Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

A good grouping game must create good groups

13

Comments

  • WizardryWizardry Member LegendaryPosts: 16,536
    Having played the most dominant grouping game in FFXI i can say that i have seen every issue.

    The ONLY thing i will agree that is a problem is it will be tougher to get  a group as say that really odd dps class when compared to a Tank or Healer.It is not that big of a deal,yes i noticed for example in FFXI that early on it was tough for Dragoons and Dark knights but they also sort of earned that because of actions they did and made it tougher for a healer,i won't go into details as to why.

    The other issue is of course SHY people,i do understand it,they don't want to take the initiative to start a group if one doesn't come their way.However you also need to understand that if your going to enjoy "the game" you SHOULD have other things to do instead of just sitting there lfg.That is a whole other problem initself,some people think the ONLY thing you should do in a mmorpg is level and that is YOUR fault not the games.

    My point is that sure  their will be some superficial problems with a grouping game,but if you TRY to form a group of friends you can stay in touch with ,you will sort it out and learn to work together.That is really what it is all  about,you need to help each other rather than insult each other,so in reality the problem is PEOPLE NOT the mechanic of grouping.However if you give it a chance,you will eventually meet some good people willing to work together,not everyone is a jerk.






    Never forget 3 mile Island and never trust a government official or company spokesman.

  • WizardryWizardry Member LegendaryPosts: 16,536
    On the idea of not enough players,i will agree that is a problem,a game should NOT be designed to RAID that alienates too many people and causes too many problems.However we can yet again point the blame at WOW,that game has created this stigma that if yorur game claims it has RAIDING ,it equates to bigger numbers.

    To that i say,even if you think it does,the market can't all be in the same target area,it would never worl.So if your developer is foolish enough to copy the other 500 Wow clones doing nothing but 24/7 instancing,then your going to lose out.

    The market imo is long overdue for another FFXI type game,we haven't seen one since FFXI 13 years ago,so yeah.1 game going camps and 5000 going WOW clones is a really flooded market in one direction and diluted in the other,we simply need some intelligent devs to figure out that hey maybe we should compete where there is little competition.

    Never forget 3 mile Island and never trust a government official or company spokesman.

  • KyleranKyleran Member LegendaryPosts: 34,320
    waynejr2 said:
    Kyleran said:
    No, in the beginning there was tanking, DPS, Healing, Pulling, Buffing/debuffing, and crowd control.

    It got later dumbed down into the "Trinty."

    The Evolution of the Trinity

    http://www.youhaventlived.com/qblog/2010/QBlog190810A.html


    Bartle apparently never played DAOC.....Pity.

    "See normal people, I'm not one of them" | G-Easy & Big Sean

    "I need to finish" - Christian Wolff: The Accountant

    Just trying to live long enough to play a new, released MMORPG, playing POE at the moment.

    Fools find no pleasure in understanding, but delight in airing their own opinions. Pvbs 18:2, NIV

    Don't just play games, inhabit virtual worlds™

    "This is the most intelligent, well qualified and articulate response to a post I have ever seen on these forums. It's a shame most people here won't have the attention span to read past the second line." - Anon






  • WaterlilyWaterlily Member UncommonPosts: 3,105
    edited November 2015
    waynejr2 said:
    Kiyoris said:
    waynejr2 said:
    both of you are wrong, Trinity in EQ was Warrior / Cleric / Enchanter, it has nothing to do with DPS, nor with mud

    an EQ chanty is a pure CC class, not DPS

    it doesn't even have anything to do with gameplay really

    the only reason we started using the word "holy trinity", was because many people started to realize that a specific class combination of War, Clr and Enc made very stable groups, it was hard to take down a group where the tank had defensive (exclusive warrior skill, still is to this day), where you had the best healer in the game (cleric) and the best CC in the game (chanty)

    hits in EQ consist of base damange (or "damage bonus") + damage interval * (random dice 1-20)

    what warrior defensive does, is bring down the damange interval, which is the spike damage, since base damange is static

    that's why warriors were part of the holy trinity, several people knew how the AC softcap and damage worked in EQ, the more powerful content, the higher the DI become, while DB remained quite low

    the harder the content, the more important the holy trinity became

    I am going to say the origins of the trinity matter.  Plus, the person at the link has industry cred and certainly more than you do.
    Going to completely disagree. Kiyoris is right and I'll back him up on this. In Everquest holy trinity referred to Warrior, Cleric and Enchanter.  It didn't have anything to do with DPS. It didn't evolve out of MUD.

