Gamasutra: The Art & Business of Making Gamesspacer
View All     RSS
October 25, 2014
arrowPress Releases
October 25, 2014
PR Newswire
View All
View All     Submit Event

If you enjoy reading this site, you might also want to check out these UBM Tech sites:

Emotions and Randomness - Concrete Examples from Ni No Kuni and Others
by Chris Grey on 04/09/13 01:20:00 pm   Expert Blogs   Featured Blogs

The following blog post, unless otherwise noted, was written by a member of Gamasutra’s community.
The thoughts and opinions expressed are those of the writer and not Gamasutra or its parent company.


Since the last essay was mostly abstract, in this one, I wanted to give a few examples of games that have implemented randomness in a way that significantly impacted the game, for well or ill.


"Gotta Catch 'em... Nevermind" - Ni No Kuni

 Adorable, but hard to catch.

I heartily recommend this game to my friends, albeit with one severe caveat: don't try to recruit all of the monsters. There are approximately two to three hundred to catch, depending on the amount of leveling the player wants to do. The recruitment rates are constant (with a one time ability that boosts all rates a fixed percentage), so we are firmly in the constant drop percentage scenario I wrote about in the previous essay.

The problems I mentioned about constant drops are compounded in this case because there are so many different monsters that, even if the drop rates look high, the player is literally more likely to win the Powerball jackpot than to not get severely unlucky attempting to recruit at least a couple of creatures. This issue is again compounded by the fact that the player will be getting several taming events from some of the other creatures while there will generally be one creature in the region that will elude capture. The game is giving mixed signals; taming is presented as easy to get because taming events happen very often while there are several specific creatures that will elude you. This sours the moment-to-moment play experience because the player will be focused on how unlucky they feel, due to the fact that the designers have basically ensured that approximately every couple of regions, some creature will take much longer to catch, even though the tame rates probably all match.

In one single decision about randomness, the designers here go a long way toward spoiling all the excitement and wonder everything else in this carefully crafted, beautiful game is working to build.

In case you were wondering why I chose to talk about Ni No Kuni instead of Pokemon, the latter softens this problem somewhat by having items like Master Balls, which are items that allow guaranteed capture of any creature they are used on. In the long run, the deeper the player gets into Pokemon, the more they will experience something similar to Ni No Kuni, especially if the player wants to tame Pokemon with specific natures or only train their Pokemon against specific creatures.


Low Constant Drops Done Well - Castlevania: Symphony of the Night

 It's a fountain.

In Castlevania: Symphony of the Night, almost every enemy has a couple of items that they can potentially drop when defeated. Noteworthy here are about twenty possible high level weapons and other powerful items that are only obtainable from enemy drop. The weapons available through the store or found in the course of the game are powerful enough that there's no need to have the drop-only weapons, and there is no achievement for doing so either, as this was a PlayStation 1 game. The dropped weapons are generally more unusual than excessively powerful and tend to force the player to play in a different way and look at the environment differently. The key difference here is that as all meaningful drops are quite rare late in the game, the player is taught by the randomizer not to expect anything, unlike Ni No Kuni, where the player is getting constant reinforcement that taming should be easy. Therefore, any rare drop gotten is more likely to be seen as a fortuitous gift, rather than an eventuality.

The rates are such that there is a good chance that the player is likely to get an unusual weapon at some point in the game if they tend to explore, something the game wants you to do anyway. The weapon is powerful enough to entice the player to spend some time learning to use it, and the player can never get a weapon that is totally incompatible with the character's skills. Through using this weapon, the way the player must approach the game changes dramatically. Therefore, the randomness adds a tremendous amount of replay value without requiring new maps. For a genre that has traditionally relied on hand crafted maps, this is a fantastic solution to replayability.


A Kinder Looting System for Quests - World of Warcraft


The pacing of World of Warcraft is decided by progressions of quests, and many of these quests take the form of killing some number of the same creature to gather some number of quest items they alone drop. Before the Wrath of the Lich King expansion, the drop rates for these quest items tended to be high, but like Ni No Kuni, when the designer demands the player succeed at repeated random actions, the player will go on lucky and unlucky streaks. Blizzard found out that players remember their unlucky streaks more vividly; unlucky streaks, after all, bloat and pad the time required of what is a task that was not designed to be interesting for that long. Blizzard's solution was to implement escalating drops for quest items in order to balance the pacing of these quests. This also has the effect of capping the maximum unlucky streak the player could suffer, leading to a more even game. The interesting side effect that the change had was to additionally decrease lucky streaks, as well, and Blizzard revisited and increased the initial drop rates to allow players to have more consecutive drops, allowing players to feel luckier.

