Design Variety is the Spice of Life

[Greenfire]

In designing games, one of the most important things for a lasting game is replayability- ie if a user, having finished the game, wants to play the game again. While the most important thing to have in terms of replay value is good gameplay, I'm going to discuss one of my favorite techniques in upping the interest of the player, Variety.

What is Variety?
Google's definition of variety is "the quality or state of being different or diverse; the absence of uniformity, sameness, or monotony."

Indeed, variety alleviates the monotony of playing a game, making it fresh and each playthrough unique. Without variety, the game is the same each time, causing the game to usually be much less interesting each subsequent time it is played.

How is Variety Implemented?
Variety is established in many ways, from as small as varying your word choice, changing enemy colors slightly, to even procedurally generating levels, weapons, and foes. While the later may be hard to implement, it makes the game have almost infinite replay potential if the foundation is solid. For examples of extreme variety, look to Dwarf Fortress and Starbound. Notably, each varies different things. Dwarf Fortress varies it's lore and history, while starbound varies its weapons and enemies. Both have randomly generated terrain.

Even without going so deep into changing your game, you can reap the rewards of variety by adding small things that may be different each playthrough. My favorite techniques are changing enemy color slightly, and having multiple messages appear upon actions. While these things may not be very noticeable, not having variety here may hurt your game if the player is exposed to these lines multiple times. A good example of this is Skyrim, where the passing dialogue seems like a great idea on paper, but some lines and characters are now infamous, such as "I used to be an adventurer like you, but then I took an arrow to the knee." apparently being the backstory of every guard, which they will regale to you multiple times if your playthough is long enough.

 

[Kepons]

I would like to add that having too much variety, especially randomness, can make the game unfair and otherwise ruin it. For example, if you completely randomize the enemies of your game in each playthrough and the playthrough is usually relatively short (like an endless runner), the player doesn't have enough time to observe the strengths and weaknesses of the enemies, making them unpredictable and the player's success dependent on luck. But if the player faces more or less the same enemies every time, they gradually learn how to defeat them efficiently, making them go farther and score better as they master the ways of beating their foes.

As for variety in messages and things that don't really affect gameplay, I strongly support that and try to implement it in my games. There's also a good example of insufficient variety in Dwarf Fortress because often when you brag about killing something, no matter if it's a dragon or a mere penguin, the NPCs will say "It was inevitable."

 

[Yal]

I think it's also important to decide where variety is the most important before you start adding some - Conception II's random dungeons, for instance, are all made up from rooms the same size connected with random corridors, and the fact that all rooms have the same texture AND the same layout, this makes the random dungeon aspect completely moot, because it feels like you explore the same room over and over rather than being in this new and exciting place you've never seen before.

 

[Lotias]

In designing games, one of the most important things for a lasting game is replayability- ie if a user, having finished the game, wants to play the game again. While the most important thing to have in terms of replay value is good gameplay, I'm going to discuss one of my favorite techniques in upping the interest of the player, Variety.

What is Variety?
Google's definition of variety is "the quality or state of being different or diverse; the absence of uniformity, sameness, or monotony."

Indeed, variety alleviates the monotony of playing a game, making it fresh and each playthrough unique. Without variety, the game is the same each time, causing the game to usually be much less interesting each subsequent time it is played.

How is Variety Implemented?
Variety is established in many ways, from as small as varying your word choice, changing enemy colors slightly, to even procedurally generating levels, weapons, and foes. While the later may be hard to implement, it makes the game have almost infinite replay potential if the foundation is solid. For examples of extreme variety, look to Dwarf Fortress and Starbound. Notably, each varies different things. Dwarf Fortress varies it's lore and history, while starbound varies its weapons and enemies. Both have randomly generated terrain.

