Monday, November 9, 2009

Test Skills, Not Patience: Challenge, Punishment, and Learning

You and your friends are dead. Game Over.

Difficulty in games is a popular and thorny subject. Are games easier than they used to be? Does easier mean worse? Are games being "dumbed down"? And how do the dreaded "casual players" fit in?

The problem with these questions is that it is not productive to discuss difficulty as a single quantity. The term "difficulty" as it is commonly used encompasses two almost completely separate phenomena, with profoundly different effects on the player:

Challenge: Tests of skill or knowledge that require mastery of the game's mechanics to complete successfully. A strictly-timed series of jumps, a grueling boss fight, a pernicious puzzle.

Punishment: The consequences of failing a challenge. Being returned to the start of a sequence, losing money or other resources, or even being kicked out to the title screen.

Without challenge, there can be no punishment, because it is challenge that presents the opportunity to succeed and be rewarded, or to fail and be punished. Still, these are two mostly independent axes. A game can have high or low challenge levels, at the same time that it can be punishing or forgiving. Consider two games from 2008: Prince of Persia and Braid. Both games are extremely forgiving platformers - it is literally impossible to die, thanks to the Prince's companion and Braid's time-manipulation mechanics. Yet Braid is considered much harder than Prince of Persia - its challenge level is significantly higher.

Modern videogames range widely on the challenge spectrum, and many provide adjustable challenge levels. Punishment, on the other hand, has steadily trended downward. Games have become increasingly forgiving over the years.

Most recent adventure games don't let the player put the game in an unwinnable state, which would require reloading or restarting. First- and third-person shooters are increasingly allowing full health regeneration after a few seconds of safe rest. Platformers and side-scrollers dropped limited lives, and are now moving toward the removal of death entirely (some discussion on this evolution-in-progress here). Every genre has its own punishment conventions, and thus its own way to become more forgiving.

Why do we see such clear-cut patterns? Why do challenge levels spread out, while punishment levels overwhelmingly decrease?

It's because both challenge and punishment have a powerful relationship to one of the most important elements of videogames: learning.

Most forms of media have exactly one method of interaction. Once a person learns to read one book, they are equipped with the skills necessary to read all books, regardless of the genre, publisher, or binding. From then on, any learning that occurs is declarative, rather than procedural - the reader may encounter new and unknown words and look them up, but they are unlikely to encounter new types of pages that must be turned different ways.

Videogames, by contrast, never stop teaching interaction methods. This is true on every level, from the platform (learning to use a keyboard and mouse is of only limited help in understanding how to use a Wii remote) to the genre (learning to play a kart racer will not teach someone to play a city-building sim) to the franchise (the skills required for Mario Tennis will transfer only partly to Top Spin) to the installment (Final Fantasy X plays rather differently from Final Fantasy X-2) all the way down to the specific challenges within an individual game (the stomping strategy that worked so well on the Goombas won't be helpful against Lakitu's Spinies). Players must constantly acquire new skills to progress.

It sure sounds like a lot of work, doesn't it? So why do we keep doing it?

Because learning is fun. Humans are curious creatures, eager to gobble up new knowledge and abilities. Few things are more innately satisfying than being able to do something we weren't able to do before - and then, being able to do it better. This desire to master has always been fundamental to the appeal of gaming.

"...players derive much of the fun from the gradual introduction and application of game mechanics...

...the joy of discovery and the satisfaction of finally 'getting it' are hard to resist."
Rob Zacny, The Incredible Disappearing Teacher

A challenge is nothing less than an opportunity to learn. But this is only true if the challenge level is appropriate to the player's skill level. Too low a challenge presents nothing to learn, and will likely bore the player. Too high a challenge level presents something difficult or impossible to learn, and will likely frustrate the player. Neither will engage or satisfy the player. The optimal challenge level is just above the player's skill level - obstacles the player can just manage. My friends and I have observed this anecdotally, as it quickly became clear to us that the fastest way to improve at DDR and Guitar Hero was to play songs that we could just barely pass.

Think of the challenge level as the shape of a climb. Too easy is a flat stroll, presenting nothing of interest and failing to elevate the climber. Too hard is a sheer wall, impossible to scale and thus also failing to elevate the climber. Juuuuust right is stairs.

