If a game feels unfair then few players want to play it.
What players see happening in the game and how they interpret it can often be at odds with what the game maker knows goes on under the hood. Fairness is subjective: a game is only fair if its players believe it to be so. To the player it really doesn't matter how the game engine does what it does, whether it's actually balanced or full of hacks that break balance in their favour. They only know what they see and what they think they see. That can be very hard for the maker to understand.
Designing for fairness sounds like a simple principle, but because it's about feelings rather than facts it's actually very complicated. Particularly for multiplayer games.
Lensing
Video games are different to all other forms of game in two crucial ways: (1) Most of their rules are hidden rather than exposed, and (2) how you play the game is lensed by the intermediate elements of controller and screen. This means that video game players play in worlds whose information quality is deeply imperfect.
A perfect game is one where the player knows all of the available information about the states of resources in the game and the potential outcomes of using them. Chess is a typical example, in that both players know where all the pieces are and what they can do, and so work out all the permutations of play given time.
Most games are imperfect because they involve hidden resources, random chance or unknowable variables. In Texas Hold'Em, for example, no player knows exactly who has what cards and so one of the core skills of the game is the figuring out of likely card dispositions. Athletic events are similarly imperfect because individual athletes are unsure of how others will perform on the day.
However even at that, the rules of those games can be examined. The Hold'Em player knows how the game works, what the composition of the deck is and he can see the whole field of play. The athlete knows the rules of the race, can see the full track and competition is only reliant on his physical ability.
With video games those rules are commonly not examinable. You pick up the joypad and are in the game world, but you don't know for sure that the dice rolls are being handled fairly. You don't know that the virtual deck in the online Poker game actually has 52 cards in it. You don't know what the exact value of gravity in the game is, or whether it's calculating damage for falling strictly according to physics. You can only guess.
This sense of imperfect information makes the experience of playing video games highly personal, and belief comes into it far more than fact. I call it lensing, and maintain that it is one of the six creative constants of all games: the player acts through a controller and perceives through a screen, and his ignorance of the rules is such that his imagination becomes heavily involved. His mind over- or under-interprets his experiences, which can be great (it's basically why numina can be so powerful and games are able to be thaumatic), but also destructive.
Balance vs Fairness
It is entirely possible that a game may be balanced, but feel unfair. For instance if the game contains a lot of opaque rules that the player does not understand, it will simply seem random - and randomness commonly feels unfair. If the game doesn't seem to be challenging the player appropriately, to the extent that it feels as though enemy characters are cheating to win, the game will feel unfair (even if that's not actually what happens).
The player sees a lot less of the frame of the game than you (as game maker) think he does. He thinks he sees behaviours in the game that you know are not there. He extrapolates, and that fuels his belief that the game is fair or not. So, many video games cheat on behalf of the player to correct his misperceptions.
In singular or parallel play fairness is largely a matter of tuning a system against the players' own actions. These games are asynchronous (by which I mean that the outcomes of the player's actions are not dependent on another player's actions), and so players tend to measure their sense of winning against the system rather than against others. You can correct their feelings of unfairness by altering reward schedules, quietly adjusting the damage that their weapons do and other tricks like that, and they are none the wiser. The upshot, in fact, is that they often thank you for "fixing" what they thought was a "broken" game, when in some respects you've actually done the opposite.
In serial or multiple play, however, the tension between fairness and balance is much more problematic. These games are cooperative or competitive (or both, such as in team games), which means that they are more sensitive to initial conditions than asynchronous games. If one team or player starts with an obviously-better position than others, or possesses a wild card or special ability that guarantees them a win, then the game is unfair.
As a result, the immediate instinct of synchronous game developers tends to be to aim for board game-style balance. They develop such that everyone has an even chance from the first turn, and for the complete game to be enclosed only within that scenario. The perception among developers is that this has to be the case or else players will feel that the game is unfair, but increasingly I find myself concluding that this is not necessarily true.
It's about whether players believe they have a shot at a win, not whether they actually do.
Scenarios and Campaigns
As I often maintain, video games are closer to sports than board games. They are commonly dependent on physical skills, played in some form of real time and tend to favour activity over economy. Of course these factors vary greatly, but in the main this is still the case. There is something about the way that they are played live that engages the brain in a sporty way. Also, much like sports, successful video games tend to be more about the campaign rather than the scenario.
A scenario is essentially a formal block of gameplay that contributes to long term progress. It's a dungeon in Diablo 3, a football match, a quest in Farmville or an individual match in a pool tournament. Scenarios are comprised of several tasks (score goals, win hands, complete goals, clear levels), but they also contribute to a larger game: a campaign. The Diablo 3 dungeon contributes to levelling up and exploring the whole world. The football match contributes to performance in a league. The pool match may eventually take you to the final in the tournament.
That sense of campaign matters (and above that again is the level of the epic, but that's for another discussion) because players like to feel that they are building something. This is one of the reasons why, for example, turn-based games are problematic. You can play Letterpress and it's fun. But there's no real campaign, only scenarios. There's nothing to invest in emotionally over the long term, nothing to build toward, and so the play experience becomes repetitive. In some games (Quake 3, for example) this matters less because the player feels as though he is training a physical skill, but more often the sense of repetition quickly leads to the feeling of maximum mastery. Then you just stop playing.
Individual scenarios in most popular sports are imbalanced. When two players face each other across a tennis court they are of unequal skill, experience, equipment and training opportunities. One might be a five-time world champion and the other a 16-year old straight out of an academy. Both will have different positions on the ATP rankings, and so seeding will have stacked their tournament placings. They are blatantly mismatched, and yet every individual match of Tennis feels relatively fair.
Why?
Because the rules of the sport are robust enough to act as a great leveller. It is possible that the new player will beat the champion through pure talent, that the toll of injuries or temper or some other factors on the day will lead to an upset. The same is true of racing, ball sports and athletics. Also of collectible card games like Magic: The Gathering and first person shooters with career modes.
Secondly because both players and audience are aware that tennis is really a campaign rather than a scenario, that it's about the overall winning of the Grand Slam and attaining the number one position. Judged against those goals even managing to play well against the champion and ultimately lose is seen as a form of success. Waiting to see what might happen on the day, seeing the neophyte steal even a set from the veteran translates to a sense of progress. It says maybe not today, but one day.
The trick seems to be to remember that these games are not really about one match (a scenario), but instead about many matches in aggregate (a campaign). Handled well that becomes aspirational rather than impossible, so even though the player may lose this time, he thinks he eventually will. As long as he managed to score a goal, give the opponent a bloody nose, score a head shot against the Quake 3 champion and so on, the game feels fair. Better players become heroes to the lesser ones. Maybe not today, but one day...
It only works if the rules of the game are such that campaign advantage only counts for so much. It's important in soccer that, even with more money than Croesus, Chelsea is still capable of losing a match to a middling team on a good day. This sense that there is always a chance to fell the giant, and that that chance is neither miniscule nor dependent on luck, seems to be what makes multiplayer gaming compelling. For a game to feel fair, the weak player needs to believe that there is a plausible path to victory.
Even if she does not achieve that victory in the first game, she'll train, get better, level up and this is a natural part of mastery. Eventually she'll be the one being challenged by new young players. The game is unfair if she believes that that will never happen. If it appears as if the route to that success is so vast that only cheating players ever make it. Or that practise does not seem to help her improve. Or that it's all about how much money other players spend to win.
Understanding these factors is, I think, at the heart of making multiplayer games sing, making them work in free-to-play economies and ultimately making them long-term experiences worth investing in.