Psychochild's Blog

A developer's musings on game development and writing.

26 February, 2014

Extreme Makeover: Random grouping
Filed under: — Psychochild @ 3:42 PM

I did a phone interview recently where I discussed ideas on how to improve random grouping. What I’m calling “random grouping” is perhaps best known from WoW’s Dungeon Finder (Looking for Dungeon or LFD) and Raid Finder (Looking for Raid or LFR) systems, where you indicate you want to run a dungeon and you are randomly matched up with other people. I share the common lament that random grouping systems don’t help players form social bonds within the game even if they would like to do so.

So, let me put my designer hat on and describe a system to allow for more socialization opportunities.

Why socialization matters

Even though MMOs are multiplayer, people tend to be wary of socialization. The whispered horror stories of “forced grouping” in older games have spooked people. Everyone has to pretend they are so busy today (although not so busy that they have to give up MMOs) to the point where making a social connection in a game is too onerous.

But, from a design point of view socialization is very desirable in an MMO. Players becoming part of the social fabric of the game become important touchstones for other players. Friendships forged in a game can improve the player’s life, as well as give them additional ties to the game. Especially for an older game, retention starts to become very important as the “newbie hose” slows down and fewer people replace those that leave the game. Social features can also encourage a returning player to stay with the game longer if they are able to make new connections within the game.

The problems with random grouping

Random grouping is added to the game as a way for people to see additional, group-required content within the game. Players who did not have existing friends or who did not build connections in the game while gaining levels have a hard time getting into dungeons. Waiting around for a pick-up group (PUG) to decide to do a dungeon can eat precious limited gaming time. Random grouping systems allow a player to queue up and do a dungeon conveniently. Of course, once you add a convenience like this to a game then the majority will rely on that system instead of other, less convenience methods.

Random grouping tends not to encourage socialization. The first problem is that there is little reason to invest in a long-term relationship with other random group members. Dungeon finder systems are designed to have a large pool of players available to pull from to reduce waiting times. But, this means that it pulls from a pool spread out across multiple servers and there’s a very slim chance you’ll ever meet the other people again if the only way to interact is via another encounter in the random grouping systems. The other problem is that the type of person who is interested in random grouping but who does not have friends to group with is probably mostly concerned with acquiring rewards. And, in a heavily achievement-focused game, the goal is to acquire those rewards with maximum efficiency. A focus on efficiency means that pleasantries like socialization are impediments rather than desired outcomes. This also causes problems if people don’t speak up because they don’t understand some mechanic necessary for a fight.

The last major problem is that dungeon content usually places pressure on players. One person making a mistake is likely to cause problems for everyone, probably a group wipe. This leads to people only seeing others as obstacles rather than as trusted group members. Most social-minded player are going to want to focus on performance than on socialization for fear of harming everyone else’s experience if they should make a mistake.

Goals of a new system

So, let’s assume we want to allow people to socialize with random group members more for the obvious benefits. What are our assumptions and goals?

The main assumption is that we don’t want to radically change dungeon and raid content just for this system. Smart MMO designers use data to create content that people enjoy, so trying to add yet another factor on top of that will introduce more complications. If the current content is engaging enough for players to want to play, we probably don’t want to tinker with it that much.

Let’s list out some goals for our system to increase socialization.

  • The system should not “force” people to participate in it. Although, we may encourage people to give it a try.
  • People should be able to find and keep track of other people they enjoyed grouping with.
  • We want to improve overall behavior in the random groups as well as allowing friendships to blossom.
  • In a generally achiever-focused game, we want to give some sort of reward, but we want to avoid people exploiting the system to make it an efficient way to farm.
  • Socialization should be able to happen outside of the context of the random grouping. People may want to focus more on performance in the group and socialize later.

That looks like a sufficient list of goals for now. So, what kind of system would work?

Inspiration from other games

Although I’m not a huge fan, I’ve paid attention to social games and mobile games. Social games in particular encouraged interaction, although that was mostly intended to get people to tell others about the game to increase the number of players. Some of the designs were based on sound foundations. One example is “gifting”, where you get a reward that you personally cannot use. You are encouraged to give it to someone else. Usually this takes the form of a present you cannot open yourself. You don’t have to give this item out, but if you do then someone else may give you a gift as well; you are encouraged to use it not waste it.

