1
Contents.
Part 1: Few elements of a possible anti-cheat system: concepts
1.1 By players for players: "cheater!" button (initial filtering)
1.2 Confirming a cheater
1.3 Punishments for cheating
Part 2: a few notes and my personal testimony establishing (for me, at least) obvious need of more/better anti-cheat systems (more and/or better than there currently are, if any)
=================================================
Part 1: Few elements of a possible anti-cheat system: concepts
________________________________________________
1.1 - By players for players: "cheater!" button (initial filtering)
----------------------------------------------------------------------------
The button which would read "Cheater!" or something similar ("I think this is a cheater", "This guy is cheating!", etc) - is to be made available for players in spectator mode.
- why: because PGI does not have enough people to spectate all the players regularly, nor should it have.
- effect: whenever this button would be pressed by a spectating player, spectated player would have +1 to his account's "suspected cheater score". The score should be kept invisible to the owner of the account, as well as invisible to any other players. The score will be used by automated systems and possibly PGI specialist(s) to confirm or postpone an actual judgement.
To prevent abuse of the button, the following (and possibly other) restrictions are to be implemented to the "cheater!" button function:
- the button can not be used on a same player more than once per account per month. It's best if the button would still be displayed for "already reported" players, - but would be inactive, with the button's name changed to "Reported!" or something similar. Once a month passes, same spectating player can press the button while spectating same suspected cheater once again;
- the button can not be used by any account which still has its "Cadet" bonus present;
- the button can not be used by player A, if spectated player B - is currently a member of player A's group, Merc Unit or pre-made company (because this button is not intended to be a tool for internal punishment or sabotage in any case);
- the button can not be used more than "Lvl / X" times (rounded up to integer) per 24 hours, where "Lvl" - is spectating player's level (assuming this will be implemented quite soon, as promised), and "X" - is such a number that players who have maximum possible level would have 12 uses of "cheater!" button per 24 hours. This restriction prevents people from labeling all people they meet as cheaters (some folks are just straaange, yep), and it also grants more power to label players as cheaters to more experienced pilots, improving overall reliability of the cheater score. Should a player reach maximum allowed uses per 24 hours, it would be best if the button would turn inactive, and the name of the button would change to "Max uses per day: Y (level based)" - where Y = "Lvl / X".
Notes about "cheater!" button:
- since it is only possible to spectate members of one's own team (for its own reasons) - there is no possibility that players will "punish" any enemy with the use of this button;
- the button provides an occupation for any player who were "killed" in the current game, but for some reason desires to wait for current game to end (and not to quit it and start another with a different mech). Some of such reasons - are: the desire to start next game with friends (who might still be "alive" in the current game), the desire to play the same mech in the next game, the desire to observe other players and learn from their gameplay. Great many players spectate because of those and other reasons; thus they are already automatically able to see at least some forms of cheating, and many among them would gladly use "cheater!" button should such a button be present and easily spottable while spectating someone, imho.
____________________
1.2 Confirming a cheater
--------------------------------
The "cheater score", mentioned in 1.1, would in time accumulate. No player, however, may be named a cheater based on this score alone: because, obviously, exceptionally skilled and able players will inevitably have many novice and/or silly pilots suspecting them in cheating. This happens in pretty much every large shooter multiplayer game in existance.
In my opinion, PGI specialist's judgement upon spectating the player and/or observing the player's stats - is a required action to ensure that nobody could be punished while not being guilty of cheating. However, much of this can (and possibly should) be automated by some relatively simple features:
1.2.1 [PGI] in-game accounts can be given an "observe suspicious player" button:
- which would allow the owner of the account to join a match (possibly one already in progress) as an invisible spectator, and automatically put the owner of the account to spectating the player who has high cheater score. It would be best if the player's stats would also be displayed for [PGI] observer, right in the game.
- while spectating such a player, [PGI] account owner can be given two more buttons (both optional; [PGI] account owner may choose not to press any of the two, if it's what he wants):
- - 1st one says "Cheater indeed!" (or something similar, like "PGI: confirming cheating", etc), and if pressed, then spectated account's "PGI: cheating score" would be increased by 1 (note, this is a separate score from basic cheater score);
- - and the other saying "Postpone judgement" (or something similar like "Seems normal", etc), and if pressed, it would not increase "PGI: cheating score" of the spectated account, but would simply mark spectated account as "recently observed". This button can be pressed if [PGI] account owner does not see anything suspicious AND does not want to spectate the player again in near future.
