Level playing field: Punish the better player or enhance the worse player?

Question

I am making a 1 vs 1 game where each player controls a car with an identical joystick. The objective for the players is to reach a goal destination before the other player.

The players of the game are placed into categories based of the skills controlling the car e.g. Skill level[1,2,3,4,5] with 5 being the highest skill level.

The players will not be intended to jump up and down in skill levels when playing the game. After a given period, e.g. a week, the skill levels will be updated. This will not be shown to the players.

The objective for me is to make every combination of players in the 1 vs 1 setting be a 50/50 win loss chance for the players.

Example: Player 1(Skill level 5) vs player 2(Skill level 1) should each have 50% change reaching the goal destination first.

I have not been able to find any research on whether the best practice is to limit the abilities of the good player or enhance the abilities of the bad player or maybe a combination.

The idea is to have disabled adolescents playing against their able bodied friends or family members. It is therefore not a possibility to not let good players play against bad ones. Data from field tests have shown that the disabled adolescents perform worse than able bodied consistently due to spasms.

I am looking for input on how to go about with this; research, your experiences, how other games are doing it? Is there any common recommendations for this?

Answer 1

Match making

If you can keep updated skill metrics on the players, then you are able to create a system that can peer people of similar skill.

Granted, this could lead to starvation (and by that I mean players unable to play because there is no match availed) and will not fly on couch gaming.

I believe this will not solve your particular case, so I will not elaborate any futher.

Handicaps

I would complain on any handicap system based on the skill of the players. I believe Bálint expressed the issue better than I could. With that said, handicaps in either direction are viable if they are on player volition.

That is, if the players go to a menu where they select what handicap they want and they are all conscious of that choice, then it can work. This is in particular useful in couch gaming with family or friends, because they would know each other and – given the motivation※ – they will find some setting that makes the game fun for everybody.

※: That is, if what they want is to win instead of have a good play session, then it will go wrong.

As I said, handicaps - as long as placed by player own volition - will work in both directions. That is, either making things harder for the "good" player, or making them easier for the "bad" player. However, I would side on making things easier for the "bad" player because...

Catch up mechanics

You can give boosts to the player that is behind (regardless of skill score) in order to help that player get back to the front of the race. After all, it is when players are nearby in the racetrack that the game play is more interesting.

A traditional example will be Mario Kart. In the game, players that are behind will receive better power ups which will allow them to catch up to the players that are at the front.

You might be interested in the concept of rubber banding. Which is how we call the idea of adapting events to fit player performance. Read more about The approaches to Dynamic game difficulty balancing at Wikipedia.

Address the issue

They all have cerebral palsy and is therefore heavily affected by their spasms when using the joystick

Perhaps the solution is simply to smooth the input. You can:

• If the joystick moves too abruptly, ignore that.
• Let the joystick input be a target angle, but limit the angular velocity at which the direction can change.
• Modern cars in real life already adjust the sensitivity of the input as a function of the current speed of the car, you can do that too.

Asymmetric gaming

Mario Chase from Nintendo Land is an excellent example of asymmetric gaming.

I would also start with the handicap approach. However not by giving some advantage or disadvantage to players, but by offering a different input mechanic※, and make it optional.

※: That does not imply to create or use a different input device - however, I admit that would be useful, I also recognize that might not be possible. When I say input mechanic I mean the way the game interprets user input.

For a first person shooter, consider the effects of auto aiming... it allows a player that has struggle with fine motion to enjoy the game. However, it does not offer means to improve the motor skill. Instead, I would suggest the following:

1. The player aims in the general direction they want
2. Then the game changes to an aiming mode where they can fine-tune the direction. The aiming mode may help the player by having:
   • Zoom
   • Less sensitivity
   • Snapping to important targets
3. The player shoots (or aborts out of the aim mode).

I suggest doing something like that for your car game. Let the person choose the general direction and then fine tune and apply the change. This means that for a turn, they will change mode ahead of time in the general direction of the turn, and then when they reach the curve they fine tune and apply. In fact, you could let the game gradually apply the turn as appropriate, after all, the result of applying the turn should not be to put the car into a circular path.