This system works within the concept of reciprocity. When someone gives us a gift or does us a favor, we feel compelled to respond in kind. It’s a way humans have learned to cooperate and build stronger bonds. Hmm, that sounds useful!

So, how can we use this concept of gifting in a random group?

Dungeon Rewards

Let me spec out a design summary for a system that uses gifting. I’ll throw out a few numbers here, but of course these might need to be adjusted through playtesting and observation.

My proposed system is that you let people give rewards to other people after a dungeon has ended. A dungeon will usually end in one of two scenarios: either finishing the last boss, or when someone leaves the group after a wipe. In these two cases, we want to give people a way to reward others who were helpful to the group. So, the system will have a quick pop up when the end of a dungeon is detected, and you can optionally give one other player a “pip”. The only person who knows who you gave a pip to is the person who received the pip. As you receive pips you get rewards. The rewards you get depend on a few factors: if the player has pipped you before and if the dungeon was successful or not.

The restriction on pipping the same person repeatedly is a first step in addressing exploits. We want to avoid situations where two friends or guildmates who queue up together just pip each other for an extra reward. The goal is to have people give rewards to people who are not already in their social circles. We might further restrict it so you can’ give pips to existing friends, guildmates, or people you were in a group with when you queued up.

The reward would be small for a successful completion. Anyone who gets a pip might get a temporary buff. Something nice, but not overpowering. Examples: a 5% bonus to experience gain for 2 hours, or a small faction bonus related to the dungeon. If the only person who pips you has pipped you before, you might on get 25% of the full reward.

If the dungeon is unsuccessful, then we give a bigger reward because we want to reward people who help others in the group. Obviously the reward would not be larger than the reward for completing the dungeon, otherwise it would lead to farming via intentional failure and that would be undesirable. Let’s say a successful dungeon gives 10 loot badges; after a failed run you get 1 badge per pip from someone who has never pipped you before. You can only earn a maximum of 2/3rds the typical reward from failure before you succeed again, so up to 6 in this case. However, these badges would count against future rewards from the dungeon, so if you got 4 badges from pips you would only get 6 badges the next time you successfully complete a dungeon run. Note that if you are the person who leaves the group and causes the failure, you cannot give or receive any pips.

We want people to pick out people they enjoyed playing with, and to reward those people who put in more effort. The person who explained encounters instead of just rushing in and wiping should be rewarded!

Dungeon Friend List

Anyone you pip is put on a personal list. When you queue up for the dungeon finder, you the system tries to match you up with people you have pipped before. This means people that you will be grouped more often with people you like, and with the people they like. This improves your experience by grouping you with good people, but also lets you “run into” those good people more often, leading to more opportunities for social interaction. The game would alert you when you are grouped with someone you pipped before, so that you recognize that the player is someone you enjoyed playing with before and giving something to talk about. This also has a side effect of reducing some abuse, as you’ll be more likely to be with people you’ve pipped before, and they’ll get diminished rewards if you simply pip them again.

If two players pip each other, then they get added to a supplemental friends list. This works like your normal friends list: you can see if the person is logged in, etc. But, this gives you a way to reach out to someone you grouped with before when not in the dungeon worrying about performance. You might be sitting around in a city and take a look at your dungeon friends list and find someone to chat with when you aren’t under pressure. This could lead to opportunities to make connections, plan new dungeon runs, recruit them into guilds, and other positive social interaction.

Pitfalls and problems

This is just an initial system, inspired by a phone call. There are likely problems I haven’t anticipated. But, here are a few things I would be wary of as the system develops.

One might suggest eliminating rewards and give achievements or some score. The problem is that people could turn that into another “gearscore” situation, where people expect you to have a certain minimum “pip score” otherwise they’ll just vote to kick you. This makes it harder for new people to get started in random grouping.

