CGDC '97 Roundtable Report

Unlike box games, the focus of the online game has traditionally been the community that the game engenders. But to keep the community (and the product) viable over the long term takes design and ongoing support. Costs of 25-35% of development resources spent planning for and dealing with problem users are the norm, and more than half of the long-term maintenance resources are dedicated just to dealing with problem users who can poison and destroy a game community. One of the roundtable participants made the comment, roughly paraphrased, "I've spent my career being a computer engineer, and now I find I have to be a social engineer too."

The roundtables produced a rough taxonomy of problem users. These users, broken down from least difficult to handle to the most threatening, are as follows:

Nuisance Users
Nuisances are often some of the most dedicated players, who have for some reason, whether it is obsolete information from the developer, or deductions of their own, come up with a detailed technical model of how the game works, that is intricate, exact, and wrong. This wouldn't be a problem, except that the bug reports they submit (and the "easy features" they demand) are colored by this false information. Moreover, they will often explain problems other users are having in very authoritative, and utterly incorrect, tones, based on this false information. Still, nuisances can generally be handled fairly easily by product support.

Abusers
Something you should reconcile yourself to in the online world, is that you will almost never get an immediate bug report for any bug that gives a player a tactical advantage in the game. If you are lucky, you will hear about it when enough players have found out that it is no longer an advantage but a threat. In general, abusers make up of some of the most dedicated of the players, but can also include players who have found ways to cheat by modifying the game front end. Because this group is made up primarily of serious players, in-game ramifications for not reporting or exploiting bugs and policing by product support can mitigate the problem. Don't plan on eliminating abusers, as long as there are bugs in the design or the code to exploit, the players will.

Wreckers
Wreckers are easily the most troublesome combination of size and problems. The hallmark of the Wrecker is that they aren't playing your game at all. They are using your game to play their own game, and the name of that game is "How many peoples' evening can I ruin." Their behavior will run the gamut from cheating, to anti-social behavior (such as destroying characters or other game constructs that your serious customers had spent weeks, months, or years creating), to use of profanity or epithets to drive people out of the game. In-game penalties or the closing of accounts are not usually effective with this group, simply because they expect it to happen, and often consider it a challenge to get back in.

Predators
This group was listed in the first session, by a roundtable participant working in online children's games, and are a major worry in that market. The paedophiles and other predators have concentrated in the chat rooms and services up until this point, but the increasing ease of access and continued removal of price barriers will only make the online games more of a target.

While identifying the types of problem users is a necessary first step, it doesn't answer the question, "How do you design games to deal with them". Figures floated at the various sessions are that 25-35% of your development resources go to dealing with the problems these users pose. That includes the design and development of more sophisticated logging and analysis tools, to find the incidents after the fact. It includes the time spent finding and patching problems unreported by the beta users because they were too useful. But most importantly, it includes the resources spent in analyzing every feature for its potential abuse and possible remedies.

The roundtables themselves primarily focused on the wreckers. Nuisance users can be dealt with with solid product support. Discussions of cheaters concluded fairly quickly that they can be blocked with heavy validity checking and a consideration that the front end is always untrustworthy if the game design makes that possible. And frankly, we couldn't find an even reasonably good answer to how to deal with the predators.

Means of dealing with the wreckers that were suggested or used as examples during the sessions fell into three basic categories. The first was a form of retroactive punishment. Company support staff would review logs or complaints, and take action (whether internal to the game, or involving an online service) against people violating the game rules or codes of conduct. The second was a form of societal emphasis against bad play, whether through rewards and encouragements for good play, automated means of singling out offenders for punishment or shunning, or through an attempt to convince the user base to simply ignore the most egregious problem cases without using automatic means to identify them.

The first two categories were the model for the first decade or so of the commercial online game (starting in the early 1980s). When the number of active users was in the low hundreds, out of a total game base in the thousands, these methods were reasonably effective, and certainly some of them (especially contractual requirements on game developers to monitor user actions in the game for network Terms of Service violations) are still in regular practice today. A fair amount of the discussion in the roundtables centered on means to make this particular set of models (especially the community based models, which have the advantage of reducing load on game support staff) work in the modern era of thousands of simultaneous online users out of player bases of more than one hundred thousand. Ideas included automated "wanted" lists of offenders, councils of senior players to work on a somewhat self policing system under product support supervision, and mechanisms to segregate the troublemakers away from the general pool.

The third category was automatic punishment; actions taken immediately by the system, whether that be in-game examples (refusing to reload the weapons of players who continued to shoot people on their own side), intrinsic parts of the game design which increase the time investment required before damage can be done, or "outside the rules" actions, such as locking out accounts for too many kills against other players in an RPG, with no corresponding in-game rationale. While these solutions have the advantage of working with minimal staff supervision, they are also far more limited in the sorts of behaviors they can recognize.

Beyond identifying the types of problem users, and possible means of dealing with some of the most troubling, the rest of the discussions focused on the aspects of multiplayer gaming which cause problem users to flourish. After all, recreational soccer leagues rarely worry about players intentionally scoring against their own team, the chess federations mechanisms for ejecting troublemakers seem to work well, so what is intrinsic to the online marketplace that makes us so prone to this? Again, the discussion centered on three primary issues.

The first, and easily the most important, is the anonymity and pseudonymity of the net. It's very easy to change identities, and continue to wreak havoc without real risk of any punishment other than simply being locked out of a game, and the fact that troublemakers only have to look at the ghost of the trouble they cause (rather than confronting it directly) makes it easy for troublemakers to justify it as "other people taking games too seriously". This anonymous or pseudonymous aspect makes it very easy for the same group of troublemakers to cause problems multiple times, something lacking in more conventional settings.

The second is the ease of access. Given the scope of the net, and the influx of new users, the economies of scale let like minded people clump, and easily gain access to new areas in which to wreak havoc. It is far easier to flip from online game to online game than it is from soccer league to soccer league.

Finally, there is the low time and money barrier to causing damage. This last aspect was the only one for which any solutions were proposed. One participant mentioned forbidding players from transferring properties and materials to another player for the first month of a game, to increase the time commitment required before cheating by having multiple accounts in the same game would pay off.