Double Buffered

A Programmer’s View of Game Design, Development, and Culture

Archive for June, 2009

Never Balance Cool Against Useful

Posted by Ben Zeigler on June 29, 2009

One of the more important parts of most RPGs (both online and not) is having cool rewards, often in the form of items. Great loot design can elevate a repetitive activity to be extremely satisfying (Diablo 2 is the best example of this), or can drag down an otherwise well designed game. To have a good loot system, you need a decent pool of desirable items, and systems for allowing players to acquire the item they want. There are a bunch of interesting ways to grant access to loot (quest rewards, random drops, crafting, auction house), but all design effort put into those systems is a waste of time if the end result isn’t actually compelling to the players. So, what makes an item compelling enough to drive player desire? I think the real value of an item is a combination of two components: Utility and Coolness.

I’m using Utility in the mathematical sense. Basically, the question is how will a particular item help the player achieve their goals more efficiently. Many players (especially the higher-level players) are trying to min/max their character and all they care about are the raw numbers. If you don’t give them hard numbers, they’ll come up with their own (possibly flawed) ways of figuring out what the best gear is. Also, once these players have come up with a system for evaluating item utility, any item ability that is complicated or hard to quantify will be considered of dubious value. In short, these players like Stats. The higher stats an item has, the more they want it.

Coolness is a bit harder to quantify, and refers to a cluster of different things. First of all, there is the simple issue of how cool it LOOKS. Also, items that are specifically rare (ie special mounts in WoW) will be highly valued regardless of utility. More relevant to combat design is how fun an item is to USE. Items that have cool effects (procs, conditional bonuses, active abilities with a long recharge, etc) can vary the combat in interesting ways and keep a game fresh, even if they don’t necessarily improve a player’s overall efficiency. They can even encourage a player to try out completely new tactics, essentially creating a mini-game nested within a larger game. More casual players, as well as players who crave variety, are big fans of items with cool abilities. However, it is difficult to strike a balance between making situational effects useless and making them too powerful (where combinations of effects can break the balance entirely). If you do it right, players will appreciate the unique abilities but won’t be able to abuse the system.

Okay, so there are two reasons why items are interesting: the quantitative improvement offered by Utility, and the qualitative improvement offered by Coolness. They both provide value to an item, but the perception of that value depends heavily on the player. Hardcore number crunchers will devalue cool effects for not being obviously useful to efficiency, while casual players will devalue pure stats for being boring (I think World of Warcraft equipment has actually gotten significantly more boring over time as they cater to the hardcore more). Given that we have two dimensions of value that are difficult to compare, how do we build an economy around them? You could try to estimate the average economic value of stats and cool effects and attempt to directly balance them against each other. This seems to make sense, but I believe it is the wrong approach.

If you balance Utility against Coolness, what you end up with is a system where the people who want stats will pick the items with the best stats but no coolness, while the players who want variety will pick the items with bad stats but lots of cool things. This has two horrible problems. The variety seekers ends up with a set of complicated conditional effects but will be significantly worse at everyday efficiency, which means they’ll fall behind their friends at levelling and be irritated at the game difficulty. The stat seekers will end up with no interesting combat effects, which means they’ll get bored of the combat more quickly. Both players get what they think they want, but are more likely to be dissatisfied over the long time.

The solution is to not balance utility against coolness, but to scale them both up as the value of an item increases. As a basic example, a “common” quality item should have mediocre stats and be fairly boring. If your baseline is fairly boring it gives you more space for improving your higher quality items without having to get too crazy. Then, a “uncommon” item should always have better stats and have some sort of interesting conditional ability. All “rare” items should then have better stats and be more interseting than all “uncommons”. Put another way, instead of constructing an item from one pool of “item points”, you instead construct it from the dual currency of “stat points” and “cool ability points”, which vary per item level and quality. This system results in items that are valued approximately equally by both stat-seekers and variety-seekers (and the majority of players who are in between). This means everyone strives for items that are both fun AND useful, and so everyone is happy! Well, everyone who can pay is happy, and everyone else is looking forward to being happy. In the world of MMOs, that’s even better.