I’d also recommend avoiding “negative” pips. The first problem is that griefers could use this as a tool to cause pain. Another issue is that we want to improve behavior, and negative reputation tends to lead to negative behavior as people try to get retribution. And drastically bad behavior should be handled by customer service. And really, we want to avoid the abuses of reputation systems as much as possible.

It might also be worth restricting how often someone can give a pip. That way people an avoid awkward situations where someone demands a pip. “Sorry, I already gave a pip today” is a little white lie that can defuse a situation.

The last problem is farming. I could see a situation where someone only needing 1 badge for a reward might join a group and try to secretly sabotage it in order to get someone to leave and hope to get a pip. I suspect this might be simply too difficult to accomplish, and we have to accept we can’t fix every exploit, but it is something to pay attention to.

Making friends randomly

I think this system would help improve the attitude in random grouping situations. It gives people some incentive for good behavior that goes above and beyond the minimum required to complete a dungeon. There are some risks, but overall I think this system would have minimal negative impact on the experience, and lead to some better experiences.

What do you think? Do you think this system would help random grouping? What is your suggestion to improve this system, or random grouping in general?


  1. Have you seen FFXIV’s Player Commendation system? It has the start of what you suggest.

    My take on the system:

    My issue with the rest of your system is that I am not sure that clustering people who commend each other is actually a desirable property for a random system. Like will congregate with like, meaning that strong players are much more likely to be matched up with other strong players.

    In reality, what you want is to distribute the strong players out. Some players are strong enough to carry weaker players. Adding that strong player to a below-average group makes success more probable, and thus more people have a better experience.

    So that’s the tack I’d take with random groups. Identify strong players, and use them to shore up weak groups. I posted an idea like this a couple years ago:

    Comment by RohanV — 26 February, 2014 @ 3:59 PM

  2. That’s my concern with the system as well. It sounds great on paper, but I’m wondering if an end result is that players who favor fast, efficient clears will only pip other players that produce those dungeon results.

    This system will end up grouping them together, which in a way, is great because everyone’s objective will match more closely, but I wonder if that grouping has any effect on the rest of the population.

    Basically, are players who favor fast clears the only “strong” players around?

    The common mindset is that there’s a correlation. Will there be enough patient and social-favoring “strong players” left to mix in with the general populace and teach? Will the separation of a subset of fast, efficient clearers make the dungeon experience better or poorer for the rest? On one hand, there might be less social abuse hurled around, and reducing toxicity always makes the experience more enjoyable. On the other hand, success rate and time to clear a dungeon may be poorer than the norm now.

    The other potential issue is that this may reduce the unpredictability of the PUG experience, which again is what many players SAY they want, but would that in fact turn dungeon clears into a boring repetitive experience that players get tired of sooner? Part of the insanity of PUGing is that the worse and weirder the group, the more unique the resultant story of the experience becomes, even if it was a nightmare while doing it. The smoother the group, the less memorable it becomes – we came, we saw, we cleared. Next!

    Comment by Jeromai — 26 February, 2014 @ 4:35 PM

  3. I would hesitate to make the correlation between “strong” and “speedy.” That is, those pushing for an efficient run are not necessarily the strongest players, and may be more likely to cut and run after a wipe rather than sticking around to help the group. If the goal is to increase the number of successful random groups, then by all means distribute the pipped players away from each other so they can carry weaker teams. But if you group people who “like” each other more often, the social bonds between them strengthen, increasing the social stickiness of the game, which I believe is the point of Psychochild’s proposal.

    Comment by rowan — 26 February, 2014 @ 4:56 PM

  4. I was just about to mention the FFXIV example Rohan gave, but as a con rather than a pro. According to at least one podcast I listen to, players routinely leave the dungeon immediately, even though “voting” costs them nothing and potentially gets them rewards. I hear similar laments from SWTOR PVP, which has a voting mechanic of some sort. That said, you can’t force people to vote given how much other stuff you’re tying into the voting process, so I don’t think this is worth trying to fix.

    I maintain that the problem is rewarding participants above and beyond the rewards for completing the content. As long as this is the design, expediency is not a bug, it’s the entire basis for critical mass in the system. “Pips” are a good feature for people who are actually looking to form new social ties within the dungeon finder, I just wonder if that’s such a small portion that really they’d be better served by being advised to go elsewhere.

