Wednesday, May 11, 2011

Optimization and Player Segregation, Part II

Keeping my last post in mind, I want to venture a bit further.

First, a side trip to World of Warcraft. In this recent post I had written down six ways to combat the optimization problem, or the inefficiency phobia, whatever you want to call it. However, the last post reveals another way to look at it:

WoW is, without doubt, easy to learn and hard to master (1). If you think differently, please stay away from my blog. In addition to that, it also makes a lot of difference in WoW how skilled you are (2). The difference between a very good player and a credibly bad player is tremendous. These are really two different dimensions and need to be looked at separately!
It is clear that this combination of characteristics in a game induces a run for efficiency in the player base. This run turns into a social drama, because WoW also (practically) forces players of all skill levels to play together (3) in the LFD or to raid as a guild that was not originally estabilished for raiding.

All three characteristics together create the drama, and changing any of them will reduce it. You can either make it very easy to master WoW. In that case, everybody plays rather perfectly and there is no drama. Or, you can make it hard to play perfectly, but have the benefits of doing so be low. An example would be an extremely complicated priority system in boss fights, that is 3% more efficient than a one-button spam. In this case, only few players would play perfectly, but those who did would be oblivious to those who do not. And finally, you can segregate the players according to their skill.

Now, leaving the WoW example behind, how realistic is each of these measures?

Is it feasible to make a game that is not easy to learn and hard to master? I don't think so. This characteristic is about exploration of the game and as long as there is still something to explore, players like to stay. It is about mystery. Also, one reason playing soccer at medium skill is fun, is because there are people who play at a world championship. The celebrities enrich the game - especially for the people who play at much lower skill. So, no, a good game should be easy to learn and hard to master; even if most players don't actually master it.

Is it feasible to segregate players according to their skill? Yes. The nature of the segregation can either be natural, that is, players look for similarly skilled players to play with. This is how things work in most of today's games and sports. Alternatively, or in addition to that, you can use a measuring system, like the Elo rating that is most prominently used in chess, but also by the WoW PvP match-making. These systems have several hard limitations, like requiring a high number of wins / losses to accurately determine a player's skill. There is no known system to guess the skill of a player without clear feedback, like wins/losses, by the way. Warhammer Online, as well as Rift tried to implement such systems into their public quests, but failed miserably. Rift gave up on it soon after release and now gives out 'rewards' for participation alone.

Finally, you can try to keep the benefits of mastery low, but still have the game be easy to learn and hard to master. Of course, the benefits must not be too low, otherwise the players don't care for mastery, and in that case, the game could as well not be hard to master. Also, this is incredibly difficult to design! Firstly, it is hard to measure skill in the first place. Secondly, the benefits of skill are hard to measure. Finally, in a non-boring MMORPG, there are lots of strongly connected activities that skill can be attained in. To balance such a system against other classes/speccs/skill combinations is already hard. To also balance them carefully for benefit of additional player skill in various connected activites, is probably a nightmare.

Finally, I'd like to say that not all activities in a MMORPG have the mentioned problem, in the first place. Take in-game economics as an example. Imagine a player-run economy that offers a very easy-to-use interface to the economically non-interested player, but offers a complicated and deep buy-order/sell-order system to the interested player. In this case, additional player skill isn't a problem, due to player expectations. Since players know from the real world that making less money is perfectly ok if you are economically less skilled, they don't complain about it in-game at all! That is funny in so far as the gold coins in a MMORPG are, of course, gameplay-wise indistinguishable from any other arbitrary points. But from a simulation-aspect point of view, it is a currency. And players have very specific expectations about currencies. For example they expect their wealth to decrease from time to time and have no problem with it at all.

Another example are organization skills. Similarly to the economics example, players have very strong expectations about the influence of organization skills in-game. Nobody complains that the student has an advantage at organizing 40 people, because he has more time. While this is obviously the case, it is also, obviously, impossible to change. So expectations cannot be messed up by the developer here.

Conluding the last few paragraphs, it is possible to have activities in-game that are
- easy to learn hard to master,
- bring along powerful benefits for mastery,
- don't segregate the players,
- and don't create a drama

as long as these activities befit from strong already-estabilished expectations.

Segregation also, doesn't necessarily mean that players have no contact with each other. On the contrary: the hard working organizer has contact with a lot of people and plays a very complicated and deep game with its own benefits for mastery. But he doesn't play against the other players, but together with them. The better he is at organizing, the better off are the players he organizes.