Posted in Game Design, MMO Design | Tagged: , , , , , , | 3 Comments »

The Players Are Wrong, But Listen Anyway

Posted by Ben Zeigler on June 16, 2009

When you’re deep in Beta, or you’re just taking the unusual step of actively seeking out what people think about your game, you’re going to interact with direct user feedback. Now, unlike feedback that has been filtered through OCR (which sometimes improves it and sometimes obscures it), direct user feedback tends to be either VERY FORCEFUL or just a bit confused. Based on my observations of fellow employees it seems to be standard policy for a new developer to seek out feedback, realize that some of it is hateful or confused, and then conflate the rest of your player base with the minority (well, or majority depending). The human brain really likes to make those kind of assumptions, so it takes conscious effort to get over the impression that not everyone who disagrees with you is a worthless human being. It turns out that most of the direct feedback you get from users DOES have value, if you just know how to mine it. Anyway, here’s my informal guide to Actually Learning From Users, broken down into helpfully pedantic steps:

1. Note Context

The context of the feedback should have a large impact on your interpretation of it. For the MMO-playing audience this can be broken down into Official Forums and Email (except for those bizarre companies that think they’re a waste of time), MMO Forums (F13, Fires of Heaven, Massively), and General Forums (Kotaku, Joystiq, etc). Feedback on the official forums tends to come from either low post-count users with specific issues or high post-count users who are part of the forum community. MMO forums tends to be full of cynical posters with a high level of knowledge and distrust, but with the occasional great suggestion. General forums are full of people who sort of heard something about your game and are a good way to get a feel for how well your marketing (formal and word of mouth) is working.

2. Ascertain Motive

Once you know the context of a posting you can make a quick guess as to the motive of the poster. The first motive is the always exciting Trolling. From my personal experience this is actually pretty rare on official boards, but this happens a lot on the general boards. If the goal of a poster is to instigate conversation or argument without actually contributing anything, go ahead and ignore them. A related but much less evil motive is Conversation. These posters will have very high post counts and are an integral part of the community, but most are useless for feedback purposes because message board posting, much like Blogging, is all about volume instead of content. A small number of high-volume posters graduate past Conversation into Aggregation. These users (I was one at one point) actively enjoy collating feedback for developers, and some are very good at this. During City of Villains Beta I trusted the bug-aggregate threads far more then I trusted our Q/A department, sad to say. Past those, another motive is Confusion. These users are often new and have a lack of knowledge. Because of this they will often be ostracized by the other members of a community, but to a developer they’re invaluable. If a number of posters are confused by some part of the game, there are no plans to make it better, and your game isn’t Darkfall, you Have A Problem. Either the system isn’t being explained well enough, or more likely your system is just too damn complicated in the first place. If players don’t at least have the illusion of understanding a system, they’re going to be constantly second guessing their choices and be too worried to have any fun.

The last motive is Advocacy. These users have a very specific goal, which is to get something about your game changed. These are the trickiest to deal with, because the first instinct is to ignore people who want something as being selfish and self-serving. But this is only part of it. Users who constantly gripe about their class will always exist no matter how well balanced your game is. The usefulness of Advocacy feedback comes in specificity. If a player feels that his class is gimped overall but doesn’t provide detail, that post is largely worthless. However, if a player specifies that a specific ability seems underpowered, and non-regulars agree (regular complainers will agree that everything is broken), it’s worth taking a look at. The odds are good there’s either a design bug or misleading player feedback. Oh, and as for those players who REALLY care about the way that one CERTAIN cape looks and it NEEDS TO BE RED or else they’ll QUIT FOR REAL, those players should be humored but largely ignored. MMOs are so much about personal identity that it’s inevitable you will highly annoy detail oriented people who can’t have exactly what they want. The only people who threaten to quit are those who are highly invested in your game: you should worry about the ones who don’t give a shit either way.