    I'll also disagree with the title of his article. It wasn't an evolution of any kind. I don't remember anyone using the word "trinity" or "holy trinity" early on in Everquest. It was quite a few expansions after the launch of EQ that that term started to be used. I don't remember that term being thrown around in Kunark.

    Whoever the "street cred guy" is in your link, he doesn't know what he's talking about. Just because someone has a blog doesn't mean he's correct. That person obviously never played EQ.
    Post edited by Waterlily on
  • cheyanecheyane Member EpicPosts: 6,408
    Anyone who played Everquest from 1999 knows about the enchanter,warrior and cleric trinity and as a wizard I did painfully too when I looked for groups. That person in the blog is talking about later use of the term but that is not how it was known in Everquest.
    image
  • nariusseldonnariusseldon Member EpicPosts: 27,772
    Wizardry said:


    My point is that sure  their will be some superficial problems with a grouping game,but if you TRY to form a group of friends you can stay in touch with ,you will sort it out and learn to work together.That is really what it is all  about,you need to help each other rather than insult each other,so in reality the problem is PEOPLE NOT the mechanic of grouping.However if you give it a chance,you will eventually meet some good people willing to work together,not everyone is a jerk.






    I have done that. Too much work for too little fun. I would much rather just PUG, and quit if i meet people i don't like. Others are free to do that to me. 
  • Flyte27Flyte27 Member RarePosts: 4,574
    cheyane said:
    Anyone who played Everquest from 1999 knows about the enchanter,warrior and cleric trinity and as a wizard I did painfully too when I looked for groups. That person in the blog is talking about later use of the term but that is not how it was known in Everquest.
    DPS was a fairly large group, but most classes could solo.  The Warrior, Cleric, and Rogue flat out couldn't.  They really needed to group.  The Rogue had it hardest as they needed to be in melee range and had no AoE.  If you had a high level Rogue you had dedication.  The same with a Ranger who didn't have anything to offer that a Druid or Tank class couldn't do better.  Ranger in Vanilla struggled to solo after a certain point and really wasn't wanted in most groups.  Melee DPS was no match for that of a Wizard or Mage.  That's kind of what made EQ fun though.  There was never and intention of this class or that class to be best or worst.  Players just found the best uses for each through experimentation.  It was great not to have classes that were so similar, but had true diversity.
  • QuizzicalQuizzical Member LegendaryPosts: 21,108
    edited November 2015
    mmoguy43 said:
    Wow, there are so many topics brought up and given solutions; I don't even know if this thread is for discussion. Some solutions may or many not work for many or any games, either.

    Auto-group matchmaking
    I also have wondered about enhancing match making so that it will have a tendency to match players you 'Liked' playing with more and less often match players you 'Disliked' playing with.
    Except the flaw is that It'd be rare for me to happen to play with the same player more than once.(If the population was low it would, but then again if the game was poorly received it wouldn't have existed) So the effort to put the feature in would largely be wasted and even more so if it was in an MMORPG instead of a MOBA.

    Yes, you don't repeat group with people very often.  But that's exactly why I want to rely on a machine learning approach to make an educated guess at who you'll like grouping with.  If you liked grouping with player A but not B, and neither A nor B are available right now but C and D are, and player C is statistically similar to A while D is similar to B, then it's more likely that you'll like grouping with player C than D.

    If everyone in the queue has been in a group a number of times, then even if no two people in the queue have ever actually grouped with each other, you've got enough data to get some much better than random guesses as to who will like grouping with each other.  You don't have to get perfect groups every time.  What I'm hoping for is something considerably better than random matches.
  • AntiquatedAntiquated Member RarePosts: 1,415
    "I want things to advance, not to go backwards."

    Wisdom.
  • nariusseldonnariusseldon Member EpicPosts: 27,772
    "I want things to advance, not to go backwards."

