want to participate in the global banlist, then allow them to disable it.

For now I'm going with the script approach so no trem patching would be needed. You can simply choose to run our script to get a list or not.

[KusaKari]

Here is my bias and suposabely simple solution to aimbotting.  The first note on aimbots is their n_ commands  like n_aim.  I am wondering if a simple add to the server pure check coding could run that command on the users client.  If it says unkown cmd.  They are obviously not botting.  If it says n_aim = 2 default = 2   or w/e it can tell them they are inpure.  That would stop all of the current funlily aimbotters.   I don't imagine this would be to hard to implament and I see that most server owners would gladly through the patch in their qvm.  I don't see it requiring any mods to the client either.  This is just a simple thought though.  

I also wonder is there anyway to see the hud of the play from the server.  When speccing someone even if they have a modified hud it shows in ur hud.  Can the server see their hud settings.  If so funlily says in its hud funlily:Backport GUID     right below ammo ect.  

I also saw from MW:ne0 something today that funlily makes a default bind and mod in its source that says the name of the aimbot and where to get it.  ne0 accidently hit it in a public game and hammer was noted to the problem.  Maybe the server could check if the client has a bind that outputs anything with funlily,  then the client could be informed they are impure.  I don't know how the bind works though because it doesn't pull up in bindlist.   Nor could I figure out what key causes it to happen.  On most servers it auto says it when u enter the game.  

These are just some simple suggestions that would weed out the majority.  
I don't know the plausability of them so please don't call me a stupid troll lol.

I still think the global ban list is a great idea but I also think this would help it greatly.  Its sad that the aimbotters leave the responsibility of cleaning up after them to the server operators and devs.  I hope we can get this cleaned up fast no thanks to that stupid asshole funlily.

[David]

Sadly none of that info is available to the server.  If it were it would lag the server every time someone joined, and aimbots would just lie.

[KusaKari]

Sadly none of that info is available to the server.  If it were it would lag the server every time someone joined, and aimbots would just lie.

I thought it could check for the commands, can't the server at least take notice if someone runs an n_ command so it can kick them immediately and warn the admins?  

I don't see it lagging the whole server to check one command.  I figure the HUD stuff is radical checking and that could cause lag, but I see no reason that the server can't just run a command to see the results or how that would cause lag.  It shouldn't cause any more lag than the current pure server check does.   From my understanding the server has a lot of power over the users client and the client is designed as a base to allow everything the server does.  There is even a large glitch that can allow the server to run basicly whatever it wants from the clients machine if its windows that was carried over from Quake 3.  I heard rumors that tjw's version fixes this but I'm not certain.

[David]

I meant sending the full cvar list would cause lag.
And when you make a bind or set a cvar the server doesn't get told, unless that cvar happens to be set as needing to be added to the clientInfo, and it can only be set as such client side.

But as said before, nothing the client sends can be trusted.

[Patriotpie]

The server cannot manipulate the client's machine in such a way that I know of. It can do pretty much anything it wants to the trem client itself, and despite what boastful server ops will tell you, they cannot do anything (remotely execute/modify files) to your computer. I suppose it is possible to make the client download a .pk3 with an auto-extracting virus in it, but why would they want to do that?

Punkbuster (god forbid) takes a regular SS of exactly what the player is seeing on their screen, but punkbuster is a clientside mod as well as serverside... it also lags server like crazy. I'm not by any means saying we should use PB, but if there was a way that this particular feature could be implemented, it could be an rcon command to take a screenie of a client's screen.

As for adding a regular n_command check to the standard repertoire of sv_pure command checks... shouldn't be that hard.
Current funlily leechers won't be expecting this (if it's implemented) so of course while it won't be as effective in the long-term, we can at least slow cheating down for now.

[David]

But adding n_ cvars would require a new client.  And not just a mod, but a new tremulous.x86.
A new client will need the aimbot to be recompiled anyway, and he will just change the cvars it uses.
Tremulous has many settings that do nothing witch could be reused.  Or it could read the config from a file.  Or hook the console so its cvars aren't 'real' cvars.

