Soapbox: Designing an MMORPG Feedback Rating System

MMORPGs are afflicted by a host of real-world social problems. Dishonesty. Theft. Malice. Petty disregard. The list is endless, and we've all experienced its bite in one way or another. There's the person who abandons his group after achieving his own objectives, despite an agreement to play through to the end of a dungeon instance. Or the person who loots a valuable item they are not entitled to, then teleports away. Or the immature player who takes offense at an innocent comment and proceeds to harass you incessantly. For every real-world vice, there are a thousand virtual-world cancers. After all, when nobody can see your face, there's not much stopping you from being a jerk.

There are systems available for dealing with these issues, but they are incomplete at best. You can silence a player who is harassing you, but that isn't very useful when you're stuck halfway through a raid with them in your group. You can report particularly nasty behavior to a GM, but it's the more subtle stuff that ruins an MMORPG over time. An onerous “master loot” system can protect group members from loot theft (“ninja-ing”), but it requires everyone in a group to trust one person, which isn't always practical. Given these issues, perhaps it's time for MMORPGs to try a feedback rating system, a la eBay. This article will examine some of the issues with and potential objections to such a system.

The Foundations of a Reliable Feedback System

When playing an MMORPG, I should be able to give a positive, neutral, or negative rating to anyone who has been in my group for more than thirty minutes. Negative ratings could be characterized via a multiple-choice list of common gripes (i.e. “loot theft”, “abusive language”, etc) -- a feature now built into the Xbox Live feedback system. However, it isn't clear that a good feedback system requires this level of depth; there's an argument to be made for simplifying the process as much as possible.

A summary of all the ratings that a player has ever received (and given) should be a prominent part of his/her online profile, and clearly visible to anyone who wishes to group with them. Showing the ratings that players have given will discourage pettiness, since overly critical and/or sensitive players will soon find themselves barred from most groups. Criticism should have a price.

Repeat ratings (i.e. from players that have grouped with and rated you in the past) should count for successively less in the overall rating, or should be disallowed entirely. This will help prevent the reputation system from being gamed by groups of players seeking to dramatically alter someone's rating. Likewise, making group participation a condition of giving a rating will help stifle abuse. And if minimum time requirements on the life of the group prove insufficient defense against abuse, perhaps other dependencies could be added or substituted; for example, requiring that groups visit an instance together and/or kill X number of mobs. This would also help “encode” a reasonable amount of depth into the ratings people give one another, which would raise the predictive value of those ratings. Admittedly, it would also preclude the feedback system from discouraging malice during quick and/or informal encounters, but an abuse-resistant system that significantly improves group play is still much better than no system at all.

A Nuanced System for a Nuanced Virtual World

You can protect the feedback system by restricting the give/take of ratings to a minimum character level (say, 10 and higher), which makes it harder for people to create characters with the sole purpose of giving out good or bad ratings. It also gives players time to learn the game's functionality and its cultural conventions before other people can start rating them.

Ratings can also be designed to fade over time, and/or fade with character advancement, and/or fade if the rating-giver has not played for a certain number of hours within the past month or two. This would make the system more forgiving and more difficult to abuse. Since your ability to re-rate other people is already limited, a fading scheme further reduces your ability to prop up a friend's rating over time. Making the life of a rating dependent upon the continued (at least minimal) activity of the rating-giver reduces the incentive to create new characters for the sole purpose of handing out ratings, just as a minimum character level requirement does.

Minimizing the Distaste of “the Transaction”

There's something less-than-romantic about giving someone a rating at the end of a joint quest. One way to combat that sentiment is to employ a little bit of storytelling in the rating system. In a fantasy-themed game, you wouldn't be “giving someone a rating,” you'd be “carrying word of their deeds back to court (or the tavern)” – or something like that. In fact, if you really wanted to be cute, you could design the system so that reputation stats aren't propagated until a player reaches an inn or speaks with a bard. And ratings should be displayed in as subtle a manner as possible, except when players actively choose to drill down into them. Perhaps a small tool-tip-like window (summarizing the ratings given and received by a player) could appear when your mouse hovers over that player's name, or the crest on their armor.

Exceptionally-respected players could be rewarded with a visible title, in much the same way that titles are granted to proficient players (i.e., “The Distinguished Captain Jack”, as opposed to merely “Captain Jack”, or just “Jack”). Because this reputation-based title would derive entirely from group-performance, it is best to avoid words (such as “honorable”) that imply goodness on a global level; after all, a good group member may still be a fraudulent tradesman or an abusive parent. Other honorifics that might work within the system I've described: “renowned”, “remarkable”, etc.

One important decision is whether to make ratings character-specific or account-specific. Character-specific ratings would enable people to get a “fresh start” when they want one. It also permits for role-playing; i.e., maybe I want to play a character that behaves obnoxiously, as well as another character that behaves like an angel. On the other hand, account-specific ratings would make the system harder to abuse, and would also discourage account-sharing between friends and families (which might appeal to the profit-minded, but might frustrate customers). I believe that with a reasonable limit on the number of characters per account, the controls outlined in this article would support a reliable, character-specific rating system, but the right decision ultimately depends on developer preference.

If It Works, It Will Pay For Itself Many Times Over…

I understand concerns that a reputation system might drive away paying customers who get bad ratings, but the risk seems negligible to me. A system that makes the game more tolerable for everyone is ultimately going to encourage population growth, not sabotage it. Not everyone has the time to join and participate in a good guild, nor can everyone play the game when their guild-mates are readily available. There is PR value in a well-designed system as well (just imagine: “video games encourage kids to play nice!”) The bottom line: if it works for eBay, it can work for MMORPGs. And let's face it… if the system doesn't work, you can always scrap it. Worse things have happened in successful games.

_____________________________________________________

[David Edery is Associate Director for Special Projects at the MIT Comparative Media Studies Program.]