Even without going so deep into changing your game, you can reap the rewards of variety by adding small things that may be different each playthrough. My favorite techniques are changing enemy color slightly, and having multiple messages appear upon actions. While these things may not be very noticeable, not having variety here may hurt your game if the player is exposed to these lines multiple times. A good example of this is Skyrim, where the passing dialogue seems like a great idea on paper, but some lines and characters are now infamous, such as "I used to be an adventurer like you, but then I took an arrow to the knee." apparently being the backstory of every guard, which they will regale to you multiple times if your playthough is long enough.

I'm not sure that Starbound is a good example of variety. The colors can differ a lot, but gameplay wise there isn't much progression at all. After you get the double-jump and other movement options you've pretty much done all the real progression there is, and anything left is just getting new items to access new planets (whose terrain shapes look suspiciously similar to earlier planets) and grinding through the armor tiers, which is a boring task in itself.

 

[Greenfire]

I'm not sure that Starbound is a good example of variety. The colors can differ a lot, but gameplay wise there isn't much progression at all. After you get the double-jump and other movement options you've pretty much done all the real progression there is, and anything left is just getting new items to access new planets (whose terrain shapes look suspiciously similar to earlier planets) and grinding through the armor tiers, which is a boring task in itself.

Yeah, Starbound does leave much to be desired with it's progression as of the time of writing. I was mostly trying to cite mainly the procedural (?) weapons and monsters, since they (should) have a multitude of different configurations and thus be a good example of (graphical) variety.

 

[Lotias]

Yeah, Starbound does leave much to be desired with it's progression as of the time of writing. I was mostly trying to cite mainly the procedural (?) weapons and monsters, since they (should) have a multitude of different configurations and thus be a good example of (graphical) variety.

IIRC, monsters are no longer procedurally generated (and it definitely seems that way). And the only procedurally generated weapons are two-handed ones, where sometimes the secondary ability can be randomly picked. Unfortunately there's not a lot of variety there either, because there are only one or two abilities per weapon.

 

[Snail Man]

There's an interesting concept I've been thinking about in randomly generated games that I call 'macro variety' vs 'micro variety'. Macro Variety is like the general category of thing, and Micro Variety is the details of that thing. For example: Minecraft world generation; No matter what you do, you're not going to get the same world twice, because it's randomly generated. You'll always get different hills, caves, and ore distributions no matter what; this is Micro Variety. Macro Variety in Minecraft is represented by Biomes: No matter how much Micro Variety there is, if it's all the same biome, like a grassland, it's going to feel same-y. Biomes add Macro Variety which, because it's on so much of a larger scale, breaks up monotony much better than Micro Variety alone, but they must work together. Minecraft without Macro Variety would be one endless varied grassland. Mincraft without Micro Variety would be a smooth grassland next to a perfectly uniform mountain range, next to a featureless tundra.

In your example with randomizing enemy color: Yes, it makes things slightly more interesting to completely randomize the color of enemies, so that they could be any color of the rainbow at any time, but on large scales (eg- a crowd of enemies) they still look like a uniform mass. By adding Macro Variety, you could make it so that all the enemies in, say, the swamp area, are green, while all the enemies in the desert area are red. This makes the world feel more coherent, and makes it feel less like the enemies are all the same on large scales. Of course, the most effective thing would be making it so all the swamp enemies were slightly different shades of green, therefore layering Micro Variety on top of Macro Variety.

 

[Greenfire]

IIRC, monsters are no longer procedurally generated (and it definitely seems that way). And the only procedurally generated weapons are two-handed ones, where sometimes the secondary ability can be randomly picked. Unfortunately there's not a lot of variety there either, because there are only one or two abilities per weapon.

I'm pretty sure they only added some non-procedural enemies, while the majority of the foes are still the monster mashups. As for the weapons, I think the randomly found ones are still random. Who knows though, I just started starbound to check and all of my hotbar items are gone and they changed the hotbar completely. Tomorrow they might decide to make all the monsters different colored boxes for all we know.