    Comment by Green Armadillo — 26 February, 2014 @ 5:02 PM

  5. I think a lot of the comments here are focusing on how players will perceive such a system. The design motivation is to encourage social connections, but I think the smart way to present it to players is by emphasizing that you’ll group with people you pip more frequently. By letting the player choose to pip who they want to group with more frequently, they should get a better experience. I think the rewards add an extra layer of appeal to people who might not otherwise think they would care about the system, but once they use it find that it enhances their experience and makes playing more fun. As I said, we’re encourages to present ourselves as busy, too busy to socialize, but I think there are a fair number of people who would appreciate building relationships in a game if it were a happy side effect of this type of system.

    RohanV wrote:
    Have you seen FFXIV’s Player Commendation system? It has the start of what you suggest.

    I think I remember reading that on your site. Maybe I was channeling that a bit. :)

    I think there is an important difference: my system rewards failure, too. Sometimes one person can make a failed run a lot less painful because they were trying to help or had a better attitude. I think the rewards are especially important here, because sometimes having good people makes failure a lot more tolerable. I also think it’s better to have some of the rewards be more immediate, and not be a separate system, otherwise you get people who really just want the rewards for the sake of the rewards.

    Jeromai wrote:
    That’s my concern with the system as well. It sounds great on paper, but I’m wondering if an end result is that players who favor fast, efficient clears will only pip other players that produce those dungeon results.

    I see this as a feature, not a bug. You can pip whomever you want. Maybe you want to pip the person who did the top DPS and didn’t die. Or you can pip the person who actually chatted in the group, whatever you want. Depending on your tastes and style, you’ll be grouping more frequently with people you like.

    Ask most people if they’d rather PUG with random strangers or group with people they know, I believe most would say they want to group with people they know. This system tries to get it so that you can get to know people easier by running into them more frequently.

    I think there’s also a false dichotomy here between “efficient players” and “people who like to chat”. It’s possible for someone to be in both camps. It’s possible for someone who is efficient to want some sort of interaction, too.

    rowan wrote:
    But if you group people who “like” each other more often, the social bonds between them strengthen, increasing the social stickiness of the game, which I believe is the point of Psychochild’s proposal.

    Precisely. You group with the people you like, which should improve your experience. And, you might make a friend along the way.

    Green Armadillo wrote:
    “Pips” are a good feature for people who are actually looking to form new social ties within the dungeon finder, I just wonder if that’s such a small portion that really they’d be better served by being advised to go elsewhere.

    As I said, one of the design goals is that this system shouldn’t be forced on people. There will be those who just want to run dungeons with random strangers and get loot. That’s fine. But, as I said, I think there are some people who might like the benefits of the system even if that isn’t why they initially used the system.

    Thanks for the comments so far! :)

    Comment by Psychochild — 26 February, 2014 @ 7:35 PM

  6. WoW’s example is a solid on if only because the odds of you ever meeting those people again is next to nil. Advocate a need for rebuilding coommunities, a social structures re-emerge. If it’s just a bunch of random faces, with no consequences, then GIFT right?

    A system that heavily weights friends, pips/colleagues/commendations, server folk – in that order – would be a leap beyond today’s reality.

    Past example with Rift. I hit max level relatively early on my server and was one of 3 healers who could heal. LFG made some friends as I met people again and again. We then built shortlists across guilds to do more dungeons/trade. That lead to alliances and cross-guild raiding. It just snowballs.

    Comment by Asmiroth — 26 February, 2014 @ 9:43 PM

  7. The answer is probably (sickeningly) to take a cue from Google, Facebook and the NSA.

    Spy on your players.

    Somebody who spends a lot of time in battlegrounds is probably a PvPer. Match him with other PvPers first.

    Somebody who uses emotes a lot is probably a roleplayer. Match her with other roleplayers first.

    Somebody who spends time in Raid instances… you get the idea.

    Comment by Rich Bryant — 27 February, 2014 @ 7:18 AM

  8. @Jeromai, I think there are two misconceptions: First, in mathematical terms, being grouped with is a symmetrical relation while pipping is not. That means, you will be more likely to be grouped with players you never pipped just because they pipped you.

