Monday, January 31, 2011

Game difficulty (Part 1)

Some people love a challenge in their games.  Some people delight in pushing their skills, tactics, management and reflexes to the limit when they play games so they can have a sense of satisfaction when they overcome the great difficulty of a game.  This is why we have games like Super Meat Boy, Ninja Gaiden and Demons Souls'. These games have a difficulty above and beyond the standard fare of games around, and there are players who flock to such titles with glee. On the other end of the scale, you have gamers who really don't have the patience to perfect their skills, or the dedication to learn and study complex tactics. They're playing the game for the fun of the entertainment medium, and possibly to experience the spectacle and/or the story of the game. And of course, there's a whole gamut of players in-between.

Now, nowhere should a judgment be made on what type of game or type of difficulty is "best". There should be games to provide entertainment to cater for all tastes. Some people simply wish to play the game for fun, whereas others play for the challenge and to push their own mastery of the controls. Designers (and players) should not denigrate a player no matter their skill level. So to any elitists out there, that's right, even n00bs have a right to play games.

Ninja Gaiden: Designed to be difficult.

So here's a little message to players: Don't be abusive to other players in a multiplayer game. If you don't like them or they aren't very good, instead of unleashing a tirade of vitriol, how about you help them instead?  Players of DotA, and its clones like Heroes of Newerth and League of Legends, please heed this - these communities are among some of the worst and most obnoxious players when someone on their team is new. In the case it does disadvantage all players on the team, but just remember, everyone was a n00b once. Screaming or typing obscenities at other players only serves to reduce the amount of players in your game of choice, not to mention helping perpetuate bad stereotypes about gamers. Don't be an Angry German Kid.

Okay, so with the community service announcement over, it is important for designers to consider what can be done to make a single game appeal to lots of different people on this "desired difficult curve".  No matter how difficult you make a game, some players are going to complain that the difficulty is not right. Some players insist the Nightmare difficulty in Dragon Age is too simple, while others struggle on normal. Some declare winning a game against a computer opponent in RTS games like Starcraft ridiculous, yet skilled players can handle multiple computer opponents this with ease. Keep in mind that when you're making a game, you're never going to make everyone completely happy about the difficulty of your game.

Yet this doesn't mean that designers shouldn't try to cater for players. Instead, they should help the player select the difficulty level that is right for them. As a designer you want to make your game fun for as many people as possible, and allowing players to better select their desired difficulty is an extremely effective way to do that. Over the next few posts, I'm going to discuss the implementation of difficulty in games using gameplay mechanics, covering the good, the bad and the ugly.


  1. I've spent a hell of a lot of time discussing difficulty in games with a lot of folks and the best example we could come up with was Condemned: Criminal Origins.

    In every single encounter in that game, you have the tools to beat the enemy and there's no secret boss mechanics or other trickery that requires to die (often multiple times in some games) to learn how to defeat them. If you die it's because you stuffed up, not because of some cheap mechanic.

    Games which require progression by trial and error (learn each attack pattern/weakness/enemy location through dying) or progression by grinding (repeating the same thing over and over to level so you have the abilities to defeat the boss) are both concepts that people misrepresent as "difficulty". They have nothing to do with difficulty. They have to do with frustration, time wasting and needless repetition.

    Similarly, putting 45 minutes to an hour between your save points (looking at you Ninja Gaiden - and every other Japanese game ever) doesn't make your game more difficult. It makes it more aggravating. Especially when you combine it with the two other problems of trial and error and grinding.

    A perfectly balanced difficulty curve should always have you going into conflict/scenarios with a feeling of uncertainty, but also allow to emerge feeling triumphant and relieved. And you should always have the capacity to get past it on the first try, even if it is just by the skin of your teeth.

  2. And before anyone jumps on me - I love games like Super Meat Boy where the challenge is in repetition and perfection. But Super Meat Boy counterpoints this by having short levels. It's rare to lose more than 20 seconds in a Super Meat Boy level when you die. Unlike many games where your punishment for failure is measured in minutes.

    I did read somewhere (and got into an argument about it with the person) that the only real punishment in games is loss of time, so that's why savepoints should be so spaced out. My response is always: "Why waste the players time? And if the only way to provoke a response from a player is to steal their time, then you have made a shitty game."

  3. My favourite topic!

    I think it's important to recognise that while there is a significant delta in player ability, in a lot of cases that delta is widened by poor game balance and wonky learning curves.

