Rep systems

 Posted by (Visited 23060 times)  Game talk
Feb 022006
 

Gamasutra -Soapbox – “Designing an MMORPG Feedback Rating System” is an article by David Edery on how we should make use of reputations systems in MMOs.

Dude, I’ve tried…! It’s a wonderful excuse to dump some accumulated knowledge on you.

Here’s some of the things that people don’t tend to realize in working in this space.

Negative feedback reputation systems have been shown mathematically to always spiral into chaos. The reason? People with bad reps just start up a new identity. Effectively, this means a free pass on bad behavior. And let’s not kid ourselves — there’s something called the “fundamental attribution error” by the psych types, which says that we mistakenly think that nobility and good behavior is inherent in people’s character. But it’s not. Given the opportunity, most people will do something that can be classified as “bad.” I know this is a kinda cynical thing to say, but it’s what the research shows.

Positive feedback reputation systems do work, as long as there’s an expectation of repeated interaction (in other words, if you think you might see the other person again). In a large enough population that isn’t segmented into cohorts or groups of appropriate size, this repeated interaction may not be a reasonable expectation. In other words, too mobile and big a group of players on your server, such as interacting with a different group of folks every night, and the incentives to get a good rep go away. (In fact, they might actually reverse: farm good rep in order to blow it on a big bad action).

Single-stage reputation systems are vulnerable to farming. This can be summarized by pointing to eBay. In a single-stage rep system, your “star” rating is based purely on how many folks give you a positive rating. The quality of their rep doesn’t matter. The result is that you can farm positive reputation pretty easily, because any new account can give positive transaction ratings.

In general, in a large enough system, farming becomes statistical noise. Of course, another way to look at it is that a guild that all rates its leader highly every day is technically not farming as far as the algorithm is concerned — they actually are showing their support.

In addition, in any environment where there is a low transaction cost, what starts to happen is the system norms to maximum. Any rating below the max is seen as negative feedback relative to the norm. This means that soon the ratings become meaningless — everyone is perfect. Over time, there’s an inflationary effect, because more and more ratings enter the system.

Of course, the core challenge here is finding a transaction to hook your reputation rating to. Many of the things that we would like to rate in MMOs, such as a random obnoxious chat remark, aren’t really peer-to-peer transactions. But giving reputation without a transaction (as in Cory Doctorow’s concept of “whuffie”) just exacerbates the problems listed above. So you have to hook reputation into specific actions you can hang a code trigger off of — rating a grouping session, or a purchase, or a murder.

More successful rating systems have either had a multi-stage system (such as Slashdot’s Meta-Moderation system) to act as a corrective, or like Advogato and other such systems, rely on authentication of reputation from trusted sources. This solves that guild farming issue.

This latter approach is very useful under certain conditions. The way it works is that when glancing at a rep rating, it displays to you the rep based only on sources that you yourself have rated. A high rating by someone you consider a jerk won’t show as a high rating to you. People your friends like will show as higher. Effectively, it’s rating mediated by social network. Unsurprising, because it actually derives from security models for untrusted peer-to-peer networks.

The downside, of course, is that if you aren’t hooked into a social network, you’ll effectively have no ratings. Since one of the primary values of a rating system to is provide cues to newbies as to the social landscape (“is this guy trying to scam me?”) this method fails that general usability test; instead, it works best for intermittent interactions in a large population of veteran users. Uh, something like maybe Eve Online might benefit from it.

This method can also get computationally expensive, of course, particularly depending on how far out through the network you search the data… does your friend’s friend’s friend’s friend’s rating count?

Ironically, a reputation system is fundamentally just about conveying data in a rapid abstracted manner. If your group of players is tight-knit enough — like say, a cohort of players levelling at the same rate through the same zones through their career — real world word of mouth is more efficient. This is a large part of why I believe that we have not seen the same sorts of misbehavior in the level-based systems that we have seen in the classless game systems; the latter do not tend to naturally create groups of a size smaller than Dunbar’s Number that are guaranteed repeated interaction, and thus peer pressure serves to police misbehavior. In other words, if you can get your players into clumps of 150 or less who tend to play mostly with each other, an abstracted rep system is effectively overkill at that point — and so is much of your customer service concern.

