Hack Cam Interview
Questions by Rocky and Cobblers, all replies by OutOfAmmo.
Hack Cam is an exciting new approach to catching cheaters. Rather than an never ending chase to identify new cheats and then counter them, Hack Cam introduces a new concept - spot a cheater by what he does, not what he uses. To explain the details behind Hack Cam, it's developer OutOfAmmo agreed to be interviewed- so here it is, enjoy!
Q0. Can you tell us who is on the Hack Cam project and their backgrounds?
Currently the only designer of the HC System is myself. I have several people involved in on-going testing with current releases on private networks elsewhere additionally. I have been programming actively for roughly 16 years thus far with concentration in C++ and ASM coding. I am currently the Chief Systems Officer for a networking security company with concentration in cryptographies, network infrastructure, and secure content handling and dissemination. I also code in VB, C, C#, COBAL, PASCAL, BASIC, Java, HTML, PHP, CGI, SQL, and Oracle systems.
Q1. What makes Hack Cam fundamentally different from previous anti-cheat efforts?
Traditional anti-cheat systems concentrate on finding a cheating device of sorts on a user's computer prior to gameplay. While effective, the inherent problem with it is that if a new cheating device is created the system will need to be updated to recognize it as well. If the cheat itself is never released there is a high probability that it will never be found by the anti-cheat designers, letting one more cheater slip through the system, and as we can all agree, one cheater is one too many.
The HC System takes the opposing standpoint. There may be thousands of cheats available for players to use, but in the end, they all do the same things. A thousand different wallhacks will always do the same thing, show a player behind the wall. So why search for the thousands of cheats available, when you can just search for what they do? That's the principle around which HC has been designed. Don't find the cheat, find out what it lets a player do or see, and then test whether or not they are indeed seeing or doing those items. If a player opens fire on a door and kills an opposing player behind it, do we really care which cheat they are using if any, or do we just care about whether he is cheating or not?
Q2. Hack Cam is becoming a bit of a buzz phrase now, but the foundations for it were actually being constructed about 7 years ago. Can you tell us a little about that, what parts of the Hack Cam code have been pulled from which separate non related projects?
The HC System actually came into design a little over 2 years ago now. As you mentioned, some components were in design long before the concept was created. One of the main components in the system is the visual recognition core. In principle, it simply looks at the screen and figures out what is what from the players point of view. For instance, if someone were to throw an HE Grenade at you, the core would identify and track that grenade in the players field of view.
The original purpose of this coding was a Computer Science course at the University I attended. The goal was to simply create an application that could find a human face in an image. Once that was completed, and in my spare time, I went about implementing more features, such as being able to identify multiple faces in a frame and specifically identifying those faces. The same type of system is used in most high-tech security systems for identifying employees through CCTV. At the time of HC's inception, I figured that if they can do it at the FBI headquarters, why can't I do it in game?
Q3. If Hack Cam works by identifying kills that are above normal skill expectations, can you be confident that extremely skilled players who, for example, know certain maps inside out, will not be penalised for simply being very good at what they have practiced long and hard at?
Extensive testing has been performed thus far on the system to guarantee that good players will not be registered as a cheating player. The HC System runs on a point basis, running from 0 to 100. Through testing from demos and live gaming a baseline was constructed. 95% of all the players scanned by the system scored a 50 or below on the point chart. As new routines were added they too were balanced to ensure this line of normality.
Demos from CAL matches have also been processed in earlier versions with the end result of a 65 being the highest value found. The system decides a player is most likely cheating when their score passes the 'golden line' as it's called at 85. The simple reasoning behind it being that if 95% of all players fell below a 50, and professional players fell below a 65, a score of 85 should be so unattainable that only cheating players could reach it.
Q4. So, for example, how would this scenario play out in the eyes of Hack Cam? A player approaches a corner and in anticipation of finding an enemy sitting at a commonly used camp location. He does this with his weapon already raised and pointing in that direction before he actually rounds the corner. As he turns the corner he immediately lets off two quick rounds, and gets a quick kill because someone did happen to be there. Would hack cam penalise that player, perhaps mistaking his familiarity with that particular area of the map for a wall hack or similar? Although this is a pretty detailed scenario, I know that a lot of gamers will recognise such a situation and be very interested in how Hack Cam deals with it.
