Friday, January 7, 2011

The problem of subjectivity in dungeon finder policing

Heroics haven't been a completely pleasant experience as of late, since not all players have been willing to adjust to the demands of the new design of heroics. As discontented people are wont to do, many people have been trying to think of ways to improve the heroic experience. Some of the ideas for ways to make the dungeon finder "suck less" have included systems for rating players; systems where you can "friend" players, i.e. mark them as good players and be more likely to be grouped with them in the future; and other ideas, most of which are based on the idea of players policing themselves.

The intentions behind these ideas are good, and in theory, they could work, but they all share the problem of having the potential to be abused by players. Some people will always look for a way to game the system if they can, and some people are just jerks, so no matter what Blizzard might do to try to prevent the system from being used as it wasn't intended to be used, someone will either benefit unfairly or be unfairly penalized. It's tempting to think the system could be designed to avoid that, but no design is foolproof. As an example, someone in the comments section of the WoW Insider post above had the idea to make it so that you can only mark players as bad players, which will increase their queue time. There would be no consequences for new players under this system because they would not be locked out of fast queues by a lack of up-ratings, which would be one of the worst consequences if players could be rated as good players, in addition to bad players. After all, up-rated players would be given shorter queue times (or else why up-rate players?), and if the up-rated players are being grouped together, that would give the newer players no one to group with.

But as another commenter pointed out, some players would just down-vote everyone they group with just because they can. Even worse, since down-votes would move the down-voted player farther down the queue, any down-rate you give another player would theoretically move you up the queue, assuming the down-rate isn't reciprocated. This would give players real, tangible motivation to down-rate people who don't deserve it; in other words, it wouldn't be just the jerkass players who down-rate everyone. In an attempt to moderate this kind of abuse, I had the idea of changing the system so you can still mark players as bad players based on their performance, but the more negative votes you give players, the less your negative votes will count against other players. This would discourage players from trying to game the system because the more they try to do so, the less of an effect they will have. Of course, players could abuse this system as well by just abstaining from down-rating players, which would give their votes more weight, and then using that weight to grief players they personally don't like by voting them down. Did another player win the roll for an item you wanted? If they're the only person you've ever down-rated, that could really increase their queue times. Isn't vengeance sweet?

Now, sure, Blizzard could tweak the formulas to prevent this kind of abuse, too, but the issues of players always looking for a way to game the system aside, this kind of system would still suffer from the issue of being a subjective system. A down-ranking system like the one described above depends on players down-ranking other players based on criteria shared by every other player, but people are subjective, and this might not be the case. Some players would down-rank players based on lack of skill, others would down-rank players for their attitudes, and some would probably only down-rank good players just to be ironic. Rather than a unified system for judging players, we would get players using their own standards to rate other players, standards which are not universal, and thus won't improve the dungeon finder for everyone.

What is needed is an objective system to help weed out the bad players (or at least increase their queue times). Some players have suggested that certain statistics be used to rank players, rather than a rating system. These statistics could include whether or not they use crowd control, whether they were healing efficiently, average healer mana when the tank pulled, etc. These are nice ideas, but they also have their own issues. What stats will we choose to use to judge a player? What if the player runs with a guild that chooses to run heroics in an unconventional way, a way that is successful but won't reflect well on their statistics? Or what if they consistently run with bad players and thus can't use crowd control, because their crowd control ends up being broken anyways? The fact is that it's very difficult to objectively measure player skill, no matter what statistics you use.

Another possibility is an idea I had a while ago, inspired by the achievement Ready for Raiding. What if a series of quests was set up to help players learn the skills they need to run heroics? These quests would be phased quests in which players would have simulated party members, and they would run through various typical heroic situations and train players in the skills they need to be successful in heroics. Tanks could learn to wait for the healer to regain his mana, how to hold threat against multiple targets, how to position targets, etc. DPS could learn how to crowd control, how to move out of the fire, how to prioritize kill order, etc. Healers could learn how to heal efficiently, how to best use their cooldowns, etc. All this could be done in a phased environment where the player would really be tested on whether or not they had the skills needed to run heroics. Now, players would probably feel uncomfortably forced if these quests were mandatory to run heroics, so perhaps they could be option and could reward some valor points to sweeten the deal, as well as an achievement--perhaps called "Ready for Heroics".