The same applies to the skilled and hard-working in-game merchant. The real-world expectations about currencies and economics are mentally transferred and keep players from complaining about his 'unfair benefits' from skill. At the same time, everybody benefits from the liquidity he creates and indirectly interacts with him.

---
Check out Klepsacovic's posts on a very similar topic.

17 comments:

  1. "WoW is, without doubt, easy to learn and hard to master"

    Relative to the amount of time people play and the resources available? The skill progression may be there, but it isn't particularly steep when you think about it. A player can be proficient in their role, get an optimal build online, and do just fine without any extreme skill-training effort.

    I guess mastery and proficiency are two different things, which is part of the problem, a player can be proficient without being a master, so reaching that level doesn't really have as much distinction as it should.

    ReplyDelete
  2. Gilded, there is a step up from running a rotation on a target dummy to running a rotation in a movement-heavy fight. From my own experience, I know this can be a big jump to make. Now throw in target switching, so then new choices have to be made on ability use, and mastery can come into play. Add the element of reflexes and quick-thinking on top, and there's a clearer distinction between mastery and proficiency.

    ReplyDelete
  3. "there is a step up from running a rotation on a target dummy to running a rotation in a movement-heavy fight"

    Is this step really that large as compared to mastery in other genres? I understand what you mean by movement, like the tank having to get out of the way of glowing areas, but for the most part a dps really is based on the amount of damage that its stats and rotation let it deal. Also, as far as gameplay, I wouldn't put movement in WoW as a great system which gets its challenge from its merits.

    ReplyDelete
  4. By the way I thought about rating system for quite a bit and while its not trivial I think it could be solved.

    The core idea is to use percentile distribution. You take absolute numbers of each measured stat and rank each player globally. Top 1% are highest ranked etc. This way it maintains consistency across any numbers of players

    Now the ratings number are the tricky part. The actual "what you measure" is tied to core mechanics of the game. I have in mind pvp centric game so it revolves heavily around K:D and objectives completed.

    But it also tracks players based on their specific roles. (DPS/Heal/Support). Core idea here is again rank by % , not raw numbers (but you obviously track both). And those things still comes secondary to K:D and objective completed . Precisely because they can be gamed

    The absolute rating number is only visible to the player, the bracket is displayed to everyone.


    Trick there is that even if you lower % player it does not belittle you and you are a part of a quite large bracket ( no one can tell how precisely good/bad you are).

    The clever trick is also that most active player who pvp at all would be above militia rating as the bulk of it would be completely non combat characters (so it makes average player feel pretty good :)

    All in all players form following hierarchy.

    -Militia (countrymen rallied for defence in harsh times - what they may lack in skills they compensate with bravery and courage) (first 30%)
    -Troop - the back bone of a force, they carry the load of battle( 30-70%).
    -The Guard - the grizzled veterans of many battles (70-90%)
    -The Elite - top 10% -
    -Champions - top 1% -
    -Heroes - top 0.1% -
    -Legendary heroes 0.01% -paragons of skill , courage and leadership.

    ReplyDelete
  5. Gilded, I think there is a pretty big difference between mastery and "proficiency" (I would use the word "competence" instead, I think).

    I played WoW for a little over six years and I never felt that I had mastered it. Even in my last days of playing I would still run into situations where I realized I could have played better to avoid a wipe, or played better to at least stave off a wipe for a few seconds to give the rest of the raid time to also play brilliantly to pull it out. I'm sure I've put more hours into WoW than any other game, but I always found room for more improvement in my play.

    If a person's play is not good enough to beat the hardest content in the game at all, let alone to beat it with relative ease, how could we possibly describe that person as a master? And if that is a minimum requirement for a claim of mastery, then how could we say it isn't hard to master when less than 0.1% of players can get to even that level?

    ReplyDelete
  6. Gilded, WoW arena is very hard to master by definition. And also heroic raids are very difficult, even if the difficulty lies only in the execution of fixed complex instructions.
    If you don't believe it, have a look at the number of serious raiding guilds and have a look at the distribution along the progression. It is similar to a gaussian curve.

    Likewise, the entire leveling game is not only easy, but outright trivial.
    But please don't make the common mistake to look at the leveling game alone or claiming to "just follow the youtube instructrions". Mastery of WoW is not easy - even if the difficulty lies almost completely in the execution. Which is something I dislike, by the way.