This argues that a coded reputation system is really intended for specific uses that your design may not call for. You really need it when you have a large population without repeated expectation of interaction, and you need it then for newbies. But the most secure methods won’t help, and if baddies can create newbies to do their dastardly deeds with, even those methods won’t help.

FWIW, UO went through most of the items described in the article on Gamasutra at one point or another, including fading of ratings, double-stage systems, recidivist tracking (causing permanent changes based on multiple movements between positive and negative) — and they all fell prey to the fact that it was a negative-feedback system. On SWG, we discussed trying a system like Advogato, and ended up concluding that word of mouth was what worked best anyhow.

One of the biggest problems, of course, is that what’s really needed is a lengthy background check, but we want a summary with just a five-star granularity. Enriching player profiles with past activities would go a very long way towards clarifying the meaning of the five-star rating, but even showing “rated 5/5 by 1 person” versus “4/5 by 127 people” is a bit too much data to parse in many circumstances.

I’m still a believer in reputation systems, but I think we’re going to need to look towards more robust mathematical models in order to really crack this particular nut.

  43 Responses to “Rep systems”

  1. Who Are You? What Do You Want? I was going to approach this slightly differently, but Raph made a post about social feedback

  2. Blogroll Joel on Software Raph Koster Sunny Walker Thoughts for Now Sex, Lies and Advertising

  3. Off the top of my head, what about a “laws” system, similar to what I believe LambdaMOO has.

    Laws would be proposed; the GMs, perhaps, would moderate that, and accounts would vote on it, and if it wins, say… an 80% majority, then it would be included. (It might complement an existing reputation system.)

    The laws would have to be mechanically implemented. You could have a law like “A person who has murdered a player sometime in the past month may not have a reputation higher than 60%.” This could be tracked and computed. You could have hooks all over the normal game code that plugs into a laws database for tracking anything and everything to optimize.

    So you can’t have a law of “A person who scams another player will have their reputation reduced to zero.” because the system doesn’t know how to figure that out.

    Just an off-the-cuff idea.

  4. On AmberMUSH, our advancement system worked strictly on the basis of reputation. Everyone was allowed to nominate other players for advancement (you would be erased from the nominee rolls that month, if we caught you nominating your own characters). However, a nomination wasn’t enough to get you anything.

    Once a month, one of the wizards would round up a secret panel of five random people from the player base. Each panel would be shown several slates of nominee names, and then would be replaced with a new random panel, until all the names were done (any alts of players on the panel would be cycled to the back, so another panel could vote on them).

    When a slate came up, the panel could discuss all of the names (usually five) on that list, and then they all secretly voted yea, nay, or abstain on each of the names. The character would lose nothing for nay votes, but could gain for yea votes. If a character had many abstentions, then the character would be cycled to the back, so another panel (who hopefully knew them better) would be able to vote on them.

    This worked very well, but it would, of course, be useless in a game that was large enough that most players didn’t know most players. Moreover, it would be useless as a mechanism for deciding who to group with, since if the general public knew you well enough that a random panel of them had any sort of opinion about your play behavior, then you wouldn’t need any sort of reputation system. So, I’m certainly not advocating a system like this for any other game.

    However, it does have one property that might be useful. That is, the system was extremely resistant to farming. Your improvement, such as it was, was mostly unaffected by the number of nominations you earned from the masses. This is because the opinion acquisition was strictly pull, rather than push. We asked for opinions, rather than accepting ones that were heaped on us.

    This would be akin to, say, if, after a group had been together for at least an hour, you confidentally polled each player in the group on their opinion of one (or more) other players in the group. (Yes, there are ways to game a system like that, but I’m just giving a rough example, for illustration.) You have the power to decide how often the game gives a damn about peoples’ opinions, and whose opinion it really cares about. You could use this for all manner of social engineering, depending on what you wanted going on in your game. For example, if you felt that guilds were becoming too isolationist, you could weight outside-guild votes heavier than inside-guild votes.