I have a vague memory of hearing of a game that had such mode, I have been unable to find it.

The good thing of this approach is that it is a difference in kind, however I admit this is just a rought sketch and I cannot guarantee it will work. If you take this idea, or any other similar one, you will have to take feedback from your players and iterate over it.

---

Did you consider making a different kind of game instead of a racing game? I think your objective is help these people, perhaps a different kind of game will lend itself easier to the kind of balance mechanics you need. May I suggest a “shoot’em up”? - I strongly belive that a cooperative game will suit you better.

---

Addendum

You may also be interested in Incomparables, Perfect Imbalance, Co-pilot mode, Assit Mode and The Curb Cut Effect.

Answer 2

Neither is good. Why would you try getting better if all it does is give the opponent bonuses or makes you worse? You might as well just make a casino game if only the chances decide who wins.

It also makes good players less likely to improve. If they steamroll their opponents without even trying, then they don't get any sort of feedback.

Don't let bad players play against good ones. Create some sort of automatic ranking system, where good players play against other good players.

Answer 3

Your specific scenario is to allow fun competitive play between one or more players with cerebral palsy and other players without that condition. You mention that the main disadvantage is that of tremors, which effect the player by reducing the effectiveness of steering.

Perhaps you could implement a game mechanic which simulates tremors/loss of control for players, and which typically effects steady-handed players more than non-steady players in a believable way. Example: would the enemies from Borderlands like it much if everyone in a race were driving in a non-chaotic fashion? Perhaps they would try to "encourage" racers to be a little more interesting in their driving. The goal would be to take something that might be considered a disadvantage and turn it in to part of the game play. A challenge, rather than a hinderance. I honestly don't know how viable this strategy is - I'm just throwing the idea out there.

Another idea is to process the input of each player and come up with a "steady-hand profile". The least-steady profile in a given match could be used to automatically introduce simulated tremors in to the input of other players. This would create a direct correlation in disadvantages. It would also be a very intriguing experiment in machine learning.

Answer 4

When developing AI for arcade-style racing games, it is a common design principle to focus on the AI, giving a disadvantage (speed reduction) to a leading AI and an advantage (acceleration increase) to a trailing AI, thus keeping all players (although it would be 1 human player and multiple AI players) in close positions. When the AI is slightly ahead of the player the AI's disadvantage is then removed so that the player still has to try to win (and gets the emotional reward of achieving a win).

The difference is only relative, so whether the leading AI is given the disadvantage or the trailing player is given the advantage, there is not much difference to the competitive experience. The important thing is balancing it so that skill is still required. The point of the handcap/bonus should be to narrow the skill difference, not eliminate it, then the game will still be fun and it will be a close game that either player can win depending on how the race goes.

So as for whether to give a skill reduction to the more-skilled player or a bonus to the less-skilled player, I suggest making it a factor of the game itself, not of the player's historical wins/losses.

What happens after you balance the skills and your better players are losing half of their games to your worse players, will their skill rating be lowered while the less-skilled player's skill increases? If a 1/5 player keeps playing a a 5/5 player and the score is 50/50, will both players' skill ratings be adjusted to 3/5? Then they won't have the advantage again, right?

Are the skills dynamic, based on historical wins and losses, or will they be set before playing the first game and never change?

If you do it correctly you might find a balance, but I suggest making it a mechanic of the game that leading players are slightly penalized and trailing players get some speed boosts that they can activate. You can do this in a way that is not obvious to the players, such as building up a trailing player's boost bar more quickly, so they will be able to boost more often, etc, while the leading player's handling, acceleration curve, and top speed, may be slightly decreased, then as the players get closer together during a race the differences become less.

This way of keeping races tight is good in general, not only in your situation. There are people of many skills, or someone might just get unlucky and crash early, and it's always fun to have a close race.

Aside from these balancing issues, design your game so that it tends toward close races. For example, will a player be severely punished for running over the boundary with a 15s wait time for a helicopter to come and pick their car up and put it back on the track? Or will they just bump off the side and receive only a small speed reduction for their mistake?