Though detailed in description, there are a few other points that would determine whether or not the player would be penalized and by how much. For instance, has that player found someone there previously? Has any player found someone there previously in the map? How often is that specific spot used in all runs of the map? Was there gunfire in the corridor prior to him moving out of his position at the corner? While pre-aiming at his intended target, how accurate was the aiming? Is there a shadow visible on the walls or other bullet holes that might tip our player off? Can he hear feet moving towards him or did the opposing player reload their weapon or change weapons? As you can see, HC is a bit more extensive then some people believe at first glance. Just because he fires on someone he cannot see, or pre-aim on a location prior to actually seeing them, there are dozens of other variables that must be taken into account.
Now, if we take your example as is, and assume no to all of these questions, the player would be most likely be penalized, but the penalty would not be severe enough to cause a false-positive result. If any one of those questions had a yes to it, the penalty score would be reduced, if not nullified completely. Normal players do get penalized throughout the scanning process. Let's not forget this is a computer application that relies on logic and instructions, so it will give points to players even if they are legitimate. These players receive what I commonly refer to as 'background-noise' scores. A few points towards a clean player in the end means nothing.
Q5. How about more cut and dried situations like anticipating situations with the help of sound or visual clues like footsteps or shadows? A switched on player will make the most of such clues and capitalise on them, will Hack Cam take cognisance of such details or assume the player must be cheating to have anticipated the enemies position so well?
Audio cues and visual cues are considered and handled during a scan process. If you hear someone coming, or find a room with a lot of bullet spam on the walls, you will naturally be more cautious when entering it and handling the situations inside. For instance, if you were running down a path and heard footsteps to your right in a building without a blip on your radar, you're going to know someone is there, so suddenly changing directions and firing on the room will not increase your score.
Q6. When do you think we will see Hack Cam released and in what format will it be, i.e. will it be incorporated into a new game, or available as a server/client side tool?
Talks are currently in progress and the release of HackCam 4 is slated to be sometime this month if no more delays arise. The V4 release is intended as a server plugin to allow for scanning of all players in a server during a live game. After that release is out the door, beta-testing and final release of V5 will be handled. V5 is a client-side application that can either be used for scanning of players in real-time, or for review of demos, in either 1.6 or CSS formats. The actual final design of V5 will not be decided upon until then.
Q7. Are you working with any game developers on incorporating Hack Cam into their existing or future games?
The HC System is currently slated to be implemented inside of an existing game system and we are currently in talk with 2 other companies regarding building it into their systems permanently prior to their original releases.
Q8. Are you planning on releasing a beta for download for any currently released games?
The V5 release of HC will be beta-tested prior to final release. There is currently an ongoing list of players interested in joining the Beta Tester list at http://www.az-prod.com where the HC forums and discussion are hosted. The V1-V3 editions have not been released and will not be released for design reasons.
Q9. Is there a way of creating a central database of people whose scores were around the 80%? This way, everyone can feed the list as well as feed from it.
A central database is in the works. All scores thus far have been logged with the ID's of the players, but release of this information to others is being looked at closely by the legal team. Once that is settled an interface for dissemination of bannable IDs will be designed for server admins to use themselves.
Q10. Do you think Hack Cam could be fooled by a cheater, who uses cheats to identify enemies but doesn't make the cheat blatantly obvious?
There are generally 2 breeds of cheaters. The first breed are our friendly neighborhood script-kiddies. These are the ones that download the newest cheats and use them in-game to their blatant advantage. If you've ever seen a speedhack or an aimbot in use, these are them. These are also the easiest to catch. If we look at the cheating population as a whole, 75% of our cheaters fall into this category.