There's an interesting concept I've been thinking about in randomly generated games that I call 'macro variety' vs 'micro variety'. Macro Variety is like the general category of thing, and Micro Variety is the details of that thing. For example: Minecraft world generation; No matter what you do, you're not going to get the same world twice, because it's randomly generated. You'll always get different hills, caves, and ore distributions no matter what; this is Micro Variety. Macro Variety in Minecraft is represented by Biomes: No matter how much Micro Variety there is, if it's all the same biome, like a grassland, it's going to feel same-y. Biomes add Macro Variety which, because it's on so much of a larger scale, breaks up monotony much better than Micro Variety alone, but they must work together. Minecraft without Macro Variety would be one endless varied grassland. Mincraft without Micro Variety would be a smooth grassland next to a perfectly uniform mountain range, next to a featureless tundra.

In your example with randomizing enemy color: Yes, it makes things slightly more interesting to completely randomize the color of enemies, so that they could be any color of the rainbow at any time, but on large scales (eg- a crowd of enemies) they still look like a uniform mass. By adding Macro Variety, you could make it so that all the enemies in, say, the swamp area, are green, while all the enemies in the desert area are red. This makes the world feel more coherent, and makes it feel less like the enemies are all the same on large scales. Of course, the most effective thing would be making it so all the swamp enemies were slightly different shades of green, therefore layering Micro Variety on top of Macro Variety.

Yeah, a huge part of Variety is not overdoing it, completely random colors just ends up with a bunch of grays, blacks and browns. Ranges of Variety might be a good term to describe the phenomenon you mention in the second paragraph. Indeed, in my game, I usually just leave a small range of color blending to be randomized while leaving the general color the same.

 

[Lotias]

I'm pretty sure they only added some non-procedural enemies, while the majority of the foes are still the monster mashups. As for the weapons, I think the randomly found ones are still random. Who knows though, I just started starbound to check and all of my hotbar items are gone and they changed the hotbar completely. Tomorrow they might decide to make all the monsters different colored boxes for all we know.

The problem with the randomness here is that it's actually hard to tell if they're properly random. A lot of the weapons & enemies feel the same, regardless of different shapes or colors. I think variety in gameplay mechanics is far, far more important than variety in colors and shapes. Introducing a new kind of trap is going to do a lot more than a new tileset.

 

[Yal]

Yeah, gameplay variety is a lot more fun in the long term than visual variety. The Binding Of Isaac basically uses the same tileset for each level every time, but the layouts and enemy placement are never the same two games in a row so it still feels engaging. Not to mention that player powerups is the single most varied thing in the entire game - they'll be affecting your gameplay the absolute most, and they're the most varied element; thus the game never feels EXACTLY the same between two sessions. Varying the player's premises is actually a really smart idea - you only need to check that a player variation works once, while making enemies feel varied require you to make sure the enemies not only can use their ability, but that they can use it PROPERLY, not to mention that it needs to be balanced for random enemy placement to not make the enemy unfair, and such.

 

[Greenfire]

I would like to add that having too much variety, especially randomness, can make the game unfair and otherwise ruin it. For example, if you completely randomize the enemies of your game in each playthrough and the playthrough is usually relatively short (like an endless runner), the player doesn't have enough time to observe the strengths and weaknesses of the enemies, making them unpredictable and the player's success dependent on luck. But if the player faces more or less the same enemies every time, they gradually learn how to defeat them efficiently, making them go farther and score better as they master the ways of beating their foes.

As for variety in messages and things that don't really affect gameplay, I strongly support that and try to implement it in my games. There's also a good example of insufficient variety in Dwarf Fortress because often when you brag about killing something, no matter if it's a dragon or a mere penguin, the NPCs will say "It was inevitable."

The problem with the randomness here is that it's actually hard to tell if they're properly random. A lot of the weapons & enemies feel the same, regardless of different shapes or colors. I think variety in gameplay mechanics is far, far more important than variety in colors and shapes. Introducing a new kind of trap is going to do a lot more than a new tileset.

As the people in dwarf fortress say:
'I concur!' 'No doubt.' 'I agree completely.' 'Truly.'
(And I think I even missed one. Didn't realize there were that many variations until I tested it.)