    Second, being able to make the run faster is not the only criteria players may use to decide who they like to group with. I’m actually sure they will decide – at least – on how pleasant (polite, chatty or silent, etc.) they are and what country they come from (or what language(s) they speak). Also, none of the criteria is clear cut: player who continuously pulls new groups may be seen as someone who makes the run fast but also as someone who puts the group into unacceptable risk of wiping; player who keeps saying jokes may be seen as funny or annoying; the same player’s English (or whatever common language is used on the server) accent may be seen as proper or hard to understand for non-native speakers.

    And I believe the other criteria are more important for those who consider socializing rather than just using their group mates to help them to reward(s) they get for doing dungeons.

    Comment by Imakulata — 27 February, 2014 @ 12:54 PM

  9. Brian,
    I have to wonder if this is a case of treating the symptom rather than addressing the disease. Not that I’m suggesting that MMOs are diseased. I would never suggest that, no way, not me. It’s just a turn of cliché.

    This use case on ‘Random Grouping’ highlights just one of the competing agendas that MMO designers must contend with, in this case balancing socialization against achievement. The frequent result are systems that makes nobody happy. So given the basic question, how to improve ‘Random Grouping’, I got to ask what does improved mean.

    Does it mean that the groups that get formed by the system are better groups for the players or that the random grouping system works at making itself unnecessary by fostering a social environment that promotes player’s establishing their own groups.

    So much of the right answer is dependent on the rest of the game’s objectives. It’s hard to make a meaningful shift in broadening the socializer base when the rest of the gameplay favors achievers. Chances are this game has gone the ‘homogenous hero’ route and that class balance has effectively been balanced out of existence. This actually works better for automated grouping because de-emphasizing group balance allows for other, more player-centric considerations to drive the solution. A ‘Pandora’ like profile for players that tracks their progression and their play style might make for better match-making. Creating the proper ‘genome’ for a player profile would be the challenge.

    I see the decline in socialization between players as directly related to the decline in really distinct tactical class roles. Player’s just don’t need each other as much. I really can’t see the engagement between players that comes with the inter-dependency between roles being replaced with a carrot and stick game mechanic.

    The last thing I would do is ask a player to rate everyone else in the group. I just don’t like the signal-to-noise ratio on that particular data stream.

    At best, I think the game could ask a player to rate the group, provided it was a group produced through the match making system. The system knows a lot about the group’s encounter, how long the group was together, the turnover of the group, the level of success the group achieved and the number of setbacks the group experienced.

    How much of this data mining would you show to the player? I can think of reasons pro and con on this. But if it can clearly be demonstrated to a player that they are progressing faster, getting into stronger groups, enjoying more success, earning more valuable loot and, by implication having more fun, that would be a good thing to let them see.

    Comment by Kern — 27 February, 2014 @ 1:45 PM

  10. @Imakulata

    Your bringing up the potentially asymmetrical relationship of pipping raises an interesting question.

    Psychochild’s proposed system assumes that people who like one another will pip each other, and thus be matched together more frequently.

    What happens when the pipping is more of a one-way street? Say, many follower types pipping an inspiring leader that they like to follow. Or weaker players pipping a strong player, whereas the strong player doesn’t return the pipping?

    Would a one-sided pip still increase the chance of being matched together, over zero pips? If so, then would you not be slowly reducing the possibility of meeting entirely new people that might be a better match?

    Would people who have built up more pips end up getting into groups faster than people with no pips?

    Comment by Jeromai — 27 February, 2014 @ 5:51 PM

  11. Kern wrote:
    I have to wonder if this is a case of treating the symptom rather than addressing the disease.

    This is kind of by design. The goal was to improve a random grouping system, which means it’s assumed that the game is already has a random grouping system in place. Ripping it out entirely is probably not feasible for an existing game, and designing a replacement would be a much bigger task.

    Random grouping systems do have some things they do well. They let you get into content you might not otherwise see. It also prevents a lot of problems like elitist exclusion if the system is making the matches instead of the players. There’s nothing precluding people from playing the dungeons without the random grouping system once they find some people they like.