The second category is the one you mentioned, the cautious cheater, that uses it for informational purposes only. They may have a wallhack or other cheat enabled that allows them to see where an opposing player is, but are smart enough not to aim directly at them, or make other moves that would lead someone to believe they are legit. This makes up the remaining 25% population and are harder to catch, but none the less, can be caught. The reason they can be caught is because they are human. Humans make mistakes, we're fairly good at it. In the end, it only takes one really stupid mistake to be caught.
For instance, a player is running towards a set of doors during a rush. Suddenly 4 opposing players appear behind the door, much more then anticipated. A normal player would continue the rush, only expecting the 1 or no players present they've imagined. Our good cheater however will suddenly realize in a split second that he's outnumbered. Instinctively, not by choice, he changes tactics, with no visual or audio cues to grant a reason for that reaction. Right there, he would be tagged as suspect. A more common scenario is a wallhacker thinking they can see someone behind a box (since the wallhack disables to some extent the ability of depth perception) and opens fire dead on at his target, only to realize a moment later that he was incorrect and stops firing. That again, would tag them.
Overall, can HC find 100% of cheaters? I prefer to take the Listerine stance. I'm fairly sure I can catch 99.9% of cheaters, but I will not guarantee 100%. After all, I'm human as well. In the larger look on things however, each cheater found is one less we need to worry about in the future. One less, is better then 0 less any day.
Q11. Do you feel that HC will be the future of anti-hacking devices? What are the advantages over anti-cheat systems that interrogate a gamers game installation looking for rogue files as proof of cheating? I see the HC standpoint being a good start for future devices, but is it the future?
Well, only the gaming community can decide that. HC has advantages and disadvantages over other anti-cheat devices. One advantage is that it only looks for what the cheat does, so no real updating is required when new cheats become available. However the disadvantage is that in the end, you must use your own judgement on whether the player is cheating or not. HC should be looked at as an anti-cheat tool. It can give you statistical and empirical data pointing towards a player, but you must decide for yourself if you want to ban them. Whereas an anti-cheat device that looks for matches in files can be 100% accurate each time, though will require constant updating to make sure the list stays as complete as possible. I simply envision a dual anti-cheat system. Use the traditional methods to scan the users system to remove a large population of the cheaters right out. After that, HC can weed out the few remaining ones that may have been missed or overlooked.
Q12. Is Hack Cam easily scalable for use in different games? For example could it be used in previously released games or only certain future games?
The HC System, in theory, could be scaled to any FPS type game. The easiest scaling of course would revolve around the Half-Life and Half-Life 2 modifications of the engine, such as CSS, DOD, and others. Scaling to other FPS games is possible, though obviously more time and resources would be spent porting over to them.
Q13. What further developments are in store for Hack Cam?
HC will be constantly going through development to add new routines for checking specific scenarios where a score increment could be given though shouldn't or not as severely. Past that, the gaming community and the cheating community will inevitably decide the future of where it goes.
Q14 What part of Hack Cam are you personally most satisfied with? And which part gave you the most headaches?!
I am personally most satisfied with the reaction that HC has received so far from the community. For over a year now HC was a small secret tool used on The-Space.net community of gaming servers for review of suspect players. I finally decided to put out a video release to all of the gamers in our somewhat small community so they could finally see just what the system looked like and did. I had not expected at that time for the video to propagate as much as it has. To date, the video itself has had over 3000 downloads and has been mentioned or posted about on 60 different gaming community websites, including the cheating community.
Though many scoff at first glance and proclaim that it will never work, once they are informed that the reasons they give for it not working are already handled, they quietly sit back and realize just how much of a beast this truly is. Certainly the part that has given me the most headaches is the engineering behind it.
The V3 release we used on our community was a 500MB DLL that was launched prior to gaming and ate resources like they were cheetos. The last few months have been spent converting a large portion of the code out of C++ and into ASM, reducing the file size to roughly 75MB and the resource requirements less. Design is still in progress to reduce these numbers farther and farther.
Thank you very much for taking the time to answer our questions so fully. Good luck with Hack Cam and the fight against cheaters!