Answer 5

If the question is only about bonus or malus, bonus is the way to go, as outlined by FFN in a comment:

Since your system is hidden from the player, this probably won't be relevant, but it is a consensus in the roleplaying world (where bonuses and penalties are visible to everyone) that giving bonuses is by far the superior alternative. If you hinder a good player, he feels cheated because you are punishing him for being good, while he did nothing wrong. If you help a struggling player, instead, you are both empowering him (which feels good) and making the playing field level. Now the worse player has a chance to win and feel good and if the better player wins, victory is sweeter. – FFN

Ideally that bonus isn't just a change of the car's stats, but something more controllable, like a faster cooldown on user activatable special abilities. It can be hard for the user to get used to the behavior of a car if the car's speed and acceleration change every week, so that's something to avoid.

My advice is to think beyond bonus and malus. There are some alternatives.

• In racing games, poor players tend to perform less bad if all players are given slower cars, primarily due to less frequent fatal crashes.
• Features like auto brake (reduce speed to less than "suicide" before a turn), or auto/manual gear can and will affect proficient and novice players differently.
• Some racing games feature a built-in drawback for the car in the lead, such as a user launched homing missile or lightning strike that seeks out the car in first place (IIRC Mario Kart, and Blur), or the ability to summon doom on the track ahead (Split Second).
• NPC cars crossing intersections are a great way to hide preferential treatment of players.
• Some racing/fight games have a large yet non-obvious element of chance involved, which allows weaker players to win occasionally. I'm thinking of the original Carmageddon where the first place car getting wrecked in the first turn by the cars in the back was a rather frequent occurrence.
• Alternate game modes like the ones in Flatout also allow weaker players a chance to score

Answer 6

Bálint gives an excellent answer, and I'd like to add why I think so.

You'll find that in many games, such quantitative changes as you suggest are an extremely poor substitute for the qualitative difference in play style, especially as a broad/sweeping/unreasoned approach that you suggest, and to the degree that you suggest. It tends to either overkill good players who are being unfairly limited in crucial aspects of play, or not do enough to solve that which makes them highly effective at play. This sort of catch all handicap system has been tried since the early 1980s at least and it has very rarely if ever been shown to be a useful approach. If applied very selectively in limited areas of the game mechanics, for example multiplying a noob FPS player's health by double or triple compared to a moderately better player, it may be useful.

Balancing this brute-force balancing attempts to work across all the differences in human skillsets is nigh-impossible. So again as suggested by Bálint, match players against others by proven skill.

Possible solution

In order carefully limit where you apply bonuses / penalties as suggested...

I can suggest a sort of intelligent approach to evaluating playstyles, and bolstering key values accordingly. For example, if it is found that a player reacts to the need to turn, too slowly, try to give them some sort of a preemptive hint that they need to start turning, which others don't get.

In Doom, a similar system was used to overcome the limitation of not being able to look up or down - if there was an enemy up or down from the current crosshair position, the game would auto-aim up or down, making the game easier to play as no-one had to worry about vertical control (this was especially important given it was originally written to be controlled with keyboard only).

This will probably require architecting things in a way that accommodates this data gathering on weak points in each player's skillset.

Answer 7

Odds makers have faced this problem since the beginning of team on team sports, and they devised a simple solution:

Beat the spread

Due to the difference between the two skill levels, the goal is not to reach the end first, or the score the most points. It's to have a spread larger than the system says you need to win by. In games with a scoring system this is simple - you need to beat your opponent by X points to "win the match." Your matchmaking rating will go up by 0 points if you get exactly X points more, and go down if you get less than X points.

In a racing game, you would have to beat the opponent by a certain time. You could do this by giving the weak player a head start, but that's going to sound like a penalty (although it will make a more exciting finish because in theory they arrive at the finish at the same time.) You also could start them at the same time expecting the strong player to finish first. The weak player then has to make sure he finishes before his spread runs out.