The player can see the path up, and knows that it can be traversed. It requires effort, but it's attainable. And the successful climb is a satisfying process of learning and improvement. (For more on this subject, read up on Csíkszentmihályi's concept of flow.)

Every player has their own set of experiences and their own skill levels. As gaming becomes more mainstream and reaches a wider audience, we should expect to see a greater range of challenge levels, allowing more and more players to enjoy games by having pleasurable learning experiences.

That explains challenge's tendency to spread out. What about punishment's decline? Again, it's the connection to learning - and punishment, it turns out, actively inhibits learning.

Learning a skill consists of a few stages. You make an attempt, observe feedback indicating how well you succeeded or failed, adjust your strategy based on this feedback, and make another attempt. Rinse and repeat.

In Shamus Young's absolutely wonderful video essay, Reset Button: Most Innovative Videogame of 2008, he points out the absurdity of the common videogame punishment scheme by applying it to another skill-based activity: basketball. Learning to shoot hoops would be a lot less fun if every time you missed the basket, you were sent home and had to walk back to the court. It would be annoying and time-consuming. But it's actually even worse than Young indicates. Not only does punishment break up and drag out the learning, it erodes it.

If you throw the ball at the basket and the shot bounces off the right side of the backboard, you aim a bit further left on your next attempt. But if you have to walk back to the court after the first airball, by the time you get there your muscles won't remember exactly where you were positioned before, and you won't know how to compensate. The value of the feedback decreases tremendously when you have to wait to act on it.

Now imagine that on your way back to the court, you have to use a completely different skill set - climbing over obstacles and dodging careless skateboarders. How can you possibly remember how you set up your last attempt? The feedback has been almost completely obliterated. In seconds, you miss the basket again, and are consigned to your learning-disruptive walk once more.

Suppose that this had been the only way to learn basketball for decades, and then someone created a basketball court that did not eject players when they missed a shot. Would the "hardcore" basketball players bemoan the "dumbing down" of basketball? Would they spit vitriol at the "casual players" this move is ostensibly designed to attract?

After all, without punishing failure, how does one sufficiently incentivize success? Punishment as a consequence for failure creates a fear of that failure, lending urgency and emotional weight to the challenge. But is that really true? Does punishment motivate, or frustrate? Does punishment create fear or annoyance?

"Creating real fear requires immersion, and sending the player back to the loading screen kills that. A second ago they were afraid for their lives. Now they remember they're in their living room, it's all just a game, and the danger was never real to begin with. You can threaten them all you like but once you actually kill the character, the player will remember you're all bark and no bite because you can't really hurt them. The worst you can do is stop them from progressing in the game, which just isn't all that terrifying."
Shamus Young (again), You Don't Scare Me

Punishment isn't scary, but it's certainly unpleasant. It's annoying and often repetitive, and it's also insulting. It's a face-slap, a knuckle-rap, a disapproving cluck. It's penance the game imposes on the player for performing disappointingly. It motivates the player, all right - not to succeed, but to avoid punishment - which means avoiding risk of failure.

A player who dreads seeing the GAME OVER screen and being sent back to the title is much more likely to give up and visit GameFAQs than stick with a challenge until they have mastered it on their own. This player will learn less, and in a much less satisfying way. Punishment is downright damaging to the learning process. The opportunity to fail freely is vital.

"People remember things better, longer, if they are given very challenging tests on the material, tests at which they are bound to fail. In a series of experiments, [researchers] showed that if students make an unsuccessful attempt to retrieve information before receiving an answer, they remember the information better than in a control condition in which they simply study the information. Trying and failing to retrieve the answer is actually helpful to learning."
Henry L. Roediger and Bridgid Finn, Getting It Wrong: Surprising Tips on How to Learn
(referencing research performed by Nate Kornell, Matthew Hays and Robert Bjork at U.C.L.A.)

Failure itself is sufficient to motivate learning. The simple absence of success makes us want to succeed. Punishment is dead weight, and removing it streamlines the learning process, while increasing fun and satisfaction.

Until the entire videogame community understands the difference between challenge and punishment, developers will continue making the same mistakes, and players will continue to berate the developers who avoid these mistakes. Consider the case of Bionic Commando Rearmed.