 

[Nehemek]

I think that variety can be done without a lot of or randomly generated content. Variety can exist in a static game, of course not to such a great extend as procedural games but sure enough to deserve two or more plays.

An example would be games in which you can either be stealthy passing right next to the enemies without killing any, or a whole species extinguishing warrior (Iji for example has many ways to be completed or Undertale to cite Game Maker games), the only important thing is that you reward the players properly for their different actions.

 

[Lotias]

I think that variety can be done without a lot of or randomly generated content. Variety can exist in a static game, of course not to such a great extend as procedural games but sure enough to deserve two or more plays.

An example would be games in which you can either be stealthy passing right next to the enemies without killing any, or a whole species extinguishing warrior (Iji for example has many ways to be completed or Undertale to cite Game Maker games), the only important thing is that you reward the players properly for their different actions.

Gameplay-wise, procedural generation doesn't have as much variety as you'd think. Procedurally generated games all have a limit before certain elements stop being interesting or get old, or at least seem very similar to previous generations. Oftentimes, as a result of poor design choices, they'll hit that limit very quickly. Terraria does it well, but there's still a point where you've done everything and seen it all. I think the amount of relevant variety is actually comparable to static games, regardless of the different land formations that can happen or so on.

 

[Greenfire]

Gameplay-wise, procedural generation doesn't have as much variety as you'd think. Procedurally generated games all have a limit before certain elements stop being interesting or get old, or at least seem very similar to previous generations. Oftentimes, as a result of poor design choices, they'll hit that limit very quickly. Terraria does it well, but there's still a point where you've done everything and seen it all. I think the amount of relevant variety is actually comparable to static games, regardless of the different land formations that can happen or so on.

It's probably that procedurally generated gameplay (besides levels) is often bad. Many shining examples of procedural generation (Terraria, Nethack, Dwarf Fortress) all have preset weapons, armors, and enemies, while the location merely changes.

 

[Snail Man]

I think that variety can be done without a lot of or randomly generated content. Variety can exist in a static game, of course not to such a great extend as procedural games but sure enough to deserve two or more plays.

An example would be games in which you can either be stealthy passing right next to the enemies without killing any, or a whole species extinguishing warrior (Iji for example has many ways to be completed or Undertale to cite Game Maker games), the only important thing is that you reward the players properly for their different actions.

Macro/Micro Variety can apply to static games as well. For example in the Mario games, each level is obviously different (Micro Variety), but the worlds also have large differences in style and gameplay, which adds the sense of progression, and that's due to Macro Variety.

 

[Gaizokubanou]

Gameplay-wise, procedural generation doesn't have as much variety as you'd think. Procedurally generated games all have a limit before certain elements stop being interesting or get old, or at least seem very similar to previous generations. Oftentimes, as a result of poor design choices, they'll hit that limit very quickly. Terraria does it well, but there's still a point where you've done everything and seen it all. I think the amount of relevant variety is actually comparable to static games, regardless of the different land formations that can happen or so on.

Indeed, I think I replayed Wargame Red Dragon's campaigns multiple times despite them being pretty tightly scripted campaign (campaign AI makes exact same moves and it is balanced around it making those specific moves).

Europa Universalis 4 is an example of fixed start but very complex system that grows into something different each session. I must have played that for hundreds of hours.

It's interesting that most procedural generation games tend to boil down to finding some boiler plate strategy to rinse and repeat. Really good authored content is better than average P.G. stuff even for replay value cause of that IMO.

I think Xcom 2 did good job with moving into P.G. maps but even those... there are few standard map logics that you see very often. They are very good logic mind you but still.

Binding of Isaac Afterbirth is really amazing in this regard and comparing it to Gungeon highlights just how brilliant it came together. I'm not saying Gungeon is bad (I think it is quite good) but the game does boil down to handful of dominant guns you pick up (so ton of variety in them is lost). But in Isaac everything you pick up adds to the run cause you are building this singular 'gun' (your tears) out of all the parts instead of choosing from one preset thing, and it's way more refreshing each run. Pretty amazing really.