In a deathmatch, this is going to be difficult. You could say you need to win with X amount of health remaining. Much like racing, this is similar to starting with a handicap.

At the end of the day, there's no way around the fact that you're trying to make a fair fight from an unfair fight. I play chess and beat a guy on Chess With Friends 5 times in a row. He challenged me again. I beat him without moving my queen or rooks at all in the whole game. Short of me getting super drunk and playing him, or him studying his keester off, this fellow will never beat me at chess. Only by imposing some artificial limit (maybe next time I only get 1 bishop and 2 knights... probably would still beat him really... whatever) are we going to be able to make this fair.

Answer 8

Reframe the problem

In most 1v1 games like you describe, the goal is to reach a finish line first or otherwise do something the best. If you do the thing better than the other person, you win. That kind of competition is a tried and true approach. If you are trying to adjust it so that everybody wins 50% of the time, it requires a different mindset. If the competition is between me and another person, and no matter how hard I try, I always win precisely 50% of the time, and no matter how much I sandbag, I still win 50% of the time, I'm going to get bored.

The game needs to be rephrased as more of a battle with myself. I can always get better. I can always improve. If I happen to be improving alongside my disabled friend playing the same game, so much the better! I need to still keep the goal of "winning," to keep the emotional attachment to the game, but I need to decouple that from the deeper game.

My approach would be to to start with a simple racing game, where the first to the finish line wins. Then, over time, try to encourage the players to engage in activities that pit themselves less against eachother and more against themselves and the environment. Maybe add some mini-goals like trying to do a perfect 720 roll off every jump in the game, which is only given to people when that would be a good test of skill. Perhaps for those of lower skill, you might give them a goal for just a 360. Never give both players the same test on any given race. Maybe one player has to do precise 720s, while the other gets credit for drifting around the corners as much as possible. That makes the final race-time less comparable, because both players are doing slightly different things through the entire race.

If you can convince them to start thinking that way, then you can start playing pareto-optimality games. Offer "upgrades" to your vehicles which make them more suitable for the minigames, but which might negatively or positively affect race times. Maybe there's an upgraded differential that's better for drifting, but happens to decrease acceleration or something equally crazy.

The remaining key is to listen to the players. Get a sense of how much they are "digging" the minigames. If they're really down with them, you can 100% balance the final winner algorithmically without decreasing player happiness, because they aren't focused on the winner of the race. If they aren't doing any of the minigames, and just racing pedal-to-the-metal, then you aren't going to be able to optimize like this without a player noticing and refusing to play. In such a case, you will need to entice them to play the minigames. Pick minigames which don't affect the final race as much, and offer those instead. Offer slight bonuses which make your race-time shorter if you do the minigames properly (like speed boosts or whatnot) to get them interested in the games. Once you figure out that they're having fun with the minigames, you can try to decouple them from "winning the race" again by offering more interesting games of skill.

Answer 9

What you're asking for isn't possible, and is going to end badly.

The big problem is hidden, long-term-stateful leveling effects. These are going to be testable and thus noticeable by players. For example, if players accidentally or purposefully switch controllers without logging in/out of their profiles, they're going to immediately notice. See Ryanfae Scotland's comment:

Please don't make this secret advantage completely hidden from players. I can see what you are doing and why you are doing it, and your heart is in the right place, but can you imagine how soul destroying it would be if someone finally found a game they were really good at, could beat their non-disabled and (in their eyes) more talented brother at only to then find out the game had been giving them an unfair advantage because they were actually so bad at it? I reckon it would be heartbreaking to find out.

Instead, if you want to do something like this and make it work, follow the "Mario Kart approach". Don't secretly record and respond to long-term play habits and results, but instead offer immediate, visible risk for being in the lead and immediate, visible advantages for being behind. In your case where disability might seriously impair one player's ability to control the game effectively, you probably want to make the effects more pronounced than in Mario Kart. But be careful not to make them too "gameable", and whatever you do, don't try to detect spasms or other input that you think might reveal who the disabled player is. Unless the able-bodied player is known to be friendly and empathetic, you must assume there will be malicious players who will mimic/mock your disabled players' play style to get advantage, and that's going to me the whole thing an extremely not-fun experience.