    Max, the core problem at automatic determination of skill is not the theoretical possibility, but the speed of convergence. This is also why Blizzard doesn't measure your skill at playing random BGs while you play them. Since teams would change all the time, players would need to play an absurd amount of BGs to be fairly measured. The convergence of the technique is not fast enough to be practical.

    I am not particularly fond of automated segragation techniqes, especially if there is no justification by the simulation aspect. I think one could introduce a system like the one described by you, but one shouldn't rely on it much for the rest of the game design.

    ReplyDelete
  7. "I played WoW for a little over six years and I never felt that I had mastered it."

    I think that is part of the point. Mastery is not really a necessary level of ability for content for most of the classes. A player can be proficient at the game without mastering it, which counters one of Nils points: "- bring along powerful benefits for mastery"

    I wouldn't really say that there are benefits for mastery in WoW besides more efficiency clearing the same content.

    ReplyDelete
  8. Let's not make this about semantics, please. Imagine a usual learning curve. WoW's curve is not correct on this picture. It starts out extraordinarily flat and stays like that until max level. At this point, if you want to raid or do professional PvP, it suddenly makes a very, very steep jump. Not as steep as Eve Online's learning curve, but still very steep.

    It then continues to climb at a high speed, similarly to Eve Online. But while difficulty in Eve is mostly the grey theory and only a little bit execution, in WoW it is mostly the execution, in WoW raiding it is all about execution.

    At the very, very utomost top of the WoW learning curve, planning and stretegic thinking suddenly start to play a critical role. At the top there are no youtube vids and no guides. If anything, you are the one who writes them !

    Well and at the top of the very very top ;), you are forbidden by your raid to write the strategy down and you are required to log out with fake speccs & co., to make it harder for your rivals to beat you.

    Of course, most players, me included, never come near to the top. Even if we wanted and invested 24 hours a day, it would probably be impossble for us to reach the pinacle of progression, because our rivals there also spend 24 hours a day.

    But make no mistake. A player fulyl decked out in the most recent heroic items is an awe inspring sight. I have never seen one, I think ;).

    When discussing WoW, I think, it is of utmost importance to avoid hypocrisy.

    ReplyDelete

  9. Max, the core problem at automatic determination of skill is not the theoretical possibility, but the speed of convergence.

    Since teams would change all the time, players would need to play an absurd amount of BGs to be fairly measured. The convergence of the technique is not fast enough to be practical.


    It does require a fair amount of games to be a good measurement, but in no way its absurd. Generally around 100 "games" (in my game "encounters" - since there is no instances or such) is a good start. If for nothing else but be a nice round number :)

    In all seriousness ladders do work, they do not require "absurd" amount of matches. Even the simplest approximations of skill(such as k:d for example ) turn out to be surprisingly reliable as an approximation of skill( or W:L if case of team encounters)

    You need not go farther then wow arena or sc2 ladders for examples.




    This is also why Blizzard doesn't measure your skill at playing random BGs while you play them.


    The reasons blizzard does/does not x or y are imho complex and varied. And not always lay in the gameplay area.

    ReplyDelete
  10. I don't believe I said that mastery is not an option in WoW. There are enough variables that "mastery" is certainly present and is not an easy thing to reach. The problem I was suggesting is that the part of the curve you are referring to before the steep incline to mastery, the part where you learn how to play the class, how to build the character, what rotations to use, etc, that's pretty much the extent of what a player needs to go through. There isn't really much merit to "mastery" besides efficiency (and besides PvP, but that's its own beast).

    ReplyDelete
  11. Gilded, I think there is a lot of merit to mastery in WoW-raiding. I tried beating heroic content myself and although I certainly wasn't the worst player in our raids, I, too, made mistakes in the execution. It is just damn hard. And since I know this, I respect every player who does beat the complete heroic content. This very respect is the merit.
    This doesn't keep me from repeating that this style of play is not what I really want from a virtual world, by the way.

    Max, this all depends on the circumstances. I wouldn't introduce a K:D system in a "default virtual world" that theoretically allows griefing, for example. I do not deny that given the correct circumstances Elo systems work and converge reasonably fast. They do in many games. But how would you make them work in a virtual world without instances and a focus on the simulation aspect?

    ReplyDelete
  12. Max, this all depends on the circumstances. I wouldn't introduce a K:D system in a "default virtual world" that theoretically allows griefing, for example. I do not deny that given the correct circumstances Elo systems work and converge reasonably fast. They do in many games. But how would you make them work in a virtual world without instances and a focus on the simulation aspect?