Bionic Commando Rearmed box art

Rearmed is an enhanced remake of the NES game Bionic Commando, released twenty years earlier in 1988. Lovingly faithful to the source material, but giving it a fresh paint job, new features, and upgraded boss fights, Rearmed was extremely well received, selling over 130,000 units during its first week. But game design has come a long way in twenty years, and some of the mechanics carried forward seem outdated now. In particular, Rearmed maintained its predecessor's use of limited lives, ejecting players from levels when they died too many times.

Nearly a year after the game's release, it was patched. The patch did a few things, but one of them was to remove the limited lives system. And what was the community response?

"Yesterday it appears that a patch was released for Bionic Commando: Rearmed over PSN. The update is a bit of a mixed bag, unless you're a total wuss and can't beat the game as it is. See it does allow for Trophy support, which is good. However, along with the Trophy support comes some changes to the games basic gameplay. Players now have unlimited lives (no starting levels over from the beginning), can swing into walls without falling to their untimely deaths and can reel out their line if it is too short. These features make the game immensely easier, betraying its old school roots.

Thankfully the update only tweaks the easy and normal levels, but that doesn't really excuse it. Who was out there complaining that BC: Rearmed was too hard, especially on easy and normal? Nancy-pants girlie men that's who. For now 360 owners can walk around claiming they have bigger nerd balls as the update does not appear to have landed on LIVE yet. PS3 owners, your cred is in question until the 360 wusses out too or Demon's Souls lands."
Matthew Razak, Bionic Commando patch is up, makes PS3 owners girlie men

This commentary is clearly not meant to be taken completely seriously, but the attitude behind it and the message it sends is clear - and thoroughly misguided. Rearmed's patch is, more than anything else, a reduction in punishment, not challenge. And the fact that Razak is thankful for - the changes only applying on Easy and Normal, and not Hard or Super Hard - is an awful, awful mistake.

By leaving punishment intact on the higher difficulty levels, the Rearmed patch reveals an assumption that follows naturally from the fallacious conflation of challenge and punishment - the player must want both or neither. There is no way to experience the game's full content and master all of its challenges, without also suffering through its heaviest punishments. The problem is that high challenge magnifies the downfalls of punishment by guaranteeing that the player will be subjected to it over and over again. The more there is to learn, the more destructive is the disruption of the learning process. Increasing both challenge and punishment at the same time is a surefire way to prevent learning by frustrating players and driving them away.

Rearmed is hardly a unique case. Throughout the gaming community, challenge and punishment are treated as equivalent components of "difficulty," and the results shortchange developers and players alike. Constructive debate is nearly impossible, as all attempts to mitigate punishment are seen as equivalent to attempts to remove challenge. Progress, therefore, is slower than it might be. But it does occur. And the market has a tendency to reward it.

Consider Braid, briefly mentioned above. A puzzle-based side-scroller developed by Jonathan Blow, Braid has some of the most forgiving game mechanics yet seen. Due to player-character Tim's limitless ability to rewind time, all mistakes can be undone. That includes death. There is no running out of lives; there is no Game Over. Yet the game presents some fiendishly challenging puzzles, requiring non-linear thinking and an understanding of the consequences of some decently-elaborate temporal manipulation.

Braid box art

In Braid, Blow quietly offered up high challenge coupled with an extremely forgiving environment in which to master it. And people loved it. Several reviewers called the game a "masterpiece." It holds a Metacritic score of 93. It sold about 55,000 copies in its first week and was "very profitable" for Blow.

And thusly the state of the art advances. The great debate rages on, going nowhere, but periodically a game comes along that lowers the punishment another level, and sells well enough to be imitated. Soon enough, the previous level of punishment feels old and outdated. So while it might not be easy to say out loud, every once in a while it's proven through action: punishment is not challenge, and its role in the modern videogame is - and ought to be - rapidly disappearing.


  1. Dr. Doctor Professor,

    I have an unexplained rash on my...oh wait...not *that* kind of doctor, huh?

    Anyway, I really like this entry, and I think we agree that punishment as a result of failing a challenge is, generally speaking, 'bad'. But what about 'punishment' as the result of a 'choice'?

    If the choices you make in the game are consequential and, most often, not fully understood - it seems like only a question of time before a player makes a choice he or she regrets. The biggest example I can think of is character creation. How do you avoid punishing players 'bad' choices and still have the choice be important/meaningful?