Note that this system only applies for quest items, which are not the powerful weapons, armor, and items that WoW players seek in the long term so this change doesn't really effect the economics of the World of Warcraft, just the pacing. It does seem rather cruel to have the randomizer teach that unlucky streaks will be capped for common quest items, but the player may have attempt for years to get the ones they really want. Part of Blizzard's motivation on not escalating the legendary drops stems in part from wanting repeatable end-game content while maintaining rarity of legendary items on servers so the comparison to single player games gets a little harder from all of the other interpersonal and economic forces involved. I'll write more about a similar situation in a game that's more single player focused in the next section.


A Couple of Shots and You're Out - Demon's Souls

 Demon's Souls

Demon's Souls allows for an interesting experiment. The game features two kinds of randomness that we've talked about: two of the items that yield the best weapons are only found from an enemy with limited respawns. It is a ten percent probability drop, which can be raised to about twenty-five percent if you understand the mechanics, and there are at most six chances you have per game to get it. The game has a New Game+ feature, which allows the same character to play through the game again while retaining items, so the drop will be available next game cycle if you didn't get it in your first. This dramatically limits your time to attempt to farm the item, and the monsters that carry the item are weak, but quick. Once the player understands the stakes, which generally takes overlooking some of these creatures or letting them get away, they tend to focus energy. The stakes are raised, and it's an experience, for well or for ill, that stays with the player. These drops have generated resentment, but in general, the pain is over quickly, and the players accept it as part of the experience. It also helps that the unique multiplayer aspect of the game has allowed people to learn to trade items with each other.

What has generated more resentment than anything else in the game is another, less powerful item, which is an extremely rare -- 1/200 under optimal conditions -- drop from an infinitely respawning monster. To give a point of reference, there is nothing else in the game that drops so small of a percentage of the time on any monster. The time to kill one of the monsters is about ninety seconds once you hone your path, so on average, the player is looking at doing exclusively that for about five hours on a game that takes about twenty hours to finish a cycle. What's worse is that there's a twenty percent chance you won't get one after doing a ninety-second loop for seven and a half hours. If you search the message boards, this item has about three times the number of rage threads that the previous items do. The experience is so uneven that for each player who gave up farming after ten hours, there's one who got it in five minutes. The variance in gameplay is huge, especially when you can see that the ten hours became rote mechanical behavior. To add insult to injury, this process, if it goes very long, will cause the player to accumulate a tremendous number of levels, almost trivializing the rest of a game whose signature calling card is its difficulty. As the downside experience is so great, any bonding done over this item tends to be resentment, instead of excitement. This dark cycle is all reinforced by achievements that require obtaining the item.


The World's Just Made That Way - The Binding of Isaac

 Binding of Isaac

This game gives an extreme example of limited availability drops. All game sessions are randomly generated and last, at most, about an hour; this sets the stage for the world to only contain a small number out of a large pool of potential items. As the items vary greatly in power, the player will have wild variability between play experiences, with a bias toward feeling underpowered. The great care taken here is that the theme of the game melds very well with the experience the randomizer gives; life is not fair, the player doesn't know what will happen in life, and the player must find a way to work with the hand dealt to survive.

The saving grace to having a terrible allotment of items is that each game is short; there isn't much sunk cost in any one run. Finding yourself in a nigh-unwinnable situation thirty minutes in creates pressure to perform, and if you fail, well, you got unlucky. If you succeed, the victory is generally a fantastic one; you really did overcome the odds. If thirty hours into a game, you didn't have a viable path to winning because the good items weren't available in this game world, the experience would undoubtedly be a bitter one instead.