 

[Zekka]

I think many of the games that do procedurally generated stats are just concerned with picking all modifiers from a group at random from an acceptable range, and in theory nothing bad happens once you've done this. There's really nothing in that to anchor you to the game. You're going to be interacting with the world the same way and you haven't been presented with anything that tickles you emotionally.

Compare Dungeon Crawl. Most of the procedural events in Dungeon Crawl don't involve wads of RNG-ed stats or special procedures -- we're talking hardcoded, simple passives. A lot of the gameplay subsystems are a little more complicated than they look at first, but it's still closer to Pokemon-level complexity than Dwarf Fortress-level complexity. The interesting thing about Dungeon Crawl is that a ton of the things you can encounter will have a long-term effect on your character. You can play it as a game about accruing procedurally-placed bonuses in the form of gods, armor, weapons, consumables, spellbooks, rings, and mutations -- you can also get cursed, non-beneficially mutated, drained, banished, and so forth, and those are all strong negative effects. This changes it from a luck game to a hedging-your-bets game.

(Dwarf Fortress IMHO does a lot of the same things, but it also focuses on what the devs call "narratively-interesting" simulation -- mostly stuff that has sensory grounding IMHO, like tasty food, musical performances, political intrigue, etc. I think that's really important and often completely neglected because it's hard to do. Another game that does an amazing job at everything I listed in this post is SS13, which is entirely about your character getting gimped until you can no longer function.)

I think Dungeon Crawl would be made worse as a hedging-your-bets game if you randomized the attributes of these miniature challenges. You can only play a hedging-your-bets game if you can predict the general kinds of things that are going to happen without knowing exactly which. There's a very tough fire resistance challenge at the end of the game, and no matter what, it's always fire. Giving people a really clear hurdle like that means they constantly have something to hedge their bets about. If the final challenge could be any effect type in the game, not just fire, then you're dealing with a luck game, not a skill game. If you'd rather it wasn't a luck game, because you want to give unlucky people a fighting chance, you have to make the challenge easier.

This kinda reminds me of the problem in joke-telling where if you have no idea where your humor comes from and just splatter the page with funny things, you get something completely unfunny. Procedural generation is usually a trick to maximize how much different mechanics in your game can interact. If you splatter your game with it for other reasons, you will not necessarily get an interesting outcome.

Starbound I think is an example of how this can work very badly. Starbound's mechanics all act almost instantaneously, the combat is teaspoon-deep, and procedurally generated zones are built out of vaults that are really large -- so there's really not a lot of room for anything to interact with anything else. Most of the things that are related to the human experience -- tasty food, musical performance, intrigue, etc -- were handwritten by the devs and not a lot of attention was lavished on them.

 

[robbertzzz]

Besides adding variety through different colours, different worlds and what not, the biggest thing to add variety that makes the game replayable is variety based on player actions. A player would never make the exact same choices at the exact same moment when replaying a game, so if you make these decisions influence the game enough, every game would be unique.
But, the aforementioned problems are even a bigger problem here: if you change the world based on player input, it'd better be clearly noticeable, or it won't have any added value compared to "normal" procedural world generation. And the world generation in itself should of course be designed very well, no matter the input it gets (either a random generator or user input).

 

[Lumenflower]

I think it depends on what people hope to get out of a game. I've seen many games with little-to-no variety in gameplay and visuals which people find very addictive. I think the biggest guilty party are mobile games, where players want to simply whip out their phone and tap away for five minutes while waiting for a bus. These are the types of players who want to get a high score and compete with their friends, but don't want to get too involved in the world of the game. There is nothing wrong with this - it's a demographic that needs exploiting.

Personally however, I play to experience new things. I treat a game as a world to explore, and want to complete levels and areas to see what fabulous creations lurk round the next corner. For me, variety is very important because it's what keeps me engaged. Too many times, I've got to a point in a game where I realise there's no longer anything new coming along, and lose interest.