3. Extract Value

Any feedback that passes the Motive check (specific advocacy, content aggregation, and confusion) contains some value to a game developer, and I refer to this as Valid Feedback. Now, I use Valid instead of Good for a reason. The posts contain useful information but there’s absolutely no guarantee that it’s correct information. For instance, a confused user will often make weird assumptions about the cause of a bug because of their lack of background. The useful kind will still report what confused them, but if you see completely illogical feature requests or feedback it’s probably from someone who is confused but doesn’t want to admit it (because of the social ramifications of admitting to being confused). In these situations you have to ask “Why would they say this?” instead of “What did they say?”. Let’s say a user angrily posts that you’re evil for removing his favorite power. Instead of ignoring him because the power still exists in the source data, you can check to see if something weird broke in the runtime to hide it from the UI.

This same principal needs to be applied to Advocates. When someone discovers something about a game that makes them unhappy, the average player will not do a very good job of figuring out why they’re unhappy. Instead, they will grasp around a bit until they find a vaguely plausible cause, and then flog that horse until it’s dead. This often results in a petition of some sort, many of which won’t make any sense. As a developer you can either ignore them for being incorrect, or you can embrace their opinion as being valid but misdirected. For instance let’s say a large number of players complain about the fact that you nerfed a particular self-healing power which was obviously broken and overpowered. You can say the players are selfish exploiters, or you can take a deeper look and wonder if the players are only upset because that broken self heal was the only way to compensate for the occasional player mistake during unforgiving combat. Players choose to post out of genuine frustration, but instead of reasoning why they tend to just rationalize.

4. Repeat

These rules are a bit verbose, but after reading a few hundred posts you can start to filter them really quickly. Most feedback will end up not being useful, which should give you enough time to think about what’s actually important. Basically, if enough sincere people post asking you do something, then there is very likely a real problem. However, the solution to that problem is unlikely to be what they originally asked for (although specific posters with deep knowledge and analytical skills are better at this then your Q/A department. Learn who they are). Extracting out the Valid feedback from the chaff is not a skill that comes naturally, but I think it’s invaluable for making a compelling long-term product. Directly following user requests and completely ignoring them are two different paths to ruin for an MMO, and the solution is to seek the middle path of interprating that feedback without being slave to it.

Posted in MMO Design | Tagged: , , , | 7 Comments »

“How We Decide” If a Game is a 9.5

Posted by Ben Zeigler on June 8, 2009

A few weeks ago I was listening to episode 4 of Out of the Game, and they started talking about “How We Decide” by Jonah Lehrer. At this point I realized I had been given a copy of it as a present, and given that it’s a book on psychology endorsed by Shawn Elliott I put it at the top of my reading stack. I’m glad I did, because I quite enjoyed it. A quick summary is that it’s a more complete, better written version of “Blink” by Malcolm Gladwell. It combines a few really good anecdotes about quick decision making  (apparently when caught in a quick moving forest fire the right solution is to light a SMALLER fire directly in front of you) with an overview of current research into both conscious and unconscious decision making. I recommend it to anyone with an interest in psychology.

The book builds what seems to be a really solid framework for making good decisions in a wide variety of contexts. There are two main processes we go through for making decisions: emotional and logical. They’re good at solving different kinds of problems, and are very complementary. The unconscious brain is very good at pattern matching and evaluating statistical models, and presents the results as input into your conscious brain as “gut feelings”. However, it is bad at dealing with falsehoods or irrelevant information. The conscious brain is good at simulating outcomes to solve problems and at regulating emotion, but it is very capable of thinking too much and incorrectly overriding emotional inputs. The general formula is to use your conscious brain to filter information and monitor emotional state (the best decisions get made while you are in a moderately excited state, as opposed to entirely dispassionate or enraged), and then let your unconscious brain think about it for a bit. The one that “feels right” will more often than not be the right decision.