    I remember playing Titan's Quest and having to spend points to increase certain skills - but I had to do so without any knowlege of what the returns would be. I could decide, 'Oh yeah, I want to be a fire guy' but...if many levels later, I realize I don't like it...what happens then?

    It seems like the ability to change my past choices at will make the choice nonconsequential. But, not having the ability to change the choice makes it seem like a form of punishment.

    I suppose you could 'fully educate' the gamer on the consequences of a choice, but I don't see how you could do that without spoiling the rest of the game.

    "If you decide to give this rose to Xanga, and continue to give her things, at the end of the game, the cut scene will show you two married. But if you don't, in dungeon six, you'll be able to trade it for an extra heart container"

  2. What about games that have become wildly popular because of their inane difficulty? If you ask any 20-30 year old man about their thoughts on Contra, just about every one of them will say something positive. And yet, there's always some kind of awe for the guy who was able to beat the game without the 30 lives code. Not many people have done that, and yet, most people have positive things to say about that game. That's clearly a game that is well above most peoples skill level, and yet still enjoyable to a vast majority of people. In fact, I'd contend that's a game that gained popularity due to it's extreme difficulty.

  3. Rob: I'm afraid I can't help you with the rash... you might want to check with my cousin Professor Doctor for that. ;)

    You've hit on something major here. Of the metanarrative traits of choices discussed here a couple of weeks ago, consequence is the only one that often leads directly to punishment. Here's the thing, though - that only happens if the choice is tied to game mechanics (and thus not free, according to my definition - see the Indigo Prophecy example from that post).

    Wanting to make choices irrevocable in order to make them feel more real is reasonable for certain types of choices - but wildly unfair for others. Character creation and skill point assignment are almost universally an example of the latter - games like Deus Ex and Fallout ask the user to pick between a variety of weapon-based and non-combat skills (such as hacking) without having any idea whatsoever how common or useful the various weapon-types or hacking opportunities will be. Ostensibly it lets you play the game the way you want to play it, but there's no guarantee the game world will be set up in such a way that your chosen build will be remotely effective.

    Choices like this - and like the rose example you give - are thus basically really mean puzzles, with no hints (so as to avoid spoilers) and no real indication of consequences for several gameplay hours. The punishment is thus particularly obnoxious, since you now must live with the subpar consequences or replay several hours of material.

    I believe this approach comes from good intentions - the desire to provide interesting choices and surprises. But there are too many conflicting goals. They can't all be accomplished at once, and the result is self-defeating.

    Choices with lasting consequences related to game mechanics should be revocable - you should be able to refund your skill points in Titan Quest. How does it improve your experience to force you to either play a character build you don't enjoy or start completely over, to improve the supposed authenticity of a decision you made with little-to-no information?

    And on the other side, choices with irrevocable consequences should not be tied to game mechanics. If giving the rose to Xanga improves your relationship with her, then it should not be tradeable for things that make the game easier - though it would be fine for it to have other plot/character/cosmetic effects, such as letting you give it to some other person instead to fall in love with them, or wear it in your lapel to be the dashing lone wolf. The use of the rose should still be mostly explained, however - there's still no justification for locking the player into a choice they didn't even know they were making.

    In short, I believe that punishment as the result of a choice is just as bad as punishment as the result of failing a challenge - because attaching punishment to a choice turns the choice into a challenge, and usually a poorly-informed one at that.

  4. Anonymous Contra fan: I agree with you. Challenge is fun! That's definitely a large part of why Contra is such a classic. And yes, I am totally impressed by anyone who can beat it without the 30-lives cheat - but Konami was completely right to put the cheat in anyway. When used, it mitigates punishment and provides an environment where it's easier to learn and practice the skills necessary to beat the game - which can then be shown off by playing without the cheat.

    Still, as many people as there are who had a blast dying over and over again to learn to survive a bit longer every time (and there are certainly plenty of them - even failing is fun as long as you feel like you're learning) I'd wager there are plenty more who were quickly frustrated or scared off, and never learned to enjoy the game.

    By having such a high difficulty level, Contra had a high barrier to entry. The result is that anyone who actually played it long enough to form an opinion and lasting memories probably loves it, but also that the number of people who pulled that off is comparatively low. If you ask around about Contra, I bet you won't hear many complaints - instead, you'll hear a decent amount of "Yes! I loved that game!" but even more, "What? I don't think I played that." (Especially if you venture outside of the 20-30 year old male demographic.) It's the rough equivalent of a film cult classic - most people don't care, but the people who like it really like it.