- The "observe suspicious player" button would automatically select only players which high cheater score who were not "judged" recently - 2 weeks may be? "Judged" here means both deemed as a cheater and deemed as a seemingly normal non-cheating player (by one of two buttons described just above).
1.2.2 A "chart" of current "leaders" in cheater-scores - could be presented to [PGI] accounts on the website (or elsewhere). It possibly should have:
- easily sortable by "cheater score" and "PGI cheater score" columns,
- marks for players who were already estimated recently by the [PGI] account owner (ideally, with a date of last visit to such players' pages also displayed for the [PGI] acocunt owner). Marking estimations done by other [PGI] account owners, - is not needed (because their judgements, if any, would already be reflected by "PGI cheater score" of players in question);
- a "Cheater indeed!" button next to each account name (or something similar, like "PGI: confirming cheating", etc), - if pressed, then the account's "PGI: cheating score" would be increased by 1 (note, this is a separate score from basic cheater score).
Such a chart would especially be needed if it's not possible to see the player's stats in the game - but also otherwise, just to a lesser extent.
"PGI cheater score" would be the main method of confirming whether a person actually cheats or not, assuming PGI specialists who will do track cheaters, - know the game and have good judgement. A score of 3 or higher is, i guess, what would be needed to "mark" the player as a "cheater" - this is additional protection against mistakes, and also still grants a chance of redemption for people who tried to cheat for a while, but then decided to stop cheating and play fair.
A player may at some point stop to be a cheater. In fact, many thousands of players try some or other cheats at some point (if not majority of players, even), - yet many of such players end up being good, fair, competitive players of the game. Therefore there much be a mechanism which, with time, would "forget" the "sins" of a player. Not too fast, of course, but not too slow either. For scores, typically, a regular % decay is the way to do this:
- I'd say, 20% monthly decay for PGI cheater score, whenever the player's score is not increased for a month. So if someone "reaches" 3, then for a month he's marked as confirmed cheater, but then, if his PGI cheater score is not increased for a full month, his PGI cheater score is dropped by 20%, - i.e. to 2.4, - and so he's not marked as "confirmed cheater" anymore. Should he remain fair playing, in just 4 months his score will drop further to ~0.98, i.e. below 1, returning him to the pool of players who are not deemed particularly suspicious by PGI staff;
- "cheater score", - the one set by players themselves as described in 1.1, - probably needs a decay too, but i guess somewhat slower one. Perhaps 10% per month, substracted at a given date (end of months? regular maintenance?).
_______________________
1.3 Punishments for cheating
------------------------------------
I believe that even with the best possible intention and execution, mistakes in judgement might sometimes happen. Therefore, punishments for cheating should imho never be permanent - especially in a game which has paid-for content aquired by some of accused players. However, even with temporary measures, it is possible to keep cheaters at bay. It certainly remains at PGI's sole discretion whether and how to punish cheaters, but here's some things i'd do to players with high (read - 3 or higher) "PGI cheater score" (as it's defined in 1.2):
- introducing a hidden feature into match maker, which would force match maker to spend a solid time (say, as much as 3 minutes) trying to put the cheater into a match which only has other people with "PGI cheater score" being equal-or-higher than 1.0. This will result in two effects for the cheater: 1st, substantially slower search for games (he's cheating the game, so why should we care?), and 2nd, isolating cheaters much to themselves (they wanna cheat? Let them HAVE the same from their own kind in return!). For community, this will result in possibly massively less cheaters present in regular games, which is mightily good;
- perhaps making "Cheater indeed!" button - the one which only PGI specialists can see and use, - to also disconnect the cheater from the current game. Silently. Punishment right "here and now", you know?
- perhaps making "Cheater indeed!" button to also block the cheater (and any other players if applicable) from seeing the cheater's personal stats (on MWO website) for a period of, say, 1 month. Silently - just remove the "stats" button for cheaetrs for 1 month, may be? I know for a fact that some players do cheat to hoard for "awesome stats". And as far as i can suspect, PGI does not have any responsibility to show personal stats to players. Not any legal one, at least, i guess?
- in worst cases - definitely obvious, prolonged, massive cheating, - temporary ban is of course in order. Temporary, because PGI do want to keep any player who is willing to stop cheating, since this is a potential for PGI to earn more money. Ban, because a nasty cheaters make people leave, thus reducing PGI profits much more than they themselves could ever bring. As for duration, well, for PGI to say, but me, i'd make 3 varieties: ban for a day, ban for a week, ban for a month. Depending on how bad cheater's cheats are, and if his "PGI cheater score" is (much?) higher than 3, and whether it is repeatable offense, i'd use one of these correspondedly;
- in "worse than worst" cases, - very bad cheats, repeated bans in the past, very high "PGI cheat score", - permanent ban might also be in order. But legal agreements must clearly make it reserved 1st (if not yet), since MWO has paid content. Other than that, - as far as i know, permanent bans of worst cheating accounts are a usual practice in many "paid for" multiplayer games.