    Wisdom.
    Things have advanced. Games are more convenient, less dependent on others for our fun, you can match with anyone in the world, and kick/quit at a click on a button.
  • QuizzicalQuizzical Member LegendaryPosts: 21,108
    mmoguy43 said:
    The more details that need to be ironed out for the 'like' group system in order for it to function properly makes it sound like it's not worth the trouble to fix. It doesn't even close to the root of the problem being: an MMORPG that isn't mostly or entirely about grouping is going to have grouping and it's content suffer. 

    There will always be bad PUGs, even in MP only games like MOBAs, but answer has always been to find a guild or group of like minded gamers if you want a consistently good grouping experience.
    Even if a game is purely about grouping, if you can't find a group, that means you basically can't play the game.  What matters in terms of finding groups is how many people are looking for a group for the content you want to do, not the total size of the player base, or even the total number of players looking for groups for other content.  Now, if 90% of the player base is off doing solo content, that does leave less of a pool to group with than if everyone were trying to group but the game had no more group content than before.  But what directly matters is the size of that pool to group with.
  • WizardryWizardry Member LegendaryPosts: 16,536
    By content we mean what,linear questing or instancing?Bah crap gaming 101.

    All you need is grouping through camping and killing nothing more.Always people dropping from groups and the group needing replacements.Mentoring allows level barriers to be removed,so their really is no excuse if the game offers the idea behind grouping to be non stop.

    If your ONLY content for grouping is something. that takes a lot of organizing or waiting for a 24 man raid group,your wasting a lot of time.If all you need is a healer a tank and some 3-4 others,you really should not have too tough a time forming a group.

    The excuses i have always heard are usually BS,you can't tell me there is 5k people online and not several of each class to choose from.What i found is the people who were afraid to be proactive were the complainers.Well yeah imagine if the entire server was full of passive people,a group would never get started.

    That is why you need group reinforcement from very early levels so players become comfortable with it,you don't do like WOW and just group to Raid and  99% of the time everyone solos.It's like asking someone who has never skated before to play hockey.


    Never forget 3 mile Island and never trust a government official or company spokesman.

  • angus858angus858 Member UncommonPosts: 380

    As for finding people with similar play styles to group with, any game could easily add a "personality flag".  Alganon did this (yes that game did have one good idea).  If you are not familiar with the Bartle Test you can google that.  Nick Yee made an even better version specifically for mmorpgs.  When looking at any kind of "LFG" window you could see if a player identified themselves as an "A" achievement gamer or something else.  I would rather group with "E" exploration gamers regardless of their class, skills, or gear than "A" gamers, for example.

    This does not solve the problem of finding healers or tanks in a trinity game, but would go a long way toward solving some other issues, and is not difficult to implement.

  • nariusseldonnariusseldon Member EpicPosts: 27,772
    Wizardry said:


    The excuses i have always heard are usually BS,you can't tell me there is 5k people online and not several of each class to choose from.What i found is the people who were afraid to be proactive were the complainers.Well yeah imagine if the entire server was full of passive people,a group would never get started.



    Yes, i can.

    Just log on WOW and queue up as a DPS vs as a healer/tank and you will see the difference.

    No social grouping is as efficient as WOW LFD, and if you have to wait in a LFD system, there is an imbalance of classes. 
  • QuizzicalQuizzical Member LegendaryPosts: 21,108
    rounner said:

    Some flaws:

    • Teams of 2 or 3 rlf who do not want to be split because of others like ratings
    • The basic premise assumes there is a large pool of choices as opposed to 1 or 2 of an essential class on at a particular time for a particular instance
    • Flexible classes eg a druid who is dps but can heal if needed isn't necessarily specced to heal, or geared to go bear and tank or even practiced at that role


    Your first point is really easy to solve.  If players manually create part of a group to say, we really want to be in the same group together, you can just score possible groups such that there's a huge positive weight to keeping manual groups together.  That way, they're guaranteed to stay together, outside of weird corner cases such as most of the people in the queue being part of a manual group that is more than half of a full group (so that two such groups can't be combined) but less than a full group.

    For your second point, that's why there are alternate versions of dungeons built around different class compositions.  If there are only half enough healers for every group to have one, then half of the groups get an alternate version of the dungeon balanced around the assumption that the group will have no healer.