  5. The best answer for now is, i think, only allow entities which you (as the designer) know are reliable to assign reputations or assign modifiers to those reputations.

    Of course, in a persistant world with players coming and going pretty much at will, there is only one reliable type of entity.

  6. […] Farming bezeichnet das endlose Sammeln von bestimmten Ressourcen an den daf�r optimalen Stellen des Spiels. Oft um es dann bei Auktionsh�usern zu verkaufen.. […]

  7. From my experiences Reputation Systems don’t work with big numbers of users. They do not even work too well on websites like ebay. If your reputation gets too bad, you have no choice but to make a new account.

    But to tell a story of my own experience. We were once doing a website for the biggest petshop franchise in europe. The website contained a community with a forum.

    To motivate the users to make good forum postings, we thought of a positive reputation system similar to many other forums. But as we knew that most poeple don’t vote for reputation in forums, we added a bonus point system. Every vote gives you from 0 to 5 points for any of your postings. At the end of the month you get the bonus points for the current avarage voting added to your account.

    Then you could buy something from the shop for your bonus points or you could donate them for a pet shelter which could buy food for these points.

    Well, we expected that there would be some trouble with these points and ratings, but nobody expected what we got in reality.

    As you could donate the points for shelters, lots of people thought of a bad rating as stealing food from pets. We had so many threads about “Why do you give me a bad rating? I donate all my points to shelters and you are stealing them food!”-posts, I couldn’t count.

    Other than that – most realy helpfull postings tend to get a low rating. Especially if they aren’t what the user asking the question liked to hear as an answer. The more friends somebody made in the forums the better the votings get, even if he/she is just spamming.

    So my conclusion is, if the reputation isn’t important nobody cares. The more important it becomes, the more trouble and bitching about points you’ll get. Either way, it doesn’t work as intended and causes more trouble than it’s worth. I don’t think I will ever see a big MMORPG with a working reputation system. I for sure will stay away from ever using them again. They MAY work with small user numbers, where everybody knows each other – but then, why do you need them at all?

  8. Maybe a partial solution is a system in which there is a real opportunity cost for giving feedback. Every day you may only provide feedback once (positive or negative). Or suppose you could only provide feedback once about each other player, but you could change the feedback you left on someone when you wanted to. This might create an incentive to “make friends” with someone who you had previously annoyed, so they would change the feedback they had left on you, thus encouraging repeat interactions.

    I can already see at least one problem with this. Gangs of nefarious highway robbers would hold up helpless adventurers and force them to hand over goods or else face negative ratings from an entire guild. Reputation points would become a weapon.

  9. Is there really a need for this kind of system?

    If your group of players is tight-knit enough … real world word of mouth is more efficient

    Don’t players tend to congregate in groups of that size anyway (guilds)? There’s the 40 players whose reputations I know directly (by word of mouth or interactions), then the dozen guilds whose reputations I know.

    Any public ranking system incites competition and exploitation. Which can be good, it gets people into the game, fuels their achiever side. But that directly contradicts the point of ranking “rep” or “honor” or whatever you want to call it. You’re trying to rank the most “honest, good people” when the people at the top of any ranking system are those that do whatever it takes to get there.

    The only time there’s a concern is someone new, starting out alone. In which case I think an apprentice would be more effective that an aribtrary ranking system.

  10. Again, though, yes, but that doesn’t help people who are not plugged into a social network. The people most in need of a rep system are the people who don’t have any interaction history, who are coming to the situation cold. That’s why eBay needed a system: otherwise, how do you know that the transaction is trustworthy?

    I have extensive notes on the nature of “trust” that I need to turn into an essay at some point.

  11. I’m on the fence with regard to rep: on the one hand, IRL you have no idea what the “rep” of John Doe walking down the street is, so why should players in a game have a flag waving over their head (if that’s what we’re talking about here) or a scarlet letter on their chest. Shouldn’t reputation just pass by word of mouth or through first-hand experience? Then again, I kind of like the concept of karma that has been implemented to greater or lesser extents in some games. It can fit into the game “mythos” and can be a badge of honor (or notoriety for that matter). I just don’t know how you work so that it doesn’t wind up being a popularity contest where a player who is a complete ass can be flagged as highly reputable simply because he got all his guildmates to vote for him.