 

[Misty]

I believe that Procedurally Generated Levels should be done as a side mode (for instance, Instant Action or Multiplayer Maps) but the main campaign should be more solid and done by hand.

Otherwise, it starts losing it's coherence and gets feelings of "splatter syndrome".

Think of it like music, you wouldn't want to listen to a whole CD of procedurally generated music.

I mean, none of our favorite childhood games became famous using procedural generation.

Videogame levels are kind of like an art gallery, and procedural levels are a bit like going to an art gallery full of paintings with no artist, or a game with only procedural levels is similar to a gallery full of splatter modern paintings and nothing else.

Varying enemy colors can be accomplished using Day Night cycles or varied Ambience settings. Stray too far from this, and the game becomes erratic, with players unable to learn or discern enemy patterns and strategies.

As Druid assisted me in pointing out, procedural generation can often backfire, and instead of creating the effect of levels and gameplay that seem fresh and new, they actually make the experience of new levels seem duller and more lifeless as a result.

 

[Snail Man]

I mean, none of our favorite childhood games became famous using procedural generation.

Don't forget that for millions of people born around 2000 and later, Minecraft is the childhood game. Just because none of your childhood games primarily used procedural generation doesn't mean that nobody's did.

Like anything, procedural generation is a tool. It can be done well, and feel like a new game every time, or it can be done poorly, and feel same-y and unfair. I'd like to think that it might be slowly moving towards the former as the industry as a whole refines their approach

 

[Misty]

Oh dear. How can I comment on the mental states on the younger generation while trying to remain tactful?

Nikki Minaj, Justine Bieber, Minecraft, are some key terms I would throw out there.
I don't understand how a mind can wrap itself around the voxellated sameness of minecraft and derive some sort of aestheary gratification from it.


I grew up with Lego, and while it's true that Lego is gridlocked, it provides a sample pallete full of curves, ramps and arches, an essential ingredient for any aesthete who desires harmony of mind. Minecraft, a second rate imitation of what Lego executed flawlessly.
The artistic design of Minecraft is no less than physically sickening to observe. Especially the movement flow, walking on a block "terrain" results in a very jerky movement flow, and the player is frequently forced to "jump" (hit the input button and put conscious effort into it) constantly to circumnavigate, this results in a very unrelaxing and exerting movement experience.
The repetitive cube like nature, causes nausea to my consciousness, I don't see how people can tolerate it, but they seem to tolerate the alarm clock music of Nikki Minaj, I'd rather sit through an hour of Justin Bieber than that. Though in alarm clock's defense I've seen a few decent tracks utilizing that instrument, such as the Dark Samus/Escape track in Metroid Prime 2 Echoes and a couple of dubstep tunes that weren't overly nauseating.

A game done exclusively in procedural generation, can be interesting in rare cases, such as a dungeon crawler. The core interest of these type of games usually revolve around survivalism, it is the type of psychology needed in order to really "get in" to the procedural generation of the game worlds. Other games, where the survilalism type psychology (or "contestant"/gameshow psychology) is not so dominant, the atmosphere is more focused on order than chaos, or balanced order and chaos, and so such procedural levels don't really give a solid grounds for player immersion, eg. Half Life 2 would not have been as solid or immersive if it used procedural generated levels in it's campaign.

 

[Greenfire]

I mean, none of our favorite childhood games became famous using procedural generation.

Actually, procedural generation has been very successful out of the mainstream for quite some time. Nethack (1987) and Dwarf Fortress (2006) are prominent examples.

 

[Misty]

Never heard of them, so they can't be that famous.

 

[Lotias]

Never heard of them, so they can't be that famous.

 

[Yal]

Actually, procedural generation has been very successful out of the mainstream for quite some time. Nethack (1987) and Dwarf Fortress (2006) are prominent examples.

I'd say Minecraft and The Binding Of Isaac are much more relevant examples due to how mainstream they are, and that sentence sounds like a quote from a wikipedia article as well. Tsk tsk.