    Dragon Age: Origins is a salient example. There are concrete reasons why some players struggle on normal and others are bored by Nightmare, and they're all theoretically avoidable. Certain builds, behaviors and the availability of unlimited restorative items trivialise the game. This means whether you fall into the "OMG DA is so hard!" or "Nightmare, more like lolmare" bucket depends more on whether you know about these things than your execution.

    Meanwhile the effects of the difficulty levels are purely quantitative, doing little to blunt the assault of an informed player. It doesn't matter if the mobs are theoretically more powerful when they're all CC'd. It doesn't matter that the boss hits 10% harder and takes 10% less damage if the player is chaining potion cooldowns on the tank and abusing dual-wield autoattack.

    DA:O's difficulty, and the delta of player skill, could be massively improved by balancing talents/spells, providing less of them to choose from at low levels (reducing the number of possible bad builds), and defaulting companions to stronger builds. If it then used its long playtime to steadily introduce and teach more sophisticated encounter concepts, it would be even better (and more fun!).

    On the flip side though, DA:O does something really great for difficulty - by offering 6 Origins, it effectively gives the player 5 restarts with fresh content. That's five opportunities to approach the choices of character creation and the early game with an informed perspective, rather than the ignorance with which you make your first choices of class, talents, skills and attributes.

    I could easily go on all day, but my key point is that most players can learn to play well, and most players will enjoy improving their skills should the game convince them those skills are useful (and here the fiction layer can play a vital role). There is a real and unavoidable spread in player ability, but the distribution we see at first glance is artificially exacerbated by poor balance and poor learning curves.

  4. Some excellent points you've both raised. I could not agree more that forcing trial and error, vast spacing between save points (heck, having save points AT ALL - let the player save whenever they want), and wasting the players time are all things I'd consider bad design.

    As for DAO's difficulty, and even RPGs in general, a lot of the difficult variance is due to player approach. Again, it's often the difference between a player who chooses to find the optimal builds and most powerful combination of spell/abilities/tactics versus the player who designs their characters around roleplaying decisions for their character build. I found this was a huge issue in NWN2, particularly when characters reach level 15 and above. As you've hinted at, the difference in combat potency between an optimised build made by a min-maxer and a casual player is extraordinary.

    Removing the number of spells/abilities available may make the game easier for the player who does not min-max, but at the same time it makes the game far less interesting for the players who like to min-max, giving them less flexibility and fewer options to pursue to find various "power builds". I'm not sure I have a solution to that problem just now, but I'll definitely be keeping it in mind as I pursue this series of posts.

  5. @AmstradHero - speaking as someone who likes to min-max, even if catering to me was somehow worth wrecking the balance of the game (it's not), it's not very useful at level 1 anyway. You can't usefully min-max a half dozen choices you need to make before play begins without having played before or researching the optimal build online, in which case there's only one or two real choices and the number of bad builds you're ignoring doesn't matter.

    Much better to introduce the choices at sensible points on the learning curve so that the player can make informed decisions. WoW's a good (if not perfect) example - you start by choosing your class, some of its abilities are introduced to you one by one over the first 10 levels, and then you choose your talent specialty at level 10 when you have at least some idea of the different ways your class can play. Ultimately you can end up simulating out stat weights and priority queues to optimise DPS, but you start out by choosing your class and hair colour. And while the delta between Paragon and your average PuG is pretty big, players we'd ordinarily think of casual/roleplayers often turn out to be very capable in a raiding environment.

    Furthermore, while it's perfectly appropriate for a gamist, performance-focused player to outperform a roleplaying/concept focused player, the assumption that the difference must be huge is an assumption that RP appropriate builds must suck. There's no reason for this to be true in a balanced game. This is something I've dealt with much more in tabletop, but I've noticed there that the problem isn't non-min-max characters are necessarily bad, but their competence in a poorly balanced game is essentially random. Two players might both choose builds from a purely concept-oriented perspective, and one ends up a demigod while the other is an active liability to the team. Neither player is at fault, but the system certainly is.

    This can be fixed by balancing the game - not necessarily PVP tournament grade balancing, but at least making sure there's a reasonable floor on character competence (this is the way D&D4 is balanced, and it's really great when you have a mix of player skills at the table) and that the most RP-appropriate builds for the setting are also quite strong.

    In DA:O, we also have the problem of it really being a handful of behaviours that really widen the gap. I like to think of players as "naive" or "informed" in a Dragon Age context. If they've read on a forum that there are unlimited healing potions available via crafting, they're suddenly much more capable without significant roleplaying impact. Likewise the only dominant mage spell with serious RP impact is in the Blood Mage line. The rest, while you might randomly miss them all choosing purely for flavour, aren't too difficult to fit a few into a concept build once you know how valuable they are.