I wrote last year about psychology and Game Reviews, and there’s a study in How We Decide that directly supports my thoughts. in 1990 Timothy Wilson put together a study comparing the ability of college students to rate jams to the ability of jam experts from Consumer Reports. When simply asked to rate the jams, the college students showed a correlation coefficient of .55 with the experts, which is reasonably high and shows that the expert’s choices matched fairly well with the average college student. Then, Wilson asked a different group of students to analyze why they preferred certain jams using elaborate questionnaires and a wide variety of categories. This group of college students showed a correlation coefficient of .11, which is essentially meaningless. After further study it turns out what happens is that the jam “reviewers” would try to describe individual components, such as “spreadibility”, that didn’t really affect their overall enjoyment of the jam. Then, as they evaluated all of these categories they tended to revise their preferences to match with what they had written in the review. The reviewers had overthought the problem and in the process had modified their initial preferences to match their specific analysis, as opposed to analyzing their preferences.

A similar study was performed by Wilson with paintings (a Monet, a Van Gogh, and 3 humorous cast posters). One group of women was just asked to choose their favorite painting, and 95% chose the Monet or Van Gogh. The second group was asked to explain why they liked the poster they chose, and that group was split 50% between the fine art and cat posters, because the cat posters had more content available for explanation (the subjects were not trained artists).  The book goes through a litany of other studies, all showing that when you try to carefully think through a complicated decision you end up making poor choices.

Let’s see, these studies are all about situations with a complicated decision and the need to generate explanatory content. In all of them, people who explained their decision before making it made worse decisions. Yeah, that sounds a lot like game reviews. If we assume that game reviewers are trained experts (most are) who have consciously trained themselves to be a good judge of games, and are not influenced by extremely strong emotions at the time of rating, their initial gut assessment of a review score is likely to correlate very strongly to the actual enjoyment of a game. However, if a reviewer rationally dissects a game into components they will be likely to rate a game higher or lower then it actually warrants, because it has “bad replay value” or something. So, game reviewers, stop thinking so much about the game and just pay attention to your emotional state while you’re reviewing a game. You’ll make better decisions.

Posted in Game Development, Random | Tagged: , , , | 1 Comment »

UFC 2009 Undisputed: Actually Very Good

Posted by Ben Zeigler on June 1, 2009

I just finished a 2 hour session of playing random people over XBOX Live. This is the first time I’ve played for longer than 20 minutes against complete strangers, and the reason I kept playing was that I was playing UFC 2009 Undisputed (UFC being the Ultimate Fighting Championship, a Mixed Martial Arts competition brand). It’s a licensed game published by THQ and not developed by Relic, which means that by previous evidence it should be a flaming pile. Despite all of that, it’s actually very good. It’s my favorite non-traditional fighting game (traditional fighting games being everything derived from Street Fighter 2) and although it isn’t perfect, it does some things better than any other game I’ve played. There is no game quite like UFC 2009, and in today’s world of refined experiences it’s great to see something new.

The inspiration for this game came very clearly from the UFC PPV/television product. The first components to the quality of UFC 2009 is presentation. First of all the fight trappings are spot on. The graphic packages are identical to what you’ll see in a real UFC fight, and the recorded audio announcing by Joe Rogan and Mike Goldberg is the single best announcer track I have ever heard in a game. It seems to use a mix of existing commentary and specially recorded bits, and it meshes quite well. UFC 2009 has really reminded me why presentation matters in games: it sets the emotional context for your actions. The fighter models and animations are great on average, and they mostly did a good job on the likenesses. I worked my created character up the Middleweight division (losing a few heartbreaking matches to last minute submissions) until I finally fought Anderson Silva for the Title. Bruce Buffer announced that I was challenging for the title, the announcers talked about previous Anderson Silva title matches, and I felt pumped. After I managed to knock him out with a solid uppercut from the clinch, and they replayed the highlights of the match, I felt extremely satisfied. If you’re familiar with the UFC product the presentation of this game will get you in the exact right emotional space to really care about what is going on, and there is nothing more you can ask from graphics and sound.