    As the cost of videogame development has skyrocketed, this is no longer a safe strategy for the major publishers - so they instead develop games designed to appeal to as wide a swath of the market as possible. Smaller, indie developers are where you have to look now to find niche-marketed titles poised to become cult classics.

    I know it's weird to think of Contra as a cult classic, given what a major player it was back in the day, and what a huge proportion of its market liked it - but the videogame market is a lot larger than it used to be. You have to ask those 20-30 year old men for an opinion on Contra - but there are a lot more groups you could ask about, say, Animal Crossing or World of Warcraft.

  5. Great post - well written and you clearly did your homework. I came across your blog via a mutual friend at Cornell.

    However, I must respectfully disagree with your conclusion that challenge and punishment are independent. I think they are fundamentally linked: The punishment, or cost of failure, is a property of the challenge itself. There are certain challenges that you simply cannot have without certain punishments, and to have punishment "disappear" from gaming would fundamentally kill off a large portion - if not all - of challenges.

    Let's take your basketball example. Increasing the punishment for missing (having to run across the court again) _changes_ the challenge itself. The challenge is no longer to make the free throw. It is now to run across the court, and then make the free throw. You have to do both successfully in a row!

    Now, that's not a very interesting challenge, since running across the court is relatively trivial compared the free throw. But the point is, requiring succession (ie. you must do multiple things in succession without screwing up any of them) is a challenge in itself, and it cannot be done without punishing the player in that way (ie. sending the player all the way back to the beginning).

    I think what you're really getting at is that "challenges of succession" (maybe there's a better term for this) should be minimized and perhaps made optional for modern games. Braid is probably the most granular example of this, as it effectively makes the length of required successions infinitely small. But games have always done this. Any cheat code that gives extra lives (such as the Contra code) effectively reduced the succession requirement - you're given more chances to screw up "in the middle". Many games also go the other way with optional modes that increase succession requirements: Metal Gear Solid games always had the optional "no alert" modes, where being spotted was an instant game-over. The recent game "Torchlight" has the "hardcore" mode where death is permanent (ie. it deletes your save).

    So maybe what we're actually getting at is that the punishment's temporal scope should not exceed that of the new, interesting skill that is currently being learned. Shooting free throws always has something punishment. If you miss, you gotta shoot it again - that takes more time and more effort, and is thus a cost of failure. But this is somehow a more "fair" punishment, because it is limited to the scope of shooting a free throw. But even then, people still get frustrated and give up on basketball, so it's probably not as simple as this either.

    So's complicated :P There are various types of challenges and they appeal to various audiences. Some audiences are larger than others. I think the audience that enjoys challenges of succession is smaller than those who enjoy challenges of the Bejeweled kind, but punishment is still a part of both. I don't think it's as simple as "eliminate punishment and they will come."

    Anyway, thanks for an intelligently written blog. These issues are complicated in a rapidly growing but young industry, so we're all going to disagree, but that's half the fun ain't it? :)

  6. Just one more thing: I think the trend that you see in games isn't a decrease in punishment, but rather a shift in the types of challenges that games utilize. Challenges of succession are decreasing with the advent of regenerating health, time reversal, lives-less games, etc., while challenges of logic and motor skills are increasing.

    But that doesn't mean challenges of succession will ever be gone completely either. And they shouldn't. I think many games are made fundamentally more exciting thanks to succession/consistency/endurance/whatever you wanna call it. Racing games, for example. If a racing game gave you the option to restart at the previous lap instead of the beginning, then getting first place wouldn't be nearly as exciting or gratifying. But who knows - maybe there's a large audience out there who would still find such a racing game fun. But there is definitely an audience already who would probably find such a game to be boring and "dumbed down." A matter of taste.

  7. Steven A.:
    Thanks for your comments, Steven. You've given me a lot to think about.

    Challenge and punishment are definitely linked. Punishment is what happens when you fail a challenge that is different from what happens when you pass it - so without the challenge, there can't be punishment, and without punishment, there can't be challenge to speak of because you would pass by default. Punishment can't and shouldn't be eliminated, but I believe that it should be minimized.