Answer 10

Based on the way I read the post, the game's purpose (first and foremost) is to make it fun for both parties playing. I propose an additional goal of helping a less challenged player to feel some of the difficulties a more challenged player would experience.

Handicap/Enhance based on the other player's capability.

Assuming you can determine the line for a "perfect" run, you can determine how much a player deviates from this perfection. We will call this "drift". You then adjust the controls to mimic the drift that your opponent would typically have through the race.

How you record this drift will be specific to your game design. One method is to look at key points (curves, narrow roadways, obstacles) and find how much far player is off from the perfect line. Another method is to record average deviation throughout the run and find an end average.

Say, based on the average of the past 5 games, PlayerOne is typically off by 30 points/pixels/feet/(whatever measure) and PlayerTwo is typically off by 10 points/pixels/feet. In a head to head game, PlayerOne will experience a "pull" toward the perfect line and PlayerTwo will experience a "push" away from the perfect line. The amount and exact methodology of this push is specific to the game.

You mention specifically that there are times where a more challenged player may have sudden jerking motions. These over-reactions can also be recorded and an adjustment made on both sides. If a PlayerOne averages (say) 3 of these sudden deviations further away from perfect in a typical run, that can be recorded. When doing the next run, the game can look at a sudden deviation away from perfect and (for a time) ignore (or minimize) that type of input, drawing the person back toward the perfect line for up to (that average) 3 times. Additionally, PlayerTwo can have sudden deviations pushing him away from the perfect line to mimic the challenge that PlayerOne would experience. An in-game explanation would be "a sudden gust of wind" (perhaps even add the sound effect?)

Adjusting the "push" and "pull" through user testing is important to balance the game and not make it overly frustrating for either type of player.

Also realize that if there are two players with the same average deviation, the calculation causes there to be no correction to either one. This still keeps the playing field level automatically.

Answer 11

Different Classes

Your game is symmetrically balanced, which leads to the better player winning. A solution is to create classes of different cars. Have classes that don't take a lot of skill, and doesn't really punish the player. And create skill-based classes that are difficult to utilize optimally and/or punishes harder for mistakes of skill.

This will also have a side effect helpful to your situation: a "skilled" player may only be vary skilled in one or a few classes. If you make certain classes have an advantage against certain other classes, insomuch as to cause a 50-50 winning-losing chance for that combination, a you get closer to your goal. The skilled player either chooses their car or picks a different one he's not as skilled in. Either way the good player would be uncheatingly challenged, and the worse player has a real chance.

So classes seem great right? There is one major disadvantage: it is hard to create such a system, and much harder to create it right. This will require lots of thinking and creativity, and will never be perfect. Additionally, you will have to prevent new players from using very difficult cars, usually by a car unlocking system. And I am sure there are other disadvantages and advantages I missed or didn't think of, that will probably be pointed out in the comments of this awnser, but with dedication, you can overcome or mitigate these problems.

Answer 12

Cort Ammon's answer touched on the idea of playing against your own best, but I'd like to take that a bit further. Its success will depend on how important the theme of "racing game" is to your project.

One example would be, instead of a race game, a chase game. The lead car has a headstart but a slower car, and would be played by whoever chooses to (assuming it's a group of friendly people, the preference would be for it to be chosen by the strongest player). The chase car is faster, but due to being driven by the lesser-skilled player it will not catch up in a determinate time period, but due to the advantage the chase car has it should eventually catch up. The lead car's score is how long it evaded being caught or overtaken; the chase car's score is the opposite. Over multiple games, both players can then try to improve their score against each other. There is no need to set a baseline, as the players can do that for themselves.

A similar game style still related to racing would be having some players go the opposite way around the track, and deliberately attempt to collide with the oncoming cars, and score points appropriately. This should not disadvantage a player that spasms as much, as a spasm would make the oncoming car's trajectory more difficult for the opposing player to predict and avoid/hit.