  12. Well, one reason why we don’t need rep as much in the real world is that there’s recourse for most interactions.

    One reason why we’re needing rep less in the games is that there’s less interactions. 😉

  13. One reason that EBay sort of works is that it only allows you to give stars after a success transaction. The users can fake transactions, of course, but in such a case they are at least giving EBay money for the fake transactions, and people who sell a lot won’t find it viable to do this. Also, of course, I think people are less likely to try and cheat with something that they perceive as more “real”.

    But what if you can only give stars to one person in your group after you group with that group long enough to obtain X experience points (where X varies by your level). When leaving the group you’d get a pop-up which offers you the opportunity to positively or negatively reinforce one player from the group. In such a case it seems like you might be able to do negative feedback too as no one you didn’t group with could blast you with negative points. This might drive those with lower reputations to group together, hopefully all trying to be good groupmates so that they could get positive stars to move up to where they could group with the higher star people.

    This would work for pickup group types of interactions and would be best with a system that would give you a comprehensive ability to look for other people looking to group together for certain tasks.

  14. But what if you can only give stars to one person in your group after you group with that group long enough to obtain X experience points (where X varies by your level). When leaving the group you’d get a pop-up which offers you the opportunity to positively or negatively reinforce one player from the group.

    I really like that idea. I guess there’s no real way to completely avoid abuse/griefing … where a player always blames his group for his own poor play so he rates them negatively. Although I suppose that would dilute with time (unless of course the other players really ARE that bad).

  15. An interesting post, but I’ve got several questions.

    1. You say that “Negative feedback reputation systems have been shown mathematically to always spiral into chaos.” I’d like to read more about this mathematical proof, where is it described?

    2. You claim that negative reputations are easy to lose: just create a new identity. The the real case of real games, that doesn’t seem quite as effective. Sure, a low level character might be relatively easy to recreate, but recreating a high level character is expensive in time.

    3. You suggest that there isn’t much incentive to build up a reputation if there isn’t an expectation of repeated interaction. Why not? Having a big “9 out of 10 people who took the time to vote said I’m a good guy” attached to me seems like an advantage; it may get me into a group or guild over someone without those ratings. eBay is a much, much larger reputation system with a positive feedback and people obsess over maintaining good ratings.

  16. On 1, I think the reference can be found in the bibliography to the Small Worlds presentation. I think that from memory, it’s in the work of economist Toshio Yamagishi, but I could be misremembering.

    On 2, I may have glossed this a bit too much in the essay. People who want to cause grief don’t necessarily invest in their character. But that’s also one reason why character advancement, particularly in cohorts, can help mitigate grief play.

    On 3, it’s not incentive to build up a good reputation; it’s incentive to treat a given interaction partner well. In eBay’s case, the reputation system itself makes EVERY interaction a repeated one. Imagine if there were infinite eBays — you’d have no incentive to build up a rep on just one of them.

    The psychology behind that can be found in research into things like “iterative tit-for-tat” strategies.

  17. […] He also responds to a Gamastura article entitled Designing an MMORPG Feedback Rating System, which was also discussed on Slashdot. […]

  18. Very informative! The idea of a one-number ratings system is flawed, but the idea of a multi-number ratings system is equally flawed. Instead, you might want to use a non-numeric approach.

    I made a quick post on that subject.

  19. Good post, Craig. I didn’t even get to the possibilities of tagging and folksonomies as ways to address the issue… good thoughts!

  20. If we’re simply discussing ways for noobs to be able to readily identify ‘good’ gamers from the miscreants upon entering the game world, then I’d have to go with Gabes idea.

    It’s simple, straightforward, and not prone to farming. Simply place a few restrictions upon it such as length of group session and also not allowing player A to give player B ‘rep’ twice in a given time period(day or days).