EDIT: Spelling

[tehOen]

people who think that they can rely on the data sent by client, makes me LOL

[beerbitch]

Basically the best solution is for humans to handle this, which is why we need an organized forum where admins can share bans.

[KusaKari]

from what I see though, most people using the aimbot are ridiculously stupid.  Even more stupid than I am.  Go to their forums and read replies, how are this work.  It would be up to funlily to change the commands ect. and reupload and he doesn't seemed encouraged to do so.  This would at least slow him down and stop most of his current users.

[Superpie]

ban aimbotters, not operating systems

[Neckhole]

This can be put on a feature list,  but as an admin I'd really also like to be able to query (in real-time) the master ban list.  If some guy joins my server and acts like a tard, I'd like to know if he's been banned on other servers that I recognize (or trust to be admin'ed well).

Or at the same time, when considering banning someone it'd be worth it to me to see if that player's been banned at other places.  A guy who has been banned on no other servers might deserve a little more leniency.

Some sort of realtime query for players/guids/IPs connected to your server would be very useful.

For those of you who probably haven't played on our server (The Potato Patch), you're not aware that we have a "bot' that sits and watches the console and does a number of commands.  Wonko, who coded this bot, says it'd be relatively simple to add this query (and other functionality of a master ban list) to the bot.    If it helped, we could work to help distribute a copy of the bot designed to interact with the master ban list.

Edit:  For anyone thats interested, please stop by The Potato Patch (after working hours) and talk to Wonko and I about it.  If you're curious about the bot, you can check out some of the things he can do on our website (http://potatopatch.brianmoses.net). But the best thing to do is just stop by and we'll show you.

[Fluxflashor]

The server cannot manipulate the client's machine in such a way that I know of. It can do pretty much anything it wants to the trem client itself, and despite what boastful server ops will tell you, they cannot do anything (remotely execute/modify files) to your computer. I suppose it is possible to make the client download a .pk3 with an auto-extracting virus in it, but why would they want to do that?

Remind me not to play on any servers that you own or have ftp access to

people who think that they can rely on the data sent by client, makes me LOL

And why can we not rely on data being sent by a client? Is funlily going to go and make stuff "invisible or spoofed" I dont think so.

Humans are the best way to detect, but sometimes it is hard to detect aimbotters. Humans cannot detect wallhacks though which also give a huge advantage in game. Only way to catch a wallhacker is to spec with a wallhack. Im almost to the point of lets face it Tremulous is Ruined.

[Eeeew Spiders]

Im almost to the point of lets face it Tremulous is Ruined.

Most players do not cheat, the number of cheaters is very small, especially considering that different cheaters are often infact the same person. All is good. The player base is growing and this will lead to some growing pains. The right measurements will be taken in time to make management of players (query on IP/GUID, (global/local) authentication etc etc) easier when it keeps growing. As long as most server operators stay alert, recruit good admins, and developers keep developing nifty tools for operators to maintain their server with ease...the future looks bright, we gotta wear shades.

[David]

The server cannot manipulate the client's machine in such a way that I know of. It can do pretty much anything it wants to the trem client itself, and despite what boastful server ops will tell you, they cannot do anything (remotely execute/modify files) to your computer. I suppose it is possible to make the client download a .pk3 with an auto-extracting virus in it, but why would they want to do that?

The QVM runs in a VM so that cant be done.

people who think that they can rely on the data sent by client, makes me LOL

And why can we not rely on data being sent by a client? Is funlily going to go and make stuff "invisible or spoofed" I dont think so.

Real easy to do, so why wouldn't he?

[AKAnotu]

And now there are only us Mac players left.
All 3 of us.

When did we get a third one?

*waves hand*

[benmachine]

And now there are only us Mac players left.
All 3 of us.

When did we get a third one?