In addition to analyzing presentation, it’s obvious that the developers (Yukes) spent a lot of effort analyzing the flow and composition of real UFC fights and faithfully converting them into gameplay mechanics. The main components are striking, submissions, and transitions. For striking, they decided to keep it simple, and map each limb to a face button, with directional input to modify the attacks. Attacking without modifying gives you safe strikes that will do some damage, while modifying it makes your strikes more damaging but ALSO more dangerous for you. Best of all, the damage modeling uses realistic physics and collision detection, so strikes do damage based on momentum changes of body parts. The primary way you get knocked out in the game matches exactly with real life: Moving forward for a strong strike while your opponents hits your exposed area with a strong counter. You can also end a fight by hurting them enough that they get knocked down and stunned, where you finish the fight by standing over them and punching until the ref stops you, which is also exactly like real life. Submissions are fairly simple. In most positions you can attempt a submission at any time by clicking the right stick (on defense you have to grab an arm as a counter), and you will then button mash against your opponent. The interesting thing is that your current stamina (which goes down when you attack) has an effect on this, so it’s quite common for a fighter in the game to get overzealous in their striking and lower their stamina, at which point they are more vulnerable to a submission. Again, this is exactly like real life and the submission and striking components directly counteract each other as ways to win a fight.

The other major component is the transitional/positional game. Previous UFC and Pride games had fairly decent striking and submission components, but the positional game in UFC 2009 is extremely deep and satisfying. There are 3 standing positions, 4 clinching positions, and 12 (I think) ground positions. This sounds complicated, but the rules for each type of position are very similar, and they form clear progressions. Basically, each in each position you can actively choose to strike, go for a submission, or attempt to transition (by rotating the right stick). Also, you can defend against strikes or transitions. Striking an opponent who is guarding against transitions is very effective, as is transitioning against an opponent who is blocking strikes. So, the whole system quickly turns into an extremely psychological guessing game. Should you go for strikes that do long-term damage, or do you transition to mount to try and win the fight right away? Or you can block an opponents transition to make them waste stamina and become vulnerable to a transition.  There is no other game I can think of that places this much explicit emphasis on positional control and transitions, and learning this part of the game will definitely give you a deeper understanding of the actual principles behind MMA fighting. The fact that the combat is deep, systematic (as opposed to being overly based on special moves), and maps very closely to actual MMA fighting is why I love the game.

Outside of the core fighting mechanic and presentation, the game is a bit more hit and miss. The career mode is pretty comprehensive and gives you a fairly fun time-management layer to deal with when training your fighter. However, it features some of the most irritating menus in the history of gaming (when it auto saves, which is like every 2 minutes, it pops up SIX different save dialogs, all with slow transitions between them). The Online play is very playable (the slower pace and higher strategy of the game work well online) and has some nice ranking features, but right now an irritatingly large number of people will disconnect immediatly before losing (I have lost 3 wins to this, to add to the 7 I have that registered). These are issues that can be fixed in updates or next year’s version, but the core game is absolutely worth it. Like few other games in this generation of consoles, UFC 2009 Undisputed is a game that could not have existed 5 years ago (when the last UFC game came out). Improvements in physics have lead to a very realistic striking game. Improvements in graphics have lead to the ability to duplicate the presentation of a UFC fight without having to clutter it up with too many meters (do turn on the stamina meter, that cue is too subtle). The improved popular success of MMA and the UFC has allowed the developers to build an extremely deep combat system that can rely on lessons learned from watching real fights instead of having to copy existing fighting games. UFC 2009 Undisputed (name still kinda sucks) will definitely go down as one of the most unique and best-executed games of 2009, at least on my personal list.

Posted in Game Design, Game Development | Tagged: , , , , | Leave a Comment »