    On another note, I disagree with you both that trial and error is bad. It's fine to kill off players as long as they learn something from that death and (as with Super Meat Boy, FFXIII etc), you give them an opportunity to exercise that learning as soon as possible. Autosaves before bosses are good :)

    Finally, for this post series I'd suggest checking some of the great material online on learning curves in games. In particular anything by Dan Cook ( is gold.

  6. @Jye - I agree that at level 1 that min-maxing makes little (if any) difference. But as you progress through levels in an RPG, build viability/strength becomes more dominant in terms of character's (and hence a player's) combat power. If this wasn't the case, there would not be so many character builds and guides for Diablo, WoW, D&D, and just about every RPG in existence. I think RPGs probably deserve at least an entire post on this topic, because RPGs have the unique situation of relying on the player's mathematical (or research) ability to solve the problem: "How can I maximise my DPS?" As for WoW, I agree that casual/roleplaying gamers who can't perform in PvP are capable in a raiding environment. When I played, I was on scorch-duty as a fire mage, which meant I was less successful in PvP compared to some other raid members (though latency didn't particularly help my cause in PvP), but I could contribute quite strongly to a raid. Even within WoW, you have raid-spec and PvP-spec builds - it's breaking down into a calculation even within sub-areas of the game. And even if you are the most skilled player in the world, you can't expect to finish the hardest content without an appropriate character build and the necessary equipment.

    Trial and error might not always be bad, but in many cases it is. In the vast majority of cases it means that the player cannot reasonably be expected to overcome a challenge the first time they face it, even if they have a great degree of skill. In many cases trial and error is implemented in a manner such that the player *will* die the first time they attempt an encounter simply because they cannot react to the situation as it unfolds. This forces the player to try again and use the knowledge they gained from their previous attempt. Sure, they're getting the feeling that "oh, I learned from that encounter, now I know what to do", but used repeatedly (and in games that utilise this type of tactic inevitably do exactly that) it serves merely as a means to artificially inflate the difficulty of the game. Out of all the boss encounters of the Xbox release of Ninja Gaiden, the one I remember the most fondly was the one with the undead dragon. That's because it was the only boss I was able to defeat on my first attempt - albeit just by the skin of my teeth. Being able to adapt to the situation on the fly and overcome it meant that I got a genuine sense of enjoyment and triumph from my success. That meant far more to me as a player, and gave a much greater sense of enjoyment than defeating bosses for which it took a dozen or more attempts to refine my strategy and perfect its execution.

    I agree that designers should want to make their players fail from time to time. If the player goes through a game without failing once, then the sense of danger and achievement is lacking. But forcing the player to continually repeat challenges is bad design - unless you're deliberately trying to force players to play your game for longer. In this case, WoW's raids achieve their goal marvelously. They force players to die repeatedly until their realise that they need to change/refine their tactics and/or acquire better gear, which keeps them playing for longer. Now don't get me wrong, I have a great deal of respect for WoW, but it's important to understand how its design is driven by the need to maintain player subscriptions. Trial and error is bad design because it *ONLY* caters for the "high-end" players who want to be forced to perfect their play in order to succeed.

  7. @AmstradHero - what I meant is having a lot of options to min-max at level 1 doesn't increase the fun even for min-maxxers (or at least not enough to justify having a broad range of newbie traps). It's better to introduce those options over time, so that players of all preferences are armed with at least some experience when they make their choices. Starting with one option doesn't preclude sophisticated builds later on, so there's no necessity to [b]start out[/b] catering to the min-maxxers.

    Re: trial and error. I tend to have the opposite feeling for one-shottable boss fights; when I encounter one I usually feel it was a failure as an encounter. This tends to happen because its mechanics are not sophisticated, or its tuning is such that you can ignore them. That makes it to my mind forgettable (and indeed I can't remember that Ninja Gaiden boss, but I can certainly remember the first guy who kicked my butt until I learned how to play ^_^), and in the case of a modern game, not really worth the price of the art assets used to create it.

    The interesting thing to me though is you're assuming it's poor design to oblige a player stay longer on a set-piece encounter. Assuming we're talking designed content (ie not procedurally generated), and the player is learning and improving on each attempt (ie our mechanics are well designed and don't generally seem arbitrary or frustrating), why wouldn't we want to get more fun playtime out of our encounter? Isn't that an efficient use of our resources, particularly development time and art?