Note. I do not believe in "warnings". Because, 1st, many cheaters are simply ignoring it, or even have no chance to see it simply because they do not check their mailbox any often and recieve tons of spam mail; 2nd, because it alerts them about the fact they are caught cheating, without actually doing anything to stop them. May encourage them to seek for better, more discrete, cheat methods, too. And in general, - as can be seen from the small list just above, - i am a big "fan" of discrete methods for "handling" cheaters. If they don't even know what hits them, they have little to no chances to oppose anti-cheat system. Which, given limited amount of effort and time PGI could put into making anti-cheat systems, - is a very good thing, yes?
=================================================
Part 2: the need for anti-cheating systems.
In most, if not all, multiplayer games there are always some folks who want to cheat others. Good multiplayer games, - especially ones which are made to be a media for high-skill competition gameplay, - put much effort into design and implementation of anti-cheat systems. This was the case with shooters since 10+ years ago, in particular.
I guess MWO qualifies as a game which would benefit greatly if some good anti-cheating systems would be implemented and maintained. In Part 1 of this post, i've suggested some highly effective (in my opinion), low-cost methods to implement and maintain one quite efficient anti-cheat system. To back it up, i now must give a solid testimony about actual presense of cheaters in MWO. So, here it is:
Case 1.
A few days ago, in a game, i was spectating "through eyes" of an assault mech for some ~4 minutes. During that time, the mech inflicted large amounts of damage to 5 opponents. It had a few lasers, 2 autocannons, and some LRMs. Thing is, his aiming was clearly automatic - every time he was targeting some enemy, once the targeted enemy was visible, his reticule was placed right on the middle of the enemy, and then his reticule would follow the enemy no matter how enemy would strafe (fast or slow), no matter how far or medium range the enemy is - except there was a tiny lag, causing his recitule to get slightly to the side of any strafing target. Like, a ~0.05 seconds of inertia. But other than that barely noticeable lag, it was exactly like his recitule was "glued" to his current visible target, always staying on _precisely_ same part of the enemy (always pointing to very middle of an enemy, i.e. lower half of torso; never to upper body parts, never to legs, never to arms).
He never used any zoom, too. Not once - even while hitting moving targets some ~500 meters away.
I've been watching this guy missing lots of autocannons shots when targets strafed, and he never attempted to correct his aim. I've been watching him hitting his targets with lasers 100% reliably, with that "glued to the target" reticule of his. Together with his LRMs, and autocannons (whenever his targets were not strafing, or strafing little), he did big-time damage (his total for the game was some ~800sh, with 4 personal kills, and his team lost the match with that; should he be in the winning team, it'd easily go much over a thousand as he'd continue to chew opponents down with ease).
I am sure this guy was a cheater, and i think solid effort should be made to make such players to be unable to keep playing MWO in this manner. At least, if MWO indeed attempts to become a serious competitive shooter.
Case 2.
Some ~2 weeks ago, i've been spectating "through eyes" of a spider, for some ~3 minutes. His aiming reticule behaved in a way which is not possible for a human to produce: namely, whenever this spider 1) had a target, 2) very recently had the target in his LoS, and 3) at the moment had the target out of his LoS, - in other words, whenever his target left his field of view, - his reticule, for a couple seconds, did chaotic, multiple-per-second "instant" movements (more like, jumps) on his screen. Each time for same "distance", - visibly for ~50...70 pixels or so on my screen, - but every time in different direction, and always "around" his current torso-heading point (the torso's crosshair).
Any lag issues he or i might have would never produce so consistent "on" and "off" behaviour - "on" when he just lost the view of his target, but "off" in a few seconds later and until next similar situation. That's why i strongly suspect that this guy was a cheater. Oh, and he was very darn precise with his lasers when fighting other lights, too. Not 100% precise, but not far from it.
=================================================
This concludes my post. Thank you for reading, and please vote in the poll if you have a moment, too.
o7
Edited by FinsT, 14 October 2013 - 06:19 AM.
I don't have enough space for more recordings. A "report" function would be highly useful, being that PGI already knows exactly how often any weapon is fired, how much damage it deals, and based on this, the frequency of hits, etc. They should have enough data to review.