    What to do about flexible classes would depend on the specific details of the game.  In most cases, it can be handled just fine.
  • thinktank001thinktank001 Member UncommonPosts: 2,144
    edited November 2015
    I agree that scaling content around group size/members is part of the solution to making a successful grouping game, but I am not so sure about the rest.  I believe the only reason "group finder" exists is because developers build static content that doesn't allow for sub-optimal player groups.

    I think your "group finder" still has a lot of merit to it, but not in the context as a way to actively look for players to complete content.  It looks much more valuable as a way to help players with similar interests to connect with one another to form a community. 
  • KiyorisKiyoris Member RarePosts: 2,130
    edited November 2015
    We already know what makes a good group game.  You're like debating if the Earth is round or flat.

    All MMO vets already know what makes a good grouping game, all EQ players already know.

    Class interdependency. That means trinity gameplay like EQ and FF, not solo action gameplay like Vindictus or Black Desert. And especially not GW2.

    Well defined one-trick-pony classes. Limit hybrid classes. If you're a healer, you should do very low DPS, if you're a DPS, you should be a terrible tank.

    Difficult content. Content that requires group, CC, pulling,  strategy. No zerg rush content like most MMO of today.

    Limit soloing. Outside of a few specific classes, soloing should be extremely counterproductive.

    Slow travel and dangerous world. That means no insta ports or flying mounts, it means getting to know the people within your zone.

    Death penalty. That means a reasonable punishment for dying so grouping is not a random one-off event. You need to build a community and trust.

    No Autogrouping. No lazy autogrouping / instance port stupidity. Learn to get to know people.
  • KiyorisKiyoris Member RarePosts: 2,130
    edited November 2015
    Quizzical said:
     Rather fewer want a game where they spend half of their time trying to get a group so that they can do anything.
    It is down to the players if they can't get a group in a populated game, not the game.

    Not everyone succeeded in EQ, we had our share of people who couldn't get groups, those people eventually left for WoW where they could solo and get automatic dungeon invites.

    And that's fine, those people are much happier in a casual game. There are more casual choices than ever, WoW, Black Desert, GW2, all of those games allow for soloing and they have autogrouping mechanics.

    Some people can't ride a motorbike, they would fall off and should be kept as far away from them for their own safety, that doesn't mean something is wrong with motorbikes, it's just not for you. That's why there are casual MMO.
  • cameltosiscameltosis Member EpicPosts: 2,207
    Kiyoris said:

    Class interdependency. That means trinity gameplay like EQ and FF, not solo action gameplay like Vindictus or Black Desert. And especially not GW2.

    Well defined one-trick-pony classes. Limit hybrid classes. If you're a healer, you should do very low DPS, if you're a DPS, you should be a terrible tank.

    I'm going to disagree with these two points (sort of). I agree that class interdependency is needed, but that doesn't necessarily equate to using the holy trinity. I find the trinity quite dull so prefer more complex class types and have seen it work well. 


    With regards to well defined one-trick ponys, again, I disagree. I think flexibility is incredibly important for a good grouping game as it reduces on of the barriers for forming a group (over-reliance on any given class). However, I would say that each class should be unique in the way it plays and it's strengths and weaknesses. 



    I keep coming back to lotro for my examples, but honestly, it had the best combat system I've seen in an MMO so far. The main reason is because it didn't follow the holy trinity, it had a fourth spoke: support. There were 7 classes at launch: 1 tank, 1 healer, 2 dps and 3 support. Group size was 6, so typically one tank, one healer and four of anything else. 

    The addition of 3 dedicated support classes (1 buffer, 2 cc/debuffers) meant that group tactics played a much bigger role in determining success, but also meant much greater flexibility. For example, if you didn't have a tank online, you could bring a champion (aoe melee dps) instead. They would respec for tanking, but that would only bring them to about 50% effectiveness of a proper tank. However, you'd make it viable by bringing a burglar who would increase threat gen for the champion, or a loremaster who would debuff everything and CC extra adds, or a captain who could buff and off-heal to make up for lack of champs mitigation. 

    The flexibility of lotro's class system, along with the ability to play hybrids, meant it was much easier to form groups than in other mmos I've played. It meant that each raid boss had a ton of different strategies for success, rather than just one. Its why I die a little inside every time I see a new MMO following the trinity. Its just too unimaginative!
  • MendelMendel Member EpicPosts: 3,368
    Quizzical said:
    mmoguy43 said:
    Wow, there are so many topics brought up and given solutions; I don't even know if this thread is for discussion. Some solutions may or many not work for many or any games, either.