    It can be displayed as a simple 5 star scale made visible upon inspection.

    All the other ideas and different variations of ‘rep’ just seem a bit too convoluted to me and rather inferior to the good, old-fashioned, word-of-mouth approach which worked well in SWG.

  21. I’m struggling with this, as well. I wanted something that I had made work on my MUSH, but the MUSH is quite a lot smaller than an MMORPG.

    Fun Quotient:
    I never liked some of the arbitrary punishing on many MU*s, and I’d seen many +nom/+vote situations get abused. It took me a few months, but I made +noms (where players vote each other after RP) nearly worthless by themselves. The surprise was that players still do it, and enjoy getting the positive feedback quotes from other players. For each +nom, it’s worth 1 Cool Point, which then add up in a locked attribute on the player.

    Then there’s XP earned from being in plots, and awarded by the GMs who ran the plots. The XP build up in the locked attribute.

    There’s also Bonus points given by staff if a player does something above the call of duty. Players can +kudo a player who is running a plot. (Players can +kudo a staffer who does extra good work, as well.)

    Then there’s Demerits, which only staff can give on a player, after an investigation on behavior. Those are kept in a locked attribute.

    I then created the Fun Quotient, which is displayed in the bottom border of the +finger. This is Cool Points+XP+Kudos+Bonus Points – Demerits. Players who are more active and run more scenes and RP more have a higher FQ than those who don’t. It does work on the 20-30 playerbase.

    But how to implement this in an MMO? It would first have to give no true reward, aside from player-to-player positive feedback. That way, abuses will be few. There might have to be multiple reasons for the +nom available: Did the player organize a social event? Did the player entertain others or was helpful to newbies?

    What’s mainly missing in the MMO that you have in MU*s is the closer degree of interaction between staff/dev and players. I’m not sure the system will work in a large-scale game.

  22. You point out that deep linked rep systems work best. I agree – one simple rep system I’d like to see is the ability to, client side, rename other players. And then to share your database of renames with other players. Seeing: Fred [NinjaLooter!] show up is much more useful than Fred [1 star]. This also allows one to have duplicate names without people being able to assume others identities – the rename is attached to the characters UID, not name.

    With regard to multi-level numerical rankings, I don’t really think the lack of an initial seed is a problem so much as an opportunity. I believe that the operators of the game have a responsibility to identify and reward what they consider good play styles. What better way than having newbies default to being linked into the developer assigned “Good” network? As the newbies link to other players, there own links will be weighted as to soon overwhelm the developer approved links.

    The operator approved links can be changed to point to a small but broad cross section of champion players. The natural cascade effect will then give newbies instant access to those considered “good” by the developer, hopefully ensuring they get sent in the right direction to reinforce the norms of that server.

    I won’t say its easy. But I wouldn’t say that the lack of an initial seed is a fatal flaw of that technique. It just means that the operator has to choose which subgroups they think are “good”.

  23. One thing I see as a potential problem with the discussions in this thread is that we are only really talking about whether a player is a good groupmate or not. This certainly does not address what I believe tends to be the bigger problem in MMOs and that is griefers.

    Griefers I think are a huge issue in MMOs for doing various things in games too many to list here but we all know what they are. I certainly understand why we are all focusing on grouping, it is easier to tell… it is a transaction of sorts. But lets face it, finding out whether a player sucks in groups or not is one of the smaller worries in the game for a new player, after all a new player is likely going to suck in a group too! If developers decide to put a system into play but focus on the grouping part, I just assume they forget about it because this is not really a huge issue.. most people these days just group with guildies anyway.

    What really needs to be done is for some sort of system for real griefers and this is much more difficult obviously. Griefing takes such a wide variety of forms that it is going to be difficult to do this for obvious reasons. Slashdot and Advogato are certainly ahead of the curve but even these two are just scratching the surface on what needs to be done. We are probably at least another 5 years before someone gets close to making a doable system I think.