    Jeromai wrote:
    What happens when the pipping is more of a one-way street?

    I didn’t go into a whole lot of detail about how the matching algorithm might work, as this was already a pretty long post. But, there are a few ways to address your concerns.

    First, the system should obviously prioritize mutual pips over one-way pips. This should make sure you have people who like each other grouped more frequently. If this doesn’t produce required results, then you could adjust the system to where if one person has piped another, and the second person doesn’t pip the first after X matches, then the chance to be matched in the future is reduced. There’s still some merit to matching one-way pips over purely random matching.

    As for meeting new people, you can have your matching system choose to ignore pips on occasion. Or, you only get a maximum of half the group of people you’ve pipped before, so there’s a chance to meet new people.

    Would people who have built up more pips end up getting into groups faster than people with no pips?

    I don’t think this is a requirement, but it’s possible to add this to the system. I was thinking that what you would do is every 5 minutes or so the system would take a look at people who are queued and form groups with biases toward people who have pipped each other. You’d need to create some code to understand a directed graph.

    Some of the probabilities and limits could be adjusted to get the outcome that works for the game. Given that the numbers would be internal and completely server side in a proper implementation, this would be easy to adjust on the fly based on data collected.

    Comment by Psychochild — 27 February, 2014 @ 7:12 PM

  12. Some other interesting discussions happening on Google+:

    A few highlights and responses.

    Chris Smith wrote:
    This sounds a lot like a “Pandora for Grouping” system. Over time, you’d build up a system-preference list of people to consider first when using the LFG, and the more you like a person, the more you can re-pip them during subsequent runs.

    What if someone becomes a total ass-hat, though? Like if they’re efficient, but quiet, and you pip them a few times, but on the 5th match-up they turn out to be a real asshole. Could you un-pip someone??

    I see no reason not to let people manage their “shared pips” friends list like a normal friends list. There might also be some decay in the system, so that if I’m grouped with someone I’ve pipped before and don’t pip them again within a limit, they get removed out.

    Brenda Holloway wrote:
    People play MMOs for different reasons, and those reasons may differ from day to day. The EQ-style of extreme difficulty where grouping with people you did not know well probably meant disaster led to socialization, but as has been noted, also leads to severe risk aversion.

    Most groups in EQ that were composed of random people who stayed in one spot and killed the same monsters again and again. You started chatting out of desperation, but most of the time, you didn’t do much more chatting than “inc” or “ch” or “mezzed” — most people had hotkeys set up to announce stuff the group needed to hear.

    Most of the chatting and socializing was actually done while looking for a group, waiting for a respawn, or just hanging out with friends in the zone.

    That is absolutely never done by any MMO ever anymore. They accomplished that by removing risk for non-raid content.

    EQ was also different in that there were actually very few active players allowed per server. Modern MMOs are set up for thousands of players at a time, but the original EQ had nowhere near that capacity. A hundred people in one zone would bring it to its knees. The lower capacity meant there was an excellent chance you would see people you knew when you logged in even if you didn’t have them on any sort of list. There were so few people that everyone had a reputation, and if you didn’t, you were viewed with some amount of suspicion. Reputation is everything.

    In modern MMOs, I’m fairly certain I will never, ever see a random teammate, ever again no matter how long I play.

    Modern MMOs have made the social experience impossible, and it shouldn’t be shoe-horned in. The social experience has moved to Twitter, G+, forums and so on.?

    I disagree that the social experience is impossible. I think the brevity of early EQ interaction was due to wanting to get info out quick and not being able to just say something. What needs to happen, though, is that designers need to focus more on creating opportunities for people to talk. The old methods of making people wait don’t work anymore.

    Richard Bartle wrote:
    Who’s the tank going to pip? The healer or a random dps?
    Who’s the healer going to pip? The tank or a random dps?
    Who’s the random dps going to pip? The tank, the healer or a fellow random dps??

    It depends on the player, doesn’t it? Maybe I have an unnatural fondness for the letter “P” and primarily pip people presenting pseudonyms prefaced with ‘P’. Then in future runs I am mostly surrounded with glorious “P”s. That’s a feature, not a bug.