    I think most of where we disagree comes down to a differing set of definitions. You speak of "challenges of succession" as a type of challenge, while I say that linking challenges is a type of punishment. But this is really just a difference in the names we use. Multiple joined challenges, or a single multi-stage challenge - it's just a question of which way of looking at it is more useful. To me, it seems more useful to define the challenge as scoped to the specific skill involved, so I think of it as multiple joined challenges. Usually. :)

    What I find really interesting is your idea of limiting the punishment's temporal scope to that of the new, interesting skill being learned. That sounds like a great rule of thumb to me. And leads to similar results whichever way you choose to define the granularity of the challenges themselves.

    My issue with your racing game example is that in most racing games, the game's defined goals are on the level of the race, not the lap. It's easier to take first place in the race if you also take it for each lap (probably) but it is by no means required. If you had to get first place in each lap, and failing to do so for a given lap put you back at the beginning of the race - that'd be a smaller-scale version of putting you back at the start of a tournament for failing to take first in a race, instead of giving you a chance to try the race again. I think what makes this scheme frustrating is that it rewinds you past previous failure points. It undoes your successes, and makes you work to even get back to what you were trying to do - and creates the possibility that you won't be able to do so.

    If the player wants to go for the arguably-greater accomplishment of stringing their challenges together and passing them all in a row, they can certainly do so. But having the game enforce it on everybody all the time just seems like a really bad idea to me.

  8. Thanks for the reply. It's good to know someone else on the internet able to discuss these things intelligently and concretely.

    Assuming that punishment is temporally responsible (ie. the game doesn't just make you sit there for some amount of time, which actually a lot of games do with loading screens), I think the granularity/definition of a challenge is a very significant to its perceived difficulty. It is precisely how the game chooses to define its goals - and thus its "minimal punishments" - that determines whether or not it is a "difficult" or "easy" game (well, logical difficulty is another factor that I think is orthogonal to this..."structural difficulty").

    Then to minimize punishment, one may conclude that all games should make their challenges as finely grained as possible. So basically, if every game had Braid-style rewinding (not much of a stretch with technology these days), all our problems with games and audience accessibility would be solved, right?

    I think not. To a very real extent, I think players need to be told how to enjoy a game. If rewind was available for Super Mario Bros., players would use that, and thus make the gmae too easy for themselves and thus boring (according to Flow theory). SMB was designed to be played in a certain way for a certain audience, and I think it can only be made more accessible by significant design-tuning - not any rules of thumb.

    I think ultimately, you need to choose some groups of audiences to appeal to (depending on how much resources you have), and just design your game to cater to their respective skill levels. This is best done through focus testing and iteration. Some audiences are easier to appease than others - the super-hardcore are probably the easiest to appease if you just give them an "INSTANT DEATH" mode/achievement or something like that.

    So I guess what I'm saying is that, yes, games are dumbing down! I don't think it's so bad - it's good for the industry to expand its audience, and for my personal skill level, I find plenty of games I still enjoy. But you can't make everyone happy.

    As for the Bionic Commando comment, I do think that's a little ridiculous. The game was designed well for a certain audience, and that audience still has the option to play it like it was before.

    Uh...hope that was coherent :P I need to get back to work.

  9. Steven A.:
    Punishment scope does affect how difficult or easy the game is perceived to be, but it's not the only factor that does so, or even the most important one. As we seem to agree, the punishment in Braid is (in most cases) effectively scoped infinitely small, yet the game is certainly not perceived as anywhere near infinitely easy. But (to channel LeVar Burton) don't take my word for it:

    "[Blow] says that Microsoft asked him to add clues after user testing suggested that players sometimes needed help. . . . Because Braid can be difficult, some players may not be able to finish and address their questions about the plot. Mr. Blow says that's intentional. 'Let me provide a longer-term challenge,' he says. 'That's why there aren't any hints.'"
    -Jamin Brophy-Warren, Time Out of Mind

    The Wall Street Journal isn't the only entity to call Braid difficult. The gamers also find it hard.

    Why? Challenge. Skilled play is needed to solve many of the game's puzzles (even after you twist your brain enough to figure out what to do).

    Super Mario Brothers was designed to be played without rewind, but I'm not convinced adding rewind would ruin the experience. It would make it different, certainly, but I don't buy that it would suddenly become boringly easy. The game is hard for more reasons than the fact that you die and run out of lives. Again, it's the challenge: the skill required to pass the game's obstacles.