    K:D was just a general example . I specifically mentioned that the rating has to take into account concrete game systems and mechanics

    In my case for many other design reasons activities have explicit "completion" points. - For contract enforcement ,accountability , etc .Most organized pvp is objective based in form of : attack/defend caravant , outpost , resource point etc).

    Assassination (e..g "griefing") are contracts too. Some of them explicit (you have to pick it), some are not(automatically activated) ,some are both (for example you could get a contract from city to defend caravan, or you could join defence if your run into attacked one)

    So a lot would revolve around objective completion ratio. With some weighted adjustments based on rating (this is more about segregation of player based on skill) . Objective completion is very hard to game with proper controls in place.

    It also allows to reward support roles as well as teamwork etc .If your team is better at completing objectives - you are better player. That is very simple concept reflecting the battle performance. Better players play in better teams.

    p.s. "Team" is 5 players

    ReplyDelete
  13. Max, having objectives that cannot be gamed and having clear feedback of 'accomplished' or 'not accomplished' would certainly work. When it comes to the speed of convergence the question is: How dependent is the outcome on factors that are outside of your personal skill?
    If there is a low dependence then convergence is quite fast. If there is a high dependence convergence can be quite slow and even unfeasibly so.

    ReplyDelete
  14. When it comes to the speed of convergence the question is: How dependent is the outcome on factors that are outside of your personal skill?


    As much as possible . its natural when encounters are small (1vs1 , 1 group vs group, but start breaking down with uneven numbers)

    I carefully considered general metagaming factors which break it. Such as "zerging": for example in "world" pvp is generally better to press on with superior numbers. And real "game" so to say is largely dependent on logistics ,politcs ,organization ,spying (hello eve) ,etc. - All those things which are outside of influence of average player

    And the solution is in several ways. First encounters are designed with tiers in mind (individual ,group, raid, faction)

    In an alpha -beta stage balance works is generally done to tuning that funneling of players (e.g. if you only have 1 caravan and 200 players want do it at same time it obviously would serve no good to design it as group).

    The tuning of caravans tasks would ensure that there is enough for most groups to get their own.

    What will generally prevent the zerging of single caravan is that rewards would be scaled on participation. Implicitly -as say caravan/mine designed would 1 group would only carry enough resources to justify 1 group commitment.

    Of course one side can choose to zerg a particular encounter with +x groups more than needed. But that would let the other side spread their groups more evenly and would be overall winner even if they have no hope winning that particular one.


    For larger scale encounters there are objectives assigned to teams (such as hold tower x) by battle commanders. While they overall flow might not go well for your whole battle your little group can still be appreciated if it does its job. If the teams perform well but battle is lost - that goes into battle commander rating.

    At global level faction balance comes into play my faction system is designed in such way that there are ways to bolster the failing side (and there is 2 and half sides, the half side being there for balancing)

    ReplyDelete
  15. p.s. I should add there are multiple approaches to ensure that in general the flow of encounters goes in the designed way ,if the funneling will start failing

    General design philosophy for pvp is:

    -Its fun as long as its against similar skill opponents

    -its fun as its generally against manageable numbers

    -If 2 above are maintained its fun as long as there is 50% W:L

    Design is aimed to achieve that.

    Of course there are cases when deviating from this would be even more fun (such as beating impossible odds in numbers), but that would be as flavor, as you cant maintain this for both sides.

    ReplyDelete
  16. How about making roles more varied? For example, EvE includes some skill-intensive activities (like PvP) and some that are useful but require little skill (like mining). That way, people could choose the type of activity that matches their time commitments and skill level. Obviously this would result in "segregation," but players of different skill levels would still need to interact (PvPers need ships from crafters, who need ore from miners, etc.).

    Another possibility would be to focus on open world content without fixed group sizes. For example, keep battles in LotRO usually involved people of lots of different skill levels because numbers were important. Every extra person was welcome even if not particularly skilled (although skill certainly helped).

    ReplyDelete
  17. Hmm, looks like Blogger devoured my earlier post. Just a couple of quick ideas:

    (1) You could make roles more varied, so that some are skill-intensive (like EVE PvP) and some are not (like EVE mining);

    (2) You could focus on content without strict numerical limits, like keep sieges, so that even unskilled players are welcome.

    ReplyDelete