    Achievers can pip people who were most efficient.
    Explorers can pip those who praised (or at least didn’t criticize) their odd builds.
    Socializers can pip friendly souls who were fun to chat with.
    Killers can pip victims who endured the abuse without submitting a ticket.

    And they’ll run into these people more frequently in future runs. Everyone is happy, and retention improves. Sounds like a win all around, except for maybe those Killer victims!?

    I wrote:
    I think there is some misunderstanding about social interaction by the players. The first is that social gamers need other gamers. Some of us old hands get around this by having friends we join in whatever game. But for people who don’t have that advantage, modern games don’t provide enough opportunities for people to interact. As +Chris Smith points out, you need to run into people multiple times for that to happen normally.

    The second misunderstanding is that social interest is more common than people believe. Again, I think part of this is because some people have their own established groups from prior games and don’t see a need to go outside of them. I also think that some people don’t even realize they would enjoy socialization because modern MMO fashion is to minimize socialization to reduce friction of playing the game. So, people who might normally enjoy socialization in games aren’t given the opportunity to do so. It’s like only having a Starbucks in town and seeing people only drink coffee then drawing the conclusion, “Well, nobody likes beer in this town!”

    The funny part is that it’s actually really easy to design socialization systems because the players mostly do the hard work themselves. All you need to do is provide communication channels and opportunities to interact and it pretty much goes automatically. The pip system is really just about providing those opportunities, but in a way that doesn’t make the system only suitable for socializers.?

    Brenda Holloway also wrote:
    Ya know, we hear a lot about how G+ and FB winnow the posts you see depending on … factors. Here on G+, we find people we want to circle based largely on shares or comments on our own posts.

    WHAT IF…..

    What if you didn’t see everyone in the zone all the time? If it’s too crowded, you see some based on some sort of weighting algorithm.

    Every time you group with someone, they are weighted a little heavier — or lower if it went badly (default is heavier). Eventually, if you play enough, the world will tend to become filled (as far as you can tell) with the kind of people with whom you like to group, and their connections will organically become your connections.

    Spammers, with no connections, will reach very few people. Wonderful people who play a lot will get many connections just by playing and not making people angry.

    An organically generated, weighted social connection graph. They generate them from e-mails and social network posts already.

    Now THIS would be a way to bring the social into MMOs.?

    Interesting expansion on the concept. The one concern I’d have is that the pool of people that could be in the same zone is probably smaller than the pool of people who participate in random grouping systems. This would lead to more pronounced positive feedback loops unless you find a way to regularly inject fresh people into the system. The nature of WoW’s cross-server LFD/LFR means that you will get a lot of churn, so even a person who has pipped several dozen people will still get new people in their group without the designers having to design that into the system specifically.

    Comment by Psychochild — 27 February, 2014 @ 7:33 PM

  13. I am sure a significant amount of pips will be asymmetrical although I don’t know whether or not it will be anything close to majority; just because someone appreciates e. g. people who can keep conversation going in a group, doesn’t mean they are good at the same thing themselves. Also, if players have limited amount of pips to give, ones who are pipped a lot just wouldn’t be able to pip everyone back (and at least some of the unpopular players will still want to pip others).

    I think an important question for the design is: How will players decide whom to pip? I think many assume there is a single scale (usually “makes the run go quickly”) and players will pip others based on how they (i. e. the others) did on the scale. On the other hand, I think Psychochild assumed they will naturally form a kind of “circles”, where players having a certain quality will appreciate the same quality in others. I also mentioned some players may use a multiple scale model (with scales such as “makes the run go quickly”, “is polite”, “speaks English well”, “has a nice moustache”) with each scale given a weight by a player which depends (although not exclusively) on how well they themselves do on it (a player with an ugly or no moustache will not care about moustache quality, on the other hand player who only has a mediocre moustache may appreciate a nice one and player with a nice one may not care about others’ moustaches at all).

    Of course, any model that uses scales will have asymmetrical pips, so if there is enough players using such model, there will be quite a few of them.