To Sum Up

The key point I'd like to make with these examples is that understanding the longer term consequences of the randomness used in your game is extremely beneficial to adjusting the moment-to-moment play experience. There is no kind of randomness that is universally wrong to use for drops, but the context of the design vision is king here and should be used to guide your selection. When there is a disconnect between what you are teaching the player and what the randomizer teaches the player, there will be unnecessary dissonance between the game the player was attracted to and the game the player is playing. Ni No Kuni's moment-to-moment play experience felt like a chore, which goes against the magical feeling presented by the story and art. Symphony of the Night's randomizer reinforced the idea that the Dracula's Castle was ever-shifting because you generally didn't see the same rare drops twice in two game cycles. Just as people rightfully deride ludonarriative dissonance, a randomizer can also subtly undermine what the story and background of a game are attempting to create.

As I wrote in the first article, people are not taught to understand randomness particularly well so experimentation and simulation of the designed randomness will go a long way in ensuring that your design vision is realized as a cohesive whole.

Related Jobs

Digital Extremes
Digital Extremes — LONDON, Ontario, Canada

University of Central Florida, School of Visual Arts and Design
University of Central Florida, School of Visual Arts and Design — Orlando, Florida, United States

Assistant Professor in Digital Media (Game Design)
The College of New Jersey
The College of New Jersey — Ewing, New Jersey, United States

Assistant Professor - Interactive Multi Media - Tenure Track
Bohemia Interactive Simulations
Bohemia Interactive Simulations — Prague, Czech Republic

Game Designer


Ian Snyder
profile image
I really enjoyed the article -- thinking about randomness can get complicated, so it's nice to see some concrete examples of how it can affect gameplay :)

Robert Marney
profile image
I love having several examples next to each other of different approaches to the same problem of doling out end-game content through infrequent random drops.

It's interesting to note that Dark Souls (the successor to Demon's Souls) moves towards a more Castlevania mechanic, where the tiny drop chances are tied to unique and bizarre weapons, while the rare crafting materials you'll need to finish the game are mostly earned through side quest rewards, encouraging the achievement-hunting player to repeated use of the New Game + option.

Ron Dippold
profile image
I came to the same conclusion with Ni no Kuni. Gave up on recruiting all the monsters very early on, and gave up on doing all the alchemy quests in the endgame when it became obvious just how ridiculous it was going to be to get all those rare drops (even with Swaine's steal shot).

The real kicker is that you're much better off just finding a few good ones and raising the heck out of their level rather than trying to level up a whole bunch. I saw a Puss in Boots, said, 'Dang, that looks useful - especially since his rapid hits are kicking the crap out of my spellcasters,' recruited one, and he remained one of my best melee guys through the rest of the game, including the bonus world end boss. Repeat occasionally. Since the game forces you to do some recruiting and leveling of other pets for other quests you do get a taste, but it also reinforced my suspicion that most of them weren't worth it.

I just couldn't see any benefit to fighting those low odds, especially for no benefit except an achievement.

Chris Grey
profile image
I agree with you on the level a few monsters up approach. It would have encouraged much more experimentation if they'd rubbed-banded experience for low level monsters gain more. It was too bad; there were several interesting-looking monsters I wanted to try out, but I couldn't be bothered to sidetrack solely to level them up enough to see if they were useful.

I got lucky in that my rag-tag group of creatures managed to have two Upsy-Daisies between them. It was apparently a rare spell, and it allowed me to fight the post game battles without necessarily having great end-game scaled creatures or xp grinding.

I stuck with the Puss all the way through the game. So adorable, and the cuteness and attack speed are what attracted me, too. It took me getting a tank with a taunt trick to realize how the game wanted you to play the combat, though. For all of the tutorials about trivial things, I was surprised that the game never bothered to talk about the tank/glass cannon/healer trinity they balanced the combat around.

Jonathan Jou
profile image
I want to point out that, strangely enough, the DS version of Ni no Kuni does not have a random recuitment process. Instead, one member of the party has the ability to recruit familiars, and there's a specific but deterministic ritual to winning the enemy over.

I have no idea why they didn't add even a little determinism in the PS3 version, and it saddens me, given how incredible and delightful the hardcopy book's art and descriptions are, that people are inclined to not go around catching them all. I sure did.

Chris Grey
profile image
Thanks for the insight, Jonathan. I may have to track down the DS version; it sounds like it'd be worth checking out.

As for taming all of the monsters; I certainly started the game wanting to. I killed about 80 of the banana bunch family monsters and never recruited one. That experience went a long way toward overriding the desire to tame everything. That, compounded with the fact you had to grind to even see what a creature was capable of, meant that when you finally did catch it, you couldn't play around with it easily so there wasn't an immediate payoff in gameplay for finally catching something. Creatures just became something to ship to the ranch and never look at again.