    Auto-group matchmaking
    I also have wondered about enhancing match making so that it will have a tendency to match players you 'Liked' playing with more and less often match players you 'Disliked' playing with.
    Except the flaw is that It'd be rare for me to happen to play with the same player more than once.(If the population was low it would, but then again if the game was poorly received it wouldn't have existed) So the effort to put the feature in would largely be wasted and even more so if it was in an MMORPG instead of a MOBA.

    Yes, you don't repeat group with people very often.  But that's exactly why I want to rely on a machine learning approach to make an educated guess at who you'll like grouping with.  If you liked grouping with player A but not B, and neither A nor B are available right now but C and D are, and player C is statistically similar to A while D is similar to B, then it's more likely that you'll like grouping with player C than D.

    If everyone in the queue has been in a group a number of times, then even if no two people in the queue have ever actually grouped with each other, you've got enough data to get some much better than random guesses as to who will like grouping with each other.  You don't have to get perfect groups every time.  What I'm hoping for is something considerably better than random matches.
    I like the basic premise of creating a tool that relies on 'machine learning' to build groups, but I think that is going to require a rigorous method for evaluating play styles.  As an enchanter (in the Kunark era) who knew when his (single-target) buffs were going to expire, who managed to communicate (a lot), and entertained the group, how is that distinguished from the enchanter who does nothing else but the class role?  What happens if I'm in a crowd control situation, and have bad luck on resists that day, versus the same enchanter whose identical spell lands every time?

    I think there's too much variability in the factors that influence 'like' to build an effective grouping tool.  Too much difficulty in attaching 'statistical validity' to 'behavior'.  Too costly and time-consuming to try to build.  But I applaud your efforts to make such an tool.  Maybe you will be remembered for an important innovation.

    Logic, my dear, merely enables one to be wrong with great authority.

  • NeanderthalNeanderthal Member RarePosts: 1,797
    angus858 said:

    As for finding people with similar play styles to group with, any game could easily add a "personality flag".  Alganon did this (yes that game did have one good idea).  If you are not familiar with the Bartle Test you can google that.  Nick Yee made an even better version specifically for mmorpgs.  When looking at any kind of "LFG" window you could see if a player identified themselves as an "A" achievement gamer or something else.  I would rather group with "E" exploration gamers regardless of their class, skills, or gear than "A" gamers, for example.

    This does not solve the problem of finding healers or tanks in a trinity game, but would go a long way toward solving some other issues, and is not difficult to implement.


    I was thinking something similar to this.  You wouldn't need to have the game track player behavior.  It would be much simpler to just have players fill out something like a dating profile to match them with other like minded players.  Questions could include things like:  "how much do you like to chat with your group" and "Do you prefer to finish content as quickly as possible".  Maybe give a scale of 1 to 10 to rate those things and then the game looks for the closest matches to you.

    Now for some general rambling.  The early days of EQ was the best grouping game I've ever played.  It worked in large part because there were usually lots of other players around in any given area and because you could solo while LFG.

    You would just log in, go where you wanted to go, start asking for a group and if you didn't get one right away you would solo while occasionally announcing in zone-wide chat that you are LFG.  Admittedly there were times when you simply didn't have any luck and never did get a group but it usually worked out.  Either an existing group would invite you or you would see another person soloing while LFG and the two of you would join together to duo and then invite more people as you went along.

    I really loved EQ in those days.  You didn't have to take your own group, you just showed up and joined with the other people there.  You didn't have to play by someone else's schedule in order to have a group you just logged in when you wanted to and went where you wanted to.  It all started to deteriorate though as the game progressed and people started self-segregating along guild lines, not wanting to group outside their guilds.  Also, players began to be spread out too much so that you couldn't always expect there to be a lot of others around.

    It gradually became the case more and more that you had to take your own group with you if you wanted to be sure to have a group.  It all became less open and simple and became a lot more of a hassle.

    I rather doubt there will ever be a game that recaptures the feel of early EQ from the low to mid-levels but I hope for it because I loved those days and it's really the only reason I got into MMOs at all.