    One thing I think that would help MMOs in particular would be making it account-based instead of player based. This would be unfair to families sharing accounts, but it needs to happen eventually. It would also be nice if MMO companies could figure out a way to stop people from having multiple accounts. Like Raph mentioned, there needs to be a sense of accountability with the system and you don’t get that if the player can make 5 or 6 guys on each server without retribution. Perhaps the game could detect whether the single stars were on a different char and not count em AS against you but still count them against you so that if you had 10 1 stars on one char and 10 5 stars on a second char, maybe you’d have 3 and a half or 4 stars on the 5 star char instead. (well assuming you are averaging)

    I kind of like the idea of having one per day and I definitely think it’d have to be that you could only make a single rating on an account, but could change it at will if you wish. But perhaps as a safeguard to the guild problem you could make it so that, while you could certainly rate guildmates, they wouldn’t count towards totals of the player until they left your guild.

    The whole concept is flawed and may take a decent amount of work to get under way, but who knows what happens in the next few years… there are guys who sit there and do nothing but sit here and think about how to make these systems better. In all honesty, I think this type of system almost needs to be put into place in some form… yes it will likely get exploited in some fashion… it is hard to name a system in MMOs which is not, but it is something that I think would seriously benefit not only newbs, but everyone else who plays the game to have it. Accountability needs to be put into these games.

  24. Brask, I call those sorts of tagging systems “tickler files” because they work like the tickler files used by politicians: short-form personalized tagging that can be used to “tickle” one’s memory about past interactions.

    Exposing it completely to others does start creating a folksonomy, but folksonomies seem to be more useful for searching than for quick identification like reputation is (based on my experience with tagging systems on services such as Flickr).

    Razak, on the one per day thing — it’s an inflationary system then, though, isn’t it? How do you judge the relative weight of someone who has fewer points but different length of time in the game? Do you divide by character or account age? If it’s account-based, what happens with a newbie alt, does it reveal they aren’t actually a newb? Do you divide across all points given globally to make sure that the totals remain in a reasonable, legible value range? There’s lots of issues like these that start cropping up with any system like that.

    One possibility that I’ve tried to work with in the past is to give everyone a very limited number of votes to assign. So you’d get five thumbs ups, and you can MOVE them, but you never have more than five out there… I suspect that this scarcity might result in something more like a power law distribution, but I also think we’d see a pushback effect, where people would say “well, she’s great but she already has several good votes, so I will save mine for someone else who is deserving but hasn’t been recognized yet.” Worth trying out to see what happens, I suppose.

  25. I agree with Razak that we are talking about a way to emblazon griefers with scarlet letters. To that extent I like Brask’s idea of share-able labels client-side. Great idea! It doesn’t really matter if someone has a rep as a “great group player” because if we follow that line then that player will never be available to play with — so what does that knowledge then gain us?

  26. StGabe said:

    But what if you can only give stars to one person in your group after you group with that group long enough to obtain X experience points (where X varies by your level). When leaving the group you’d get a pop-up which offers you the opportunity to positively or negatively reinforce one player from the group.

    Did you skip my post completely, and then independently come up with almost exactly the same idea?

    This would be akin to, say, if, after a group had been together for at least an hour, you confidentally polled each player in the group on their opinion of one (or more) other players in the group.

    Jinx! 😉

  27. > People with bad reps just start up a new identity.

    Thus my suggestion that rating give/take be restricted to players of a given minimum level. More to the point, are there really that many level 40 (or 50, or 60) players that would abandon their character so easily?

    > you can farm positive reputation pretty easily, because any new account can give positive transaction ratings.

    Again, minimum level restrictions help address this, no? Not to mention my suggestion that ratings should potentially fade over time (and fade more rapidly if the rating-giver has been relatively inactive.) This would make reputation farming extremely tedious, since it would require leveling up new characters AND keeping them active.

    And, as suggested in my article, there’s always the (imperfect) option of locking reputation to account, not character.

    > FWIW, UO went through most of the items described in the article on Gamasutra at one point or another, including fading of ratings, double-stage systems, recidivist tracking (causing permanent changes based on multiple movements between positive and negative) — and they all fell prey to the fact that it was a negative-feedback system.