Or we could all just wait a few months until everyone out-gears heroics and they don't require skill to run anymore. It'll happen, it just needs time.

20 comments:

  1. Or, Blizzard could implement a system where they allow you to up- or down-vote other people without spelling out exactly what the consequence would be. Then, preferentially group you with people who up- or down-vote in the same way you do.

    They'd need a lot of data to work this out mathematically, and I'm not sure that the solution would be feasible to implement, but it sure would be neat if the people who downvoted everyone ended up grouping with people who downvoted everyone, and the people who up- and down- voted based on dps ended up grouping together. Everyone would probably end up happier for it (except the downvote everyone crowd, who would probably end up miserable).

    ReplyDelete
  2. "Everyone would probably end up happier for it (except the downvote everyone crowd, who would probably end up miserable)."

    I'd like to say that's a fair consequence for such negative behavior; after all, what better poetic justice for having an undesirable personality trait (thinking nothing but negative things about everyone else) than to be forced to be with people who share that trait? But if someone who once down-voted everyone learned their lesson and realized that they shouldn't down-vote everyone, they could potentially find themselves stuck being grouped with players who don't deserve to be up-voted for any reason. Because of this, they would have to up-vote in un-earnest ways in order to change their situation, which would probably lead to them having a history of rating players in ways that don't reflect how they actually feel about other players, which would lead to them being grouped with people who don't play like they do.

    There's also the issue that not everyone would rate players based on something that can be measures objectively. If someone down-rated players mostly for being jerks in chat, or for making racist/sexist/homophobic comments, how would the system detect that? Sure, one could potentially detect and measure homophobic comments by counting how often a player uses the word "gay", but what if the group gets into a legitimate discussion of something involving the gay community? Or what if a group gets in a discussion about the black dragonflight and, just for the sake of brevity, they refer to them as "the blacks"? Both of those situations could lead to those players' group-mates up-voting them for their intelligent discussion, only to find themselves later grouped with players who say that Vashj'ir is "gay" and say disparaging things about "the blacks". Sure, these scenarios aren't very likely, but any system that tries to punish bigoted behavior has the potential to unfairly punish people who aren't bigoted.

    Don't get me wrong. I really like your idea, when considered in theory; I wouldn't have taken the time to outline its potential problems if I didn't like it. And hell, I'd be quite happy if the system could detect that I rated players based on how eloquently they spoke (or typed), and then grouped me with players who do speak well. I'd love to be grouped with people based on the number of spelling and grammar mistakes they make, the frequency with which they use profanity, the brevity--or lack-thereof--of what they type, and the frequency with which they use acronyms like "lol". My playing experience would be much more pleasant if I were grouped with players who typed like me (few, almost never, great, never, respectively). But like I said, even a rating system like yours has the potential for unintended results and unfortunate consequences.

    Well now I've made myself start dreaming about a WoW are people are grouped together based on the criteria I outlined above. It's going to be a long fall from this fantasy when I next log in.

    ReplyDelete
  3. The LFG tool is as good as it's ever going to get. It should be left alone. Don't like running with bad players, then there's already a great solution in place - find good ones and run with them. That's what I do. It's not that hard and my heroic experience is great. Or there's this other tool called a vote.

    People looking for fixes of LFG tool are just lazy. Everyone complains bad players don't put time into learn their class, but here you are not putting time in to find good players to run with, or time to figure out how to use the tools you have. They already have consequences like not getting the shiniest gear as fast as the good ones. The reality is they pay their $15 a month just like everyone else, and that alone entitles them to the same LFG tool you have.

    I have absolutely no patience for impatient clowns who think they're "leet", always wanting to gogogogo before making sure everyone knows the fight, healer is mana'd up, everyone is buff'd etc. I've wiped far more from the impatient elitist jerks in the game than I have from a noob, and run into relative few "baddies." Never enough time to do it right the first time, always enough time to wipe over and over again.