Ramin Shokrizade
profile image
A big part of what we do is supply dopamine to our customers. The three primary factors to dopamine release are risk, reward, and novelty. The first two factors work best when they are in relative parity. Higher risk leads to higher reward. The higher the novelty of any action, especially if it leads to a reward, the greater the dopamine release. This is a delicate balance. Every time you do something more than once the potential for dopamine goes down. This is why grinding can be not only undesirable, but downright depressing for players as they experience dopamine withdrawl. Perhaps this lens can help explain what is happening in the scenarios described in the OP. Maintaining novelty often requires some for of user generated content, usually in the form of PVP action but there are other ways...

Stephen Karpinskyj
profile image
I really like this comment. Do you have the sources for this handy, or perhaps an elaboration of these thoughts somewhere on the internet?

Robert Crouch
profile image
I enjoyed the article. I have thought about this a lot, and I think it has to do a lot with how humans, players and designers alike, are generally pretty bad at intuiting geometric and exponential progressions.

From the designer who creates a random opportunity that, if a player wants to get it, might take the player thousands of hours to nearly guarantee, to the player who feels that the 1% drop rate means that after 100 attempts he should generally be successful. However, for 1 in 20 players they might go for 300 attempts before seeing the result they are looking for, and 1 in 100 go for 450.

As designers, I think it's important to think about what the end result should be. Do you want something that any given player might not see, ever, no matter how bad they want it? How does that fit into the game? How will your players feel about that? Sometimes that's OK. If you have 4000 of 5000 things that a player can never experience they are probably willing to tolerate it, the world space might feel vast. Consider Minecraft or something, even the most dedicated explorer simply will not explore the world, but that's OK. If you have 999 things that a player can get, and 1 that they can't, they will probably be frustrated, moreso if some players randomly stumble across that one but they can't.

So first determine how never getting it will affect differently motivated players. If you determine that players will be disappointed if they don't get it (perhaps it's an achievement, or an iconic reward) then determine an upper bound on how long a player would have to invest in order to guarantee its appearance. Would 8 hours be too much? 4? More? Less? It depends on the pacing of the game generally. An MMO could get away with forcing a player to farm for 16 hours over the course of time for something they wanted. A single player RPG would probably find that was too much.

I recently played through Final Fantasy XIII-2, and there is a slot machine whose jackpot odds are something like 1 in 64,000. There's a reward, 1 of about 140 "shards" that you are encouraged to collect in the post-endgame, that relies on you winning a certain amount of coin from the slot machines. You can reach this number by winning a series of the second level jackpots who have a 1 in 5000 chance. In practice the common means of winning this shard is to buy a ton of tokens and use a rubber band to hold down the auto-play button and leave the game on while you do something else. If you don't receive the shard before you run out of tokens, you can restart the game without saving and try again. The slots themselves are not horribly designed, in fact they are kind of interesting to play with multiple "victory modes" and different ways to trigger different effects. While you don't have control over the reels the game still feels somewhat interactive. However, by placing a shard (the set of which you're encouraged to collect) as a reward for overall winnings, instead of being an interesting diversion it becomes a frustrating experience.

Once you've decided on whether an upper bound is necessary and what that upper bound would be, then take into account what sort of distribution of attempt durations you would like to see. If a player gets the reward too quickly does that enhance or diminish the game? If getting the reward too quickly is fun, how frequently can the player get the reward that quickly and have it remain fun? What's a "natural" amount of time to need to spend to get the reward? Try and balance it so that most of the time it feels "natural" sometimes it feels quick, but not so often that it's always too quick, and sometimes it's a bit frustrating, but never so much that if the player gets a run of bad luck the game is unenjoyable.

I personally dislike any adaptive system that modifies future results based on past events. I find them to be disingenuous. But there's other ways to mitigate luck problems. For instance, if you had a 1/50 chance to get a reward, but every time you did some action you got a token and upon receipt of 10 tokens you would get the reward. Generally you would get the reward after 10 actions, but about 20% of players would get the reward early. It makes it feel like a little bonus on something you were getting anyways.