  • mmoguy43mmoguy43 Member UncommonPosts: 2,770
    Mendel said:
    Quizzical said:
    mmoguy43 said:
    Wow, there are so many topics brought up and given solutions; I don't even know if this thread is for discussion. Some solutions may or many not work for many or any games, either.

    Auto-group matchmaking
    I also have wondered about enhancing match making so that it will have a tendency to match players you 'Liked' playing with more and less often match players you 'Disliked' playing with.
    Except the flaw is that It'd be rare for me to happen to play with the same player more than once.(If the population was low it would, but then again if the game was poorly received it wouldn't have existed) So the effort to put the feature in would largely be wasted and even more so if it was in an MMORPG instead of a MOBA.

    Yes, you don't repeat group with people very often.  But that's exactly why I want to rely on a machine learning approach to make an educated guess at who you'll like grouping with.  If you liked grouping with player A but not B, and neither A nor B are available right now but C and D are, and player C is statistically similar to A while D is similar to B, then it's more likely that you'll like grouping with player C than D.

    If everyone in the queue has been in a group a number of times, then even if no two people in the queue have ever actually grouped with each other, you've got enough data to get some much better than random guesses as to who will like grouping with each other.  You don't have to get perfect groups every time.  What I'm hoping for is something considerably better than random matches.
    I like the basic premise of creating a tool that relies on 'machine learning' to build groups, but I think that is going to require a rigorous method for evaluating play styles.  As an enchanter (in the Kunark era) who knew when his (single-target) buffs were going to expire, who managed to communicate (a lot), and entertained the group, how is that distinguished from the enchanter who does nothing else but the class role?  What happens if I'm in a crowd control situation, and have bad luck on resists that day, versus the same enchanter whose identical spell lands every time?

    I think there's too much variability in the factors that influence 'like' to build an effective grouping tool.  Too much difficulty in attaching 'statistical validity' to 'behavior'.  Too costly and time-consuming to try to build.  But I applaud your efforts to make such an tool.  Maybe you will be remembered for an important innovation.
    Yeah, that's why I came to the conclusion that it doesn't seem worth it vs just making the game a purely or much better group focused game.
    It may actually succeed at giving like minded players that are good at grouping an even better grouping experience. However with many holes, it could also give new players and unfortunate ones a consistently bad grouping experience.

    If I'm very good and experienced with grouping in past games but just started with this one, do I get no chance to group with vet players? If no, then I get matched with bad players for how long until I get an amazing group?

    If I'm a new player and not very good in groups yet but would like to play more and get better, do I have to suffer with always getting matched with bad players?

    If I start playing and am not social but want to be, how can some algorithm take that into account?


    Alternate versions of dungeons
    This would be worth exploring. I don't know how you'd plan to have this, whether it's explicit (this class composition get this dungeon mob composition) or dynamic. Mob scaling seems to work with groups but it can water down the value of progression. In a dynamic approach it could also work in open world group encounters, instead of only in dungeons. So you'd have groups of mobs but depending on the player group that attacks it, it would grow in strength to match. For example, a group with no tank or healers; no change to the mob encounter. But if it is a player group of tanks and healers or other class composition; the mob encounter would reinforce to counter the players by scaling the difficulty up to match it. I personally like mobs getting called in over simply bolstering mob stats.
  • QuizzicalQuizzical Member LegendaryPosts: 21,108
    Mendel said:
    Quizzical said:
    mmoguy43 said:
    Wow, there are so many topics brought up and given solutions; I don't even know if this thread is for discussion. Some solutions may or many not work for many or any games, either.

    Auto-group matchmaking
    I also have wondered about enhancing match making so that it will have a tendency to match players you 'Liked' playing with more and less often match players you 'Disliked' playing with.
    Except the flaw is that It'd be rare for me to happen to play with the same player more than once.(If the population was low it would, but then again if the game was poorly received it wouldn't have existed) So the effort to put the feature in would largely be wasted and even more so if it was in an MMORPG instead of a MOBA.

    Yes, you don't repeat group with people very often.  But that's exactly why I want to rely on a machine learning approach to make an educated guess at who you'll like grouping with.  If you liked grouping with player A but not B, and neither A nor B are available right now but C and D are, and player C is statistically similar to A while D is similar to B, then it's more likely that you'll like grouping with player C than D.