    ReplyDelete
  4. I RDF into a guild running low level dungeons. First thing out of the healer "tank = terrible" when I don't pull the first mob as everyone is getting ready to start. Then i proceed to be punished with no heals and funny enough an entire group wipe. I think the healer thought he could heal the DPS while allowing me to die.

    After everyone was ready and I pull the first group after the wipe I get a load screen. My only solace was knowing I got my Avenging Shield to pull the group.

    Thought I'd share some pain. But the next group I tanked were amazing! So if you get some clowns. Drop group and move on. Don't let them ruin it. Leave them behind.

    ReplyDelete
  5. I have never had problems with 'bad' players (the mere term reeks something awful). Even if the player has less skill (another reeking term) or no idea about the mechanics, they can be told what to do, and from then on, it depends on their attitude whether they try to improve or don't.

    I think the more serious problem is with players' attitude. Consequently, the following suggestion for a downranking system is only partially tongue-in-cheek:

    - leaving party upon entering: -10 points
    - leaving party upon entering, if preceded by inspects of other players: -20 points/inspect
    - every 'wtf', 'omg' utterance in party chat or in /say: -20 points
    - every 'gogogo', 'stfu' utterance in party chat or in /say: -30 points
    - every use of the abominable character string 'noob' or any of its variants: -30 points
    - failing to greet other players upon entering the instance: -10 points
    - failing to thank other players and say goodbye to them upon leaving: -10 points

    And so on, I think you get the idea.

    ReplyDelete
  6. I agree with your assesment of the subjectivity of rating the players you encounter in LFD. Personally, I find all of these "quick fixes" a sign of how badly the system works. But I'm of an oldschool opinion ;)

    By the way, the quest style of "Ready for Heroics" you suggest exists in the MMO Guild Wars in the form of Henchmen.
    But I'd say WoW already teaches you how to do heroics in the form of normal dungeons and leveling dungeons, all going from very simple to more complex in heroics.

    ReplyDelete
  7. I think a the best idea I have heard so far is the cross-server friend's list. Put a limit on how many players you can add to this list, and only match if you are on their list and in queue at same time. This adds a bit of self-selection and mutual understanding to the process, and by capping the amount of people on your list, new players are not at such a disadvantage. As a tank I would love to add good dps and heals to a list of preffered players. If they liked my tanking and did the same, they may get faster q times if we are queueing at the same times of day.

    ReplyDelete
  8. The issue with statistics is that many players can play with Terminator-like efficiency and skill, but by the LIGHT you would never want to play with them again.

    I think the number of bad votes to good votes system is a good idea, one I suggested and stand by was that people who get extensive down voting just don't matter as much, and people who get voted up extensively matter more when voting.

    Disallowing guild intervoting would prevent most abuse there, I think.

    ReplyDelete
  9. There *IS* a built-in way that Blizz trains and teaches players the skills they need for end-game content. It's called leveling, questing, and doing lower-level dungeons. All of the mechanics in the higher level dungeons have "baby-steps" versions in the lower level dungeons. CC, LOS-pulls, targeting the tank's target, not standing in metaphorical fire, etc, etc...these can be found in various dungeons and lower-level raids.

    As far as learning one's class is concerned, that's something that should happen as one levels. As a holy priest, I have 14 different healing spells. (I could spec into a few others if I wanted to.) As I leveled, these spells were introduced slowly. I worked with "spell X", then "Spell Y", then how "spell X" works with "Spell Y" while also figuring out "Spell Z".

    The problem is that instead of taking the /played 60 days it took me to level (as holy to level 80), people spend much less time leveling. By the time they reach 85 they should know their class, but many don't.

    In short, Blizz gave us the tools, but a lot of playerbase no longer uses the tools and just smashes through.

    ReplyDelete
  10. Oh and as far as down-ranking or up-ranking players...

    I'd bet that most people would just look at the DPS meters to vote on DPS. So the hybrid class that stops DPS to help out the healer (which I do a lot in Shadow spec and it's almost always overlooked especially when I've been accused of being carried while needing to spend 1/2 my time healing), would just end up getting down-ranked.

    And, what happens if someone is a hybrid spec and they're a great tank, but weak DPS? Would that make the votes a wash or would good players get penalize for trying a new spec?