    I’m hoping I was able to explain what I meant clearly, I’m sorry if I did not.

    Comment by Imakulata — 28 February, 2014 @ 2:43 AM

  14. But does grouping have to be ‘random’ or ‘anonymous’?

    Why not go e.g. DDO route and create convenient interface to advertise their own groups? Including ‘description’ field that can be used to e.g. write ‘no spoilers’, ‘speedrun’, whatever?

    Yes, it would force people to do at least minimal socialization by the way of applying to some group explicitly — but is it a bad thing?

    Comment by Solf — 3 March, 2014 @ 10:11 AM

  15. Solf wrote:
    But does grouping have to be ‘random’ or ‘anonymous’?

    Well, the dungeon (and later raid) finder were put into WoW to lower the overhead of organizing a run. WoW also pulls people from multiple different servers, which causes other issues with DDO-style group formation. Queuing for a group and then being put into one “randomly” requires the least work for the player and the most freedom in allowing them to do something else while the group is forming. So, this works well for a larger game like WoW.

    Truth be told, I do prefer to the DDO system. Although, it might be nice to have a system that would allow you to pull people from across different servers. That would be tough to do given how DDO puts you into a group as soon as you are interested, and how quests are usually a commitment so running a quest while you’re a queued for another group might cause problems.

    Anyway, my point is different games have different situations and expectations of the playerbase. WoW players obviously like the random dungeon finder okay, I’m looking for ways to improve the experience and provide long-term benefits for the game and for the players.

    Comment by Psychochild — 3 March, 2014 @ 4:39 PM

  16. Hey, is it this topic again? Grouping in MMOs, hmm.

    I don’t like “forced grouping” where I’m required to group up just to play the game, if I’ve been able to get by without it for most of the experience. I don’t mind “forced grouping” if it’s obvious at the start this game is going to require it. I would prefer grouping with people I know, but honestly with my play patterns due to working during “prime time” I’ve not had that luck too much.

    I half liked how Guild Wars 1 did it where the whole of outposts were lobbies for grouping up, and it was more than likely if you stayed in a certain place you’d find people who were interested in grouping from that point. Be it for one of the missions, dungeons, or just to kill certain targets for whatever reason, if they were sitting in the place then they weren’t likely going to be looking for people to do something halfway around the world. And there was a “Party Search” where you could post LFG or GLFM short statements. So it wasn’t as random as “find me a match, pull the lever”, it was more “let’s put up a postit and see who answers”. I really enjoyed this since it was likely more socializing could happen since you were -there- and could chat before joining up a group.

    Re: Your “pip” idea

    I’m not sure I like the idea of only one pip to give out, especially when connected to rewards. I will again note there was something one game I played (not sure which) did which was track “recent contacts” which would temporarily (for like the space of a day) hold people you PM’d or grouped with who weren’t on your friends list. They didn’t count towards a list limit.

    I like the idea of the finder weighting groups which had pip-holders previously in them. Though . . . that could prevent people from finding NEW people to join up with.

    Comment by Kereminde — 14 March, 2014 @ 7:02 AM

  17. Here’s a paper from other game designers talking about building friendships in games on a more general level:

    Food for thought.

    Comment by Psychochild — 2 February, 2017 @ 5:05 PM

Leave a comment

I value your comment and think the discussions are the best part of this blog. However, there's this scourge called comment spam, so I choose to moderate comments rather than giving filthy spammers any advantage.

If this is your first comment, it will be held for moderation and therefore will not show up immediately. I will approve your comment when I can, usually within a day. Comments should eventually be approved if not spam. If your comment doesn't show up and it wasn't spam, send me an email as the spam catchers might have caught it by accident.

Line and paragraph breaks automatic, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Email Subscription

Get posts by email:

Recent Comments


Search the Blog


January 2020
« Aug    



Standard Disclaimer

I speak only for myself, not for any company.

My Book


Around the Internet

Game and Online Developers

Game News Sites

Game Ranters and Discussion

Help for Businesses

Other Fun Stuff

Quiet (aka Dead) Sites

Posts Copyright Brian Green, aka Psychochild. Comments belong to their authors.

Support me and my work on