    But UO didn’t lock feedback to a meaningful transaction (i.e. a group session that has met certain explicit criteria for “legitimacy” — i.e. time together and/or mobs killed, etc). Or am I mistaken? I don’t fully understand why you believe that linkage to a transaction in this manner might cause more problems than it solves.

    In general, I’d love to hear more about how this stuff worked (and didn’t) in UO!

  28. I just realized that the “you” in my original post might be read as YOU, David, and it’s not what I meant. 🙂 I thought your article was quite nice, and I wasn’t trying to attack the methods you described, but rather just provide an infodump of stuff I’ve learned in trying to tackle similar problems.

    Rating/give take being restricted to higher levels doesn’t prevent farming, in the event of organzied guild farming; it just takes a bit more effort. You establish the higher level chars first, and then create new accounts for the characters to farm onto… I’ve never seen tedium stop this process, alas. 🙁

    UO locked its feedback to the meaningful transaction of a playerkill. 🙂 So yeah, there was a specific transaction. They faded only after very long lengths of time (as I said, it was a negative feedback system). People literally sat running macros for days in order to rack up the requisite online time. We then added the notion of “fast” and “slow” counts, I believe; I can’t remember exactly how that worked though. Finally, we did a “ping pong” count for recidivism, and made the result of accumulating too many ping-pongs be effectively permanent negative consequences.

  29. So do you (the collective) think rep should be a player-driven rating system or a world-driven karma system? Or both? I always liked in UO that your actions had a real bearing on your character, regardless if any other player ever saw you do anything. It was also honest (presuming that the game itself is not capable of griefing). I just feel like any time you leave it up to the players it becomes a popularity contest which opens the door to a very unpleasant player garnering an incongruously high reputation.

  30. > I just realized that the “you” in my original post might be read as YOU, David

    I didn’t take offense, either way. Thanks for taking the time to share your thoughts; it has made writing the article a much more thought-provoking experience. 🙂

    > You establish the higher level chars first, and then create new accounts for the characters to farm onto…

    By new accounts, do you mean new, full-fledged paying accounts? If not, what do you think about the idea of making ratings fade quickly when given by a (relatively) inactive character? That would seem to tremendously complicate any farming strategy. If so… wow!

    > UO locked its feedback to the meaningful transaction of a playerkill

    I think that linkage to grouping might perhaps prove more “meaningful” … at least in a game where group play is of paramount importance. Of course, the group would have to be judged “legitimate” by some standard (time together, and/or X mobs killed, or XP gained together – as some in this thread have suggested).

    Also, do you think that perhaps there was too much incentive to game the rep system in UO? If the rep system does nothing more than educate players about each other, perhaps players will be less willing to make the effort to break the system.

  31. By new accounts, do you mean new, full-fledged paying accounts?

    Yeah, I did. I basically feel that in a day where the mean number of accounts per user is 2.5 (even though the median is still 1 — the high end is extraordinarily high), you have to consider things that way.

    If not, what do you think about the idea of making ratings fade quickly when given by a (relatively) inactive character? That would seem to tremendously complicate any farming strategy. If so… wow!

    Activity is also a tricky thing. If it can mean logging in with an idle character who is kept alive via simple macros, then…

    I think that linkage to grouping might perhaps prove more “meaningful” … at least in a game where group play is of paramount importance.

    Well, yeah. The only reason why UO focused so much on playerkilling was the constant battle of reducing it. The very first pass was a “notoriety” system, basically like a karma score, that considered a whole host of actions against both PCs and NPCs. Blending these was a mistake, because you could use easily-repeated actions against NPCs to affect your score vis-a-vis other players. Quickly, the only facet in that score that mattered was the aggressive acts against other players; altruistic acts for other players were readily gamed.

    For similar reasons, I think “time together” is a very easily gamed metric. Even mobs killed can be farmed by tackling low cons. Shared XP gain seems best.

    do you think that perhaps there was too much incentive to game the rep system in UO?

    The incentive lies in the interactions. If we keep making games have fewer and fewer possible interactions, then the incentive decreases. In UO, the fact that the chief problem wqas a highly negative interaction just made it worse.