:) *waves hand*

>:\/ WAARK

[stalefries]

And now there are only us Mac players left.
All 3 of us.

When did we get a third one?

*waves hand*

Does that make me #4? Or does benmachine's cryptic message above make him #4, and me #5?

[n00b pl0x]

hi im a lamer (a.k.a. mac user) so im going to post here on this thread because someone mentioned macs

[KusaKari]

from what i've looked at though I don't understand trem code or c very well the server does have quiet a lot of control over the client.

@ david, yes funlily could easily attempt to spoof the server, but he doesn't have the go getter attitude most think he does.  Yes he made an aimbot, yes he made a tjw aimbot.  But look at the gap it took him forever to get the GUID acceptable version out.  He lacks diligince as most aimbotters do.  Because the aimbot is not for himself.  He probably doesn't even play trem.  he makes it because hes an asshole and likes hurting others.  He wont go so far to work on it to that extent I feel though.  And almost all of his current users aren't smart enough to pick up anything in his place.  Especially since he kept his source code to himself.  A simple check for n_commands wouldn't be a hard patch to make for anyone with experience and it would slow him down at least 2 months while something better is worked on.  And would discourage/stop most of his current users.  

I am currently attempting to make this n_command check but I have no experience in c or in trem source itself so I am having difficulties.  If trem was made in c++ this would be a lot easier for me.  While c is similar the syntax difference and the overall complexitiy of trem is hard for me.  

Here is my idea for a more advanced anti-aimbot.

The server can obviously see movements every client makes, or clients wouldn't be able to play with each other.  funlily's aimbot aims at the same spot in the hitbox for each model.  At first I thought anyone who hits that exact point 3 time in 1 second is obviosly using it.  But you can use his cvars to change the predictions and xy locations of where it points.  This would make it to much to check and deffinately lag the server.  Now what if instead we checked if someone hit the same x,y point relative to a hitbox 3 times or more in 1 second.  After moving of course.  No human player no matter how good their mouse is can follow a point on a hitbox exactly.  Does this make sense yet?  Lets imagine a square.  thats the hit box.  If someone aims at 10 x and 14 y inside the hitbox.  the hitbox moves the person follows and lands back exactly on 10x 14y again.  These points will have to be relative to the hitbox of course because otherwise the points would read different as the hitbox moved.  There is no way a human player could follow a hitbox further than 10 pixels and land back in the same x,y position.  So if the server saw someone hit the same point relative to a hitbox 3 times in a row in a small period of time.  When the hitbox has shifted 20+ x,y locations.  Its obvious its code doing the aiming not a human hand.  Checking for the hitbox shift also would prevent false positives of a person leaving their mouse unmoving.  Also the short timing would prevent that someone accidently falls in place of the same relative hitbox point 3 times in a match.  It would have to be in a small period of time.  

I have this whole idea in my head and no way to put it into code yet.  i'm learning fast though and I think I should be able to start testing my first anti-aimbot patch within a month.  I would like help if it can be offered.  I would also like criticism.  Counter arguements are the best way to find out whats reasonable.  

If I could get my hands on funlily's code it would be so much easier.  Instead of doing what i'm trying to do I could just have the server check for his actuall patterns of movement.  That would be the easiest but of course he could just change the bots patterns slightly and I would have to redo everything.  So the above idea to me seems as the best idea for an overall aimbot anti-patch.  It at least limits basic aimbots.  Advanced aimbots could get around it but what cant advanced aimbots get around.  Most people wont go to that much trouble to create something that will get around checks like this.  If they do they generally don't feel like sharing also.  

So just reply please and rip me apart or offer help or compliments whatever you feel is necessary.

[stalefries]

I think your idea (while good) falls into the same trap as the other ideas: somehow, the coder of the aimbot can modify that aimbot to avoid these checks. For yours, all it would have to do is slightly randomize the hit target. But I imagine if we could somehow roll out a bunch of these fixes at once, it would make it that much harder for the coder. But first, we have to do some *real code*.