    Having a bunch of tries to make a jump doesn't make the jump any inherently easier, any more than having a bunch of tries to make a basket makes the basket any easier. It just means you get better at it faster.

    Reducing punishment scope changes goals, not challenge. Suppose a level in a sidescroller has a sequence of three bottomless pits, and falling into any of them will send the player back to the beginning of the level. The player thusly has the goal "Jump across pit A, then pit B, then pit C." Give the player rewind powers, and now they will have three goals: "Jump across pit A," "Jump across pit B," and "Jump across pit C." Each jump, however, is discrete and requires just as much skill either way. The magnitude of the challenge does not change. The rewind-less game will be perceived to be more difficult, but that's exactly the problem with punishment - it inflates difficulty in unchallenging and thus uninteresting ways.

    And no, you never can make everyone happy. But the great advantage of a dynamic medium like videogames is that you can make a lot of different groups of people happy, thanks to difficulty settings and other options (with good user testing and iteration as you pointed out). (Failing to do this may well be what killed pinball.)

    You mention that people still have the option to play Bionic Commando Rearmed the way it used to be - "option" is the key word here. For nostalgia releases like Rearmed and Mega Man 9, I think there should be options to play the old school way or remove the punishment - completely separate from the difficulty settings, which affect the level of challenge.

  10. There are a few semantic arguments we can make here, but I think we fundamentally agree: To maximize your game's potential audience, you should find ways to offer various levels of difficulty that do not compromise the game.

    I guess I'm just making the cautionary point that this must be done with thought and care, since offering the option to reduce difficulty not only affects the game experience, it also affects how you design the game itself. Braid allowed infinite rewind, because Blow wanted to focus on the logical challenge of the puzzles. What about SMB? You say that having rewind in SMB would make it different - agreed. And by making it different, you take away the original game - I think that would be a shame. Some would argue, "but you can just not use the rewind!" This is true if rewind was "patched in" to SMB. But what if SMB had been designed knowing that rewind would be in to begin with? The platforming would probably because more ridiculous (consider Braid's last level). That would necessarily change the experience for those that want to play it without rewind, possibly making it damn near impossible.

    Of course, it could be done in a way that offers the best of both worlds. But it needs to be done with care. You can't just throw Braid-rewind into every game and assume it's a net win.

    So minimizing punishment or difficulty is a good idea, but there's a right way and a wrong way to do it.

  11. Yes, the praxis nature of the 'braid-rewind' is a strong skill-building process. That doesn't automatically mean all games should be completely focused, or limited, to skill-building exercises. Like the Contra example, punishment avoidance and 'succession challenges' are good design elements that shouldn't be tossed out unnecessarily. To use the basketball metaphor, you can't rely on 'rewinds' while playing a competitive game against another team, and so you must also train in a setting where losing (punishment) is a possibility.

    I can see a happy medium though. Let's stick with the Contra example. The designers could have included a 'practice' option at the start screen, where players can go and play short, tight, increasingly challenging game scenarios. This would be separate from the main game itself and have as its purpose to strictly enhance your skills without slogging through the levels as a noob.

  12. Dear Doctor Professor,

    I was wondering if you'd be interested in writing articles for If so, shoot me an email at (I don't want to put my proper email up in a comments section) and we can discuss ideas, payment, etc.



  13. About punishment as far as choices. What in the case of the player having an indication that the choice may have unfavorable results. For example in a non-linear game the player having the option to go into a very dangerous area early on but being warned how dangerous it is and that they aren't ready to brave it.

  14. @Anthony:
    Fair warning is vital. Choices are really only choices to the extent that they are informed - otherwise the player is effectively just getting arbitrarily smacked around by the game designer.

    For example, I was recently playing Final Fantasy V Advance. That game has a number of poor design decisions, but the last straw for me came when I was exploring the surrounding area, found a cave with level-appropriate encounters, and then suddenly in the back of the cave I got into a boss fight way above my level from which I could not run. Game over; load from save.

    Had I been warned that there was an extremely dangerous boss at the back of the cave, I probably would have at least saved before going in, and the experience wouldn't have caused me to finally put the game down for good.