    If everyone in the queue has been in a group a number of times, then even if no two people in the queue have ever actually grouped with each other, you've got enough data to get some much better than random guesses as to who will like grouping with each other.  You don't have to get perfect groups every time.  What I'm hoping for is something considerably better than random matches.
    I like the basic premise of creating a tool that relies on 'machine learning' to build groups, but I think that is going to require a rigorous method for evaluating play styles.  As an enchanter (in the Kunark era) who knew when his (single-target) buffs were going to expire, who managed to communicate (a lot), and entertained the group, how is that distinguished from the enchanter who does nothing else but the class role?  What happens if I'm in a crowd control situation, and have bad luck on resists that day, versus the same enchanter whose identical spell lands every time?

    I think there's too much variability in the factors that influence 'like' to build an effective grouping tool.  Too much difficulty in attaching 'statistical validity' to 'behavior'.  Too costly and time-consuming to try to build.  But I applaud your efforts to make such an tool.  Maybe you will be remembered for an important innovation.
    Statistical noise doesn't actually break the system at all, though it does mean you need larger sample sizes to get good results.  For example, take the results of an NBA season, pick 10% or 20% of the games at random, and replace the real-life result by a coin flip to determine the "winner".  This won't change the final standings very much.  It might change things so that if team A finished 1 game above team B in the real-life standings, the order is reversed in your redone standings.  But it's not going to make a championship contender look like a lottery team or vice versa.

    If without any statistical noise, a system would have said you have a 65% chance of liking to group with player A and a 64% chance for player B, that's effectively a tie.  Even if some noise in the system reverses the order, so what?  They're basically tied anyway.  If a good system would have found that you had a 90% chance of liking grouping with A and 10% for player B, maybe some statistical noise changes those numbers to 84% and 13%, respectively, but so what?  It can still tell that it's far better off pairing you with player A than player B.

    A matchmaking system doesn't need to find the best possible partition of players into groups.  It just needs to find a pretty good one.

    Random noise isn't a problem, though systematic biases to the system can be.  But that's why you'll have to constantly refine the grouping model, not just pick something that seems reasonable once and let it run forever.

    If the question is distinguishing players who communicate from those who don't, that should be pretty easy to measure, unless you're using some program external to the game for communication.  It's basically trivial to measure how much someone types in chat.  It's not terribly difficult to score whether the text looks like real words.  It is harder to distinguish word salad (with real, correctly spelled words!) from meaningful sentences, but how often is that a problem in MMORPGs?
  • nariusseldonnariusseldon Member EpicPosts: 27,772
    Kiyoris said:

    Some people can't ride a motorbike, they would fall off and should be kept as far away from them for their own safety, that doesn't mean something is wrong with motorbikes, it's just not for you. That's why there are casual MMO.
    But the question is whether there are enough of those who can (or want to) ride motorbikes. 

    Motorbikes have very little value (i.e. what is wrong with them) if there are not enough people who like them. 
  • QuizzicalQuizzical Member LegendaryPosts: 21,108
    rounner said:

    Some flaws:

    • Teams of 2 or 3 rlf who do not want to be split because of others like ratings
    • The basic premise assumes there is a large pool of choices as opposed to 1 or 2 of an essential class on at a particular time for a particular instance
    • Flexible classes eg a druid who is dps but can heal if needed isn't necessarily specced to heal, or geared to go bear and tank or even practiced at that role


    Essential classes seem to be the issue most folks have.

    So, start there. "Essential Classes" need to go.

    Turns out it's as easy as "every class has some healing capability", and let agro wander where it will. Works better if "every class has some way to dump or evade too much agro," but that isn't truly necesary with a well-managed party that looks out for each other.

    The "I'm the superstar" players will whine about no longer being center of attention, but that's just an ego thing.

    But GW2 missed the boat with all-dps groups, the most boringly familiar way to "dump the trinity".

    A tightly working group of dark and/or rad defenders/controllers was soooo much more fun to play. Tank? Healer? DPS? No thanks, no need, we gots the legendary tankmages.
    Making it so that any arbitrary set of players can group fine regardless of class is one way to get around required class problems.  But some games want to have combat mechanics that really don't allow that.  That's why I proposed an alternate solution for those games:  make alternate versions of dungeons (same map, different mob spawns) that assume different class composition.
Sign In or Register to comment.