    ReplyDelete
  11. The issue with Heroics currently isn't the instances, mechanics, or anything that really has anything to do with something that Blizzard can fix. The biggest issue is with the attitudes of players.

    Yelling and screaming at someone whose DPS doesn't match your fantasy standard isn't going to make them a better player.

    ReplyDelete
  12. I think the whole "up-vote" and "down-vote" thing is misleading; they imply some form of absolute. Instead, if one considers them as relative, there are reasonable solutions. "I would like to group with this person again" and "I would not like to group with this person again" are clear statements - especially if "not like" is softer than the current cross-faction ignore.

    Once these bonds are in place, you can then potentially use the maths borrowed from trust networks to work out "clumps" of people who are likely to get on well with one another, and put them together. It's reasonably proof against gaming, too. If you "like" everyone you group with, you're more likely to see the same faces pop up again in the dungeon finder - assuming they like you, of course. If you "dislike" everyone, you'll tend to be grouped with people you haven't seen before (they won't necessarily be people who "dislike" everyone, as you're quite likely to have met those people and "disliked" them already).

    How does this affect queue times? I'm not sure whether it does; it certainly need not, as it can simply be used to control the grouping of (say) three adjacent groups. If there's a sufficiently high throughput of players entering the RDF, there's a higher chance of grouping with more people you like; if there's a low throughput such that groups are built based on availability, the situation is no worse than it is now. I don't necessarily see this as a problem.

    ReplyDelete
  13. I really, really like the idea of a "training quest" for dungeons and heroics. And not for other people; for ME. I'm a little shy, and I hate the idea of screwing up and getting yelled at. I mostly stay out of dungeons. Having a quest that would let me practice first, with npcs who won't be annoyed or inconvenienced while I'm getting the basics down, would be REALLY helpful to me, and go a long way towards boosting my confidence enough to actually get me running dungeons with real people.

    ReplyDelete
  14. This is a ridiculous discussion. Bad players never get better if they are never told how to get better but just have their queue times increased(Who are you going to eventually group them with?). Honestly if you can't deal with the short comings of your fellow man then stop playing MMOs. There are lots of single player games for you. Anyone even considering such a system obviously feels that they are awesome and therefore shouldn't have to wait with the noobs. If you don't want to wait, roll a tank.

    ReplyDelete
  15. @dejadrew

    I totally hear you on that one. Thats how I feel about running dungeons. I want to learn! Really! but I don't want to get screamed at for accidentally doing something wrong. Or because my dps is lower because I'm trying to pay attention what I'm casting/targeting so as not to pull aggro (for example).

    ReplyDelete
  16. Zani and Dejadrew - agree here, as well. For the last several years, I've run a guild that's designed to be supportive. People can come in, learn by running instances with guild members who've learned how to give constructive feedback rather than "loln00bl2p", and then decide what they want to do. Many have stuck with the guild, but we spawn raid guilds pretty regularly once folks feel they've learned to play and want new challenges.

    ReplyDelete
  17. Or again Bliz can increase the number of servers they choose to select people from in order to decrease waiting time. The more realms/battlegroups the less the waiting time.

    ReplyDelete
  18. There are no more good players, they are all grouped with others, anyone who logged out to take gramma to kung fu class lost the timming to find good players to group for heroics, face it.

    ReplyDelete
  19. I actually think that any system of trying to group people together based on some criteria would be a mistake because it would inevitably increase queue times, which no one wants. I think it would be really neat if there was an artificial intelligence trying to put us together based on our observed behaviour, but I don't think it's practical.

    Queuing for a heroic is like showing up to a pick-up game of a sport. Sometimes you end up on a bad team. If you only care about winning and don't have fun playing then it's just not an activity for you.

    ReplyDelete
  20. I'm with Zani and Dejadrew... I would love the chance to run through the instances in "practice-mode" first to learn the fights and group roles. You can learn your class by solo questing, but it really doesn't prepare you for group dynamics.

    I used to play with the most wonderful bunch of people (sadly they've all moved on to other servers) that must have been the blue-print of theozzardofwiz's guild (or inspired by him /her, whichever came first). I miss them all so much.

    And... Kesely's "rating" system would be a fantasy come true!

    ReplyDelete