    That said, increasing information doesn't necessarily reduce punishment - it just makes it less arbitrary. I don't see why the game couldn't have allowed me to flee from the battle (and just put me back a square from where the fight is triggered, like it does with trapped treasure chests). Having death lead immediately to loading a save is also a pretty punishing model, at least when saving opportunities are limited as they are in a Final Fantasy game.

    Ultimately, warning or the lack thereof just lies on top of whatever punishment model the game already uses.

  15. I thought this article was very thought-provoking and I kept starting to write a bunch of comments since you first posted this, however in the end I wrote my own article about this, using the great game Splosion Man as a lense, & referencing your article here. :)

    Summary - I think that the elements of challenge and punishment are very complicated to separate in the way you're describing, and so I strongly agree with the points raised by Steven A above.

  16. My current game I inspired on old school arcades and all...

    Yet, LOTS of players complained that it was too hard... But one player, figured the problem: He said that it was boring, because you had to play the easy part over and over again, as every time you got a "game over" you had to start again.

    This made me realize that the problem is not really with punishment itself, but with certain kinds of punishment, first, time punishments are annoying (like loading screens...), as is some other stupid punishments... Then we have the punishment that is most common, yet most problematic: Forcing the player to do something that he ALREADY KNOW how to do.

    In that situation, he is not learning anything new, this is only irritating him...

    Contra can get away with its insane difficulty and punishment, because it is insane ALL THE TIME, if you get to the beggining, getting up to where you was, is a challenge itself, a hard one.

    Now Braid for example, the game is about puzzles, what would happen if you needed to restart the level? Well, you ALREADY KNOW the complete solution, thus doing it again would be dead boring.

    How I fixed my game? I added auto-save and a game over menu, you are given three options: losing bonuses to remain playing (really forgiving, as you can easily get them back), going to the high-scores screen and putting your score (ending your game), and note that these two options ALREADY existed (and the game was considered too hard), the third option is load the last save, thus, reloading the level.

    This introduced a issue on its own: the punishment became purely about time... With sufficient persistance you would still reach the end of the game, but I don't wnat players getting bored by finishing the game brute forcing their way, in fact, the point is not even finish the game, the point of my game is get high-scores! So the option of reloading the level takes away a bit of your score. This mean that those that only want to ignore scoring and reach the end, mostly to learn how to play, can do it freely, without screwing-up the challenge of getting a high-score for those aiming for a high-score.

    Maybe it is not a perfect solution, but noone ever said to me that my game was boring. (they still complain that it is too hard... but that is balancing issues :P)

  17. @doctor
    the informed choice is more towards your rose scenario. For example if instead of a rose it were some rare flower that was known to have potent alchemical properties. It would seem less arbitrary when later it could be used by a character you meet for the MacGuffin of alchemical awesomeness if you didn't give it away to your leading lady.

  18. I apologize for this comment coming so long after the last, but I've been reading and thinking about this article and the subject it's about for a while now, and have something to say.

    I agree that punishing a player by forcing them to start over from a long ago point after a set number of 'tries' is usually detrimental to the learning process.

    Key word is 'usually', however. I have to agree with Krio Dragon. There are exceptions to this, such as Contra.

    The problem in most games is having to redo challenges that are no longer challenging or entertaining to get back to the current 'challenge'.

    This is not a problem for games like Contra, as it is challenging every time. Even when you push through the challenges for the 30th time, they're still amusing due to their challenge, and how you can (to some degree) use different strategies to conquer them.

    But that is neither here nor there. What I wanted to say is "What if, instead of punishing failure, we reward climatic successes?" Not just completing the challenge, but acing it? The difference between a C and an A?

    For an example of this, there is Street Fighter 2: Turbo. You can continue as many times as you want, and complete the game that way. However, if you complete the game without continuing or even losing a round, you get to fight Akuma instead of the normal final boss. Akuma is a reward in and of himself, as he presents a new challenge in the form of a satisfying and challenging battle.

    I'm not a fan of the SF series myself, but I can guess that simply kicking you back to the title screen and forcing you to reply each fight until you can reach M. Bison would not have be nearly as loved as the 'reward' of Akuma, nor would requiring you defeat Akuma at the end of any playthrough be 'rewarding'.

    So what if, at the end of any given challenge, you get extras or unlock new levels/bosses when you ace a level instead of just pass? This allows casual players to cruise through the game, while providing a tentative challenge to cut your teeth on if you enjoy the game enough or consider yourself hardcore.