    It's easy to imagine a boss fight that lasts 30 minutes rather than 5 in this case, multiplying its contribution to playtime and fun without outstaying its welcome. A few of those (at the cost of an extra couple of hours development and testing on each), and someone in our position could double the length of their module. Even without a subscription fee to collect that's a good result.

    It's worth noting that wowprogress has ~60,000 guilds killing Lord Marrowgar in 25-man, and ~90,000 more in 10s. Even allowing for wonky statistics and overlap, that's many hundreds of thousands of people. This style of challenge can scale down a very long way from Paragon, and I think it's an error to discard it as a viable option.

    Not saying it can't be done poorly though :) Even I hate it in cute little Recettear, because fat-fingering can waste your entire dungeon run. The problem in this sort of context isn't killing the player per se, it's "killing" meaning too much more than an encounter reset.

  8. @Jye - Funnily enough, I think we do actually agree on some points here, though we are coming at this from different angles.

    The issue isn't if I'm forced to replay a fight - the fact that I can't get through something the first time helps reinforce that the game I'm play isn't a cakewalk. As funny as it sounds, I do actually *like* losing fights. It forces me to improve. However, I consider it to be an issue of poor design when the game has not given me sufficient information to be able to win the first time around. Losing a fight because I as a player lack significant skill is fine. And this is because of the different types of players out there - some players don't want to lose repeatedly in order to overcome a fight.

    Case in point: I received a couple of very angry complaints from players regarding Alley of Murders because they couldn't beat one fight... but they didn't attempt to change their tactics after being defeated a few times.

    WoW (or rather MMO raids) are a bit of a different scenario because of the planning/research required. The guild is either working out how the raid has to be done, or refining their teamwork and tactics in order to be successful under the guidance of the raid leader. This division of effort allows MMOs to cater for more than one group - but if this is translated into a single player experience, it's likely going to alienate the more casual player who doesn't have the same dedication as the type of person who becomes (or could become) a raid leader.

    Imagine playing WoW as a single player game, without any raid assist macros. That would force the player to identify exactly what is going on and demand they know how to react to it.

    I agree that some fights should be made such that the player is highly unlikely to succeed on the first attempt. But it should be done in such a fashion that it's due to the player's inability to adapt to the new situation rather than enemy capabilities that are certain to catch the player by surprise and kill them the first time around. It's a fine distinction, certainly, but an important one.

    You also hit on the other point mentioned earlier - if trial and error is going to be forced upon the player - the cost in terms of time and/or resources should not be significant. The player has already suffered from being killed, they don't need to be further aggravated by having to start a 30 minute encounter from the beginning. (Though once again, WoW/MMOs provide the counterpoint to this to keep people coming back)

  9. "Case in point: I received a couple of very angry complaints from players regarding Alley of Murders because they couldn't beat one fight... but they didn't attempt to change their tactics after being defeated a few times."

    The question really is whether that population is big enough to justify neutering the challenge of your game, and whether it can be reduced by good design (ie gracefully improving the player's skills as they go).

    Or if not neutering the challenge - and this is obviously something you're going to touch on in the upcoming articles - is it worth the cost and effort of implementing difficulty modes that reach down to that level (or up to Heroic Lich King brutality)? I'll be interested to see what you have any specific thoughts on doing that for DA:O, since the base difficulty levels really don't get the job done.

    (and re: WoW as a single player game. I made that module! Classic Week is 6 hours of that :) Admittedly its difficulty isn't flexible enough, being far too hard for many DA:O players but still a bit easy for the top end. That's why I'm starting to use encounter complexity difficulty modes, which hopefully will help Broken City work for both those used to the very easy Fragments as well as players who've enjoyed the challenge of my other modules).

  10. Classic Week was WoW as a single player game, but as you so aptly pointed out, the difficulty was far too much for many players. I agrre that the trade off of the time required to implement difficulty versus the increased benefit in terms of player accessibility is definitely a tough decision to weigh.

    I think one of the most difficult things in terms of "fixing" Dragon Age's difficulty for new modules is that many of the aspects that make the game easy for the knowledgeable players are part of the core mechanics of the game, and I don't like modifying those in mods. I don't think it's a matter of simply throwing more enemies at the player on the highest difficulty (though correct me if I'm wrong here) as I don't think that would actually make a whole lot of difference for those players.

    Thus I think for DAO it will probably boil down to having to change the reaction and behaviour of enemies to combat the advanced tactics utilised by veteran players in order to cater to them. At this point, it then comes back to the question of the potential audience size and whether it will be sufficient to warrant the extra effort required to implement that. Of course, the specifics of *how* to implement those changes... well, that would potentially warrant a post and/or a length discussion between modders and experienced players.