  32. Did you skip my post completely, and then independently come up with almost exactly the same idea?

    Pretty much. 🙂 Actually I did read most of your post before getting distracted by something or other (work is a pest, isn’t it?) and didn’t make it to the last paragraph. Then got to the same idea through thoughts of EBay.

    Must be a good idea if we both got there from different directions. 😉

  33. On the one per day thing… I was merely saying that I kind of liked a previous post. The idea of being able to comment on just one player a day creates a weight behind that comment. A player may be less likely to go around and comment people falsely if there is only one comment they may make a day. They are more likely to save that comment for when it matters, I’d wager for the negative unfortunately, but perhaps if someone is extremely helpful they’d use it there, or towards the end of their playtime they may also comment positively. It is intriguing to me, and I never previously thought about that. (I have previously thought of this subject before and have a paper written up somewhere on some of these ideas but who has time to find em=P) To be honest, I’m unsure if you weight a player ranking based on their own ranking… this is an issue for me as well that I have not resolved so I cannot answer your queries along these lines but it is an interesting thought line for me so I appreciate you sticking it in my mind=)

    For Account-based… I think the insinuation is that yes alts would be tagged as if they weren’t newbs because they would have rankings upon creation based on previous characters. I am inclined to say that any ratings a player would get are account based and not character based, my suggestion that perhaps a slightly degraded rating on a character was mostly a thought due to shared accounts. I feel an account-based systems biggest issue is in fact shared accounts which aren’t entirely uncommon and need to be thought about.. I don’t really have a solution to this, I was mostly arguing with myself there… if the system also had a one per day thing, I think that would mean that it would be one rating per account per day. So you couldn’t just log off and go to a different character to rate a player a second time. (I could see trading similar to the Asheron’s Call cross-server mule trades to get the higher allegiance rank if this was allowed).

  34. There are still several key aspects missing from any reputation system I’ve seen. One of the biggest is that a person’s reputation is monolithic: you can’t be “+10” (or whatever) with the Montegues and “-10” with the Capulets. In fact in some systems such ratings would average out to zero, which is wrong on several levels.

    I think we’ll get a lot further if a person’s reputation is not ssen as at attribute on them, but is instead something you believe about them (from direct or indirect experience). But even this doesn’t answer the newbie issue that Raph brings up: those who need robust reputation values the most are those who have the least context for understanding them.

  35. you can’t be “+10″ (or whatever) with the Montegues and “-10″ with the Capulets.

    So to really work then your rating would have to be reflective of the ratings of those who rated you? If you’re rated +10 by the Capulets, but the Capulets have been rated fairly low by the majority of players who have rated them, then your rating reflects downward in kind? I’d like that, but I can only imagine the cost of that.

  36. So to really work then your rating would have to be reflective of the ratings of those who rated you? If you’re rated +10 by the Capulets, but the Capulets have been rated fairly low by the majority of players who have rated them, then your rating reflects downward in kind? I’d like that, but I can only imagine the cost of that.

    Or, alternatively, you could use a real-player faction reputation system like the NPC faction reps. But, uh, that would probably be hideously impractical. 🙂

  37. […] Long story short, I’d like to know more before accepting any pronouncements. That said, I’d bet a few developers would be pleased to learn that social interaction is not so important, since accommodating social interaction tends to engender the thorniest design problems.    […]

  38. I get the feeling any kind of mechanical rep system could (therefore would) be disassembled, farmed and griefed by the player base. Putting any kind of power to mechanically affect others’ long-term play experience without the consent of the affected player just screams grief-play.

    While some groups of players are going to be annoying, etc. – ultimately I think it has to come down to players themselves forming groups. One man’s annoying is another’s funny, so a global scale isn’t going to be representative anyway.

    The problem of people being nasty isn’t one that I think we can fix with a simple algorithm. If it was, I suspect many governments would love to hear from us.

    Or maybe my faith in humanity is just lower than everyone elses?

Sorry, the comment form is closed at this time.