Alternatively, you could have something with a 1/10 chance to grant the reward, but reward a token per attempt and grant the reward after accumulation of 30 tokens, that way the 4% or so of players who haven't received the reward after 30 tries will not have to keep trying. This makes the attempts less frustrating because at least you're sure of advancement.

I think random chance events are fun, but I think it's important that they're implemented responsibly. If you want to have an event that can be retried, say, every 5 minutes, don't give it less than a 2% chance to happen unless you want it to take more than 3 hours of solid attempts for some players. If you have 200,000 players, even 1% of those players is 20,000. If you want them to be something that players can miss completely, make sure it's something that players will be OK with missing. Don't tie it to some sort of counter, don't make it the most powerful, don't make it the only thing that's frustrating in a sea of accessible content.

Then there's the ni no kuni scenario. If something just has a 1% chance of reaching no-longer-enjoying-the-game frustration levels, if you make players do it 100 times there's now a 63% chance that a player will be frustrated at least once.

Luis Guimaraes
profile image
For a 1/y event chance you can just roll a 1 to y*1.x+z value and pre-determine at which attempt it will happen. Like 1 to 125 in a 1% chance, or 1 to 6 in 25%.

You only need some math tricks to deal with changing of chance values, like re-calculating the remaining failed attempts based on what relative amount of them have already happened. Say your chances doubled after some buff, drop the amount of fails left by half. Your chances are halved, double the remained misses.

This is perfect for stuff like critical hit chances, drop chances, special effect chances and so on. The 1/100 "odds" will always be bad, but the closest ones (20%, 40%) will feel a lot better for the player, especially if lucky streaks don't feature any trick against consecutive low gaps between hits (like making it that only after 100 you can have the next 1% hit).

Wylie Garvin
profile image
Dark Souls:
My favorite weapon by far, is the Black Knight Halberd which can be gotten very early in the game (drop from the knight in the Darkroot Basin) but the drop rate is low, maybe 10% or something. You can increase it perhaps to 25% using humanity, and I evolved a strategy of starting a new game and rushing through to where that knight is in about 12 minutes, arriving with 5 humanity and cheesing the knight to its death. Then if he didn't drop the weapon I would delete my character and start again...

Anyway, great blog post and designers, please think carefully about how you use randomness in your games and what effects (intended or otherwise) that is going to have on the behaviour of your players, and on their experience: frustration, enjoyment, etc.!

Chris Grey
profile image
Thanks for the anecdote. I always love to hear about the character strategies people develop in the face of designed randomness. I loved pyromancy and arcane magic, myself.

To your end point, I'd add: if you aren't confident thinking about randomness, which most people are not, simulate it (or try it repeatedly) and feel it instead. Then consider the feelings and experiences that you're having and consider if that matches your design decision. That will be more useful than only thinking about it if you don't have the technical knowledge built up about probability and its impact.

I actually think many of the points of ill-designed randomness come from a lack of education about it: most designers either come from programming or art, neither of which is exposed to much math.

Justin LeGrande
profile image
Good article, though I don't think any of the titles mentioned are solely built around drawing the player in using randomness. Action RPG's such as Phantasy Star Online and Diablo are notorious for basing the endgame justification for "player progress" entirely around luck- dropped items which cannot ever be matched or found otherwise.

There are also other games, especially MMORPG's such as Final Fantasy XI, that forced the players to utilize scientific and tactical methods of maximizing their chances for rare random drops. The players would do things ranging from calculating and recording a somewhat complex calendar system, to creating spammable precise action macros which figured in the fastest performing abilities. Mastery of these things formed the basis for that game's economy, during it's heydey. Anyone not willing to put in the time devoting themselves towards learning how to load their dice was left in the dust.

The only randomizing element I'm fond of, without considering item drops or level generation, is unique character appearances which cannot be studied down to a science. They act more like Easter Eggs than anything else, but they can give a good opportunity to give a community a sense of "Has anyone else seen this?" in an age of painstaking walkthroughs and guides for anything which is even remotely predictable.

Carl Chavez
profile image
Crimson Shroud was mostly well-received, but a lot of people absolutely hated the second chapter, which cannot be finished without a rare, random loot drop from an uncommon, random monster. The lesson from Crimson Shroud is that a game should never have a random key drop for game progression... NEVER.

Personal experience: I spent four hours on my first playthrough of Crimson Shroud trying to get that damn key to drop, and two hours on the second "New Game+" playthrough.