Why Procedural Generation Is Never as Good as Scripting | Slightly Civil War

[@theescapist]

Watch today's episode - Can Movies Based on Video Games Actually Work? www.escapistmagazine.com/v2/can-movies-based-on-video-games-actually-work-slightly-civil-war/ Podcast: soundcloud.com/user-944993929/why-procedural-generation-is-never-as-good-as-scripting OR search Slightly Civil War on Spotify or iTunes.

[@barrybend7189]

Pokemon Mystery Dungeon why not both?

[@DPCP-h9u]

>Movies based on videogames can work? Most of the time, no. But games based on movies work amazingly. I still play The Warriors and Scarface

[@cian2168]

Jack's performance here was his best yet.

[@fireisawesome1466]

Chess already has procedural moves it's to keep things interesting what if a.i makes mistakes and has no/had a plan isn't that more interesting

[@lydiahollowmoor4883]

Might I be so bold as to suggest that watching two people forced to argue a point that they don't necessarily agree with, where even if they do happen to believe there is real merit to their core argument, are still forced to embellish that point and needlessly argue against valid thoughts the other person might have is inherently less entertaining and enlightening than if they were simply allowed to just express how they actually feel on the matter? I realize that's what the podcast is for but why is the podcast not available on KZbin?

[@jax3636]

"You're talking about procedurally generated games avoiding cliché, I'm saying procedurally generated games are basically just collections of clichés assigned in random order" Damn that was smooth ngl

[@trymbruset3868]

I am completely flabbergasted nobody mentioned replayability.

[@squiddler7731]

_Same._ Also feels like they didn't really describe procedural generation accurately either. The idea of some robot building a world autonomously without any human intervention is a myth. Procedural generation can involve just as much human effort refining and testing the system to create a desired output as crafting levels by hand, and the two overlap far more often than people realize

[@moartems5076]

Same

[@echo.1209]

But they did, though. They talked about how it all feels “same-y” after a while. If you have 9 different rooms, there are 362880 different ways to order those rooms. This is technically “replayability”, because you’re unlikely to get the same room several times unless you play for a long time. But, from a player’s perspective, it’ll get stale fast. Having 9 different rooms per level is a much more enjoyable experience than the same 9 rooms organised differently. Replayability means nothing if you can’t enjoy replaying it. I’ve put hundreds of hours into games with a linear progression and generally have had a more enjoyable experience when it’s meticulously designed to be so. This is purely my anecdotal experience, though.

[@MangaScrub]

I’m personally surprised they didn’t bring up We Happy Few. It’s not roguelike but it is procedural to its own detriment.

[@trymbruset3868]

@@echo.1209 I agree with you to an extent, I have probably played SM64 and ocarina of time at least a hundred times each, and still enjoy them. Thing is though, that I've put a few hundred hours into binding of Isaac, and to me it doesn't feel samey at all. You have different characters, modes etc, but all just using regular Isaac, the variation in floor layouts, item drops and rng elements make it just as interesting the 200th time around, if not more. You simply can't get that from scripted games, it's a different type of experience which is not in general worse.

[@HamHamDude]

1:45 "Try as we humans might, there is no new story to tell; no new game to play." Jack's right. Nothing is new. Now if you'll excuse me, I'm going to strap a vive pro on my head and slash apart flying boxes with lightsabers, just like my forefathers did.

[@thomastakesatollforthedark2231]

Youre bassicaly just a one man orchestra, playing the sound at the same time a conductor (or boxes here) tells you to while you try and match the beat

[@Sylfa]

You mean Fruit Ninja with extra dimension and new peripheral? He both has and doesn't have a point at the same time with that one, "everything is a remix", people don't have brand new ideas, they have evolution of previous ideas. It's how they connect together that makes the story new. It's how the whole Rogue-like and Rogue-lite genres came to be even.

[@confusedmatthew3736]

I have nothing worthwhile to say, but I paid two dollars for this loyalty badge and I'm getting my money's worth, dammit!

[@tetragears]

"Game developers are not factory workers, they are artists!" As a game developer i cannot disagree more.

[@thomastakesatollforthedark2231]

What gAme did you make?

[@tetragears]

Thomas takes a toll for the dark, i cant tell, NDA and stuff, but i worked on some big mobile pvp titles.

[@Sylfa]

Most people in game development are more like factory workers than artists, *even* if they are making art for the project. Sure they get to work with their chosen artistic medium, but they have to follow the design guidelines that have been decided by just a handful of people running the show. It has to mesh with the entire project so it makes sense, but if you remove artistic freedom, how much artistry is left?

[@highspeedfistfight]

@@tetragears Mobile pvp titles are more often than not gigantic greed factories that prey on addiction. If I'm wrong, feel free to start listing off all those mobile pvp games that are balanced and not designed to cause addiction instead of enjoyment.

[@LomanLawson]

@@highspeedfistfight not sure what that had to do with his statement of being a factory worker...

[@faustlican5566]

I adore just how animated Jack gets when responding to Yahtzee's quips sometimes (pun not intended)

[@jafaral-z1090]

he's only human unlike the robot called Yahtzee (has he ever laughed in a vid?)

[@SweetKiren]

@@jafaral-z1090 I believe he laughed in one of the podcasts that happen after these episodes. I don't remember wich one it was though.

[@sameggenton1077]

SweetKiren clearly it wasn’t him then...

[@SweetKiren]

@@sameggenton1077 nah you right, it was probably someone they hired, so people wouldn't find out Yahtzee can't laugh x)

[@TrueMentorGuidingMoonlight]

I think it depends on what kind of game you want. Having procedural generation would be great for an exploration based game where every player’s experience is intended to be different, especially for every new playthrough. But for a game designed around stealth or combat, you need a scripted map design. There’s a reason why “good level design” is a thing. Ever play FEAR or Thief the Dark Project/Metal Age? Those kinds of games rely on the immersion of a well-made map with realistic architecture for traversal. FEAR’s nooks and crannies set the stage for the actually-good AI enemies to take cover and flank you, using the environment to protect themselves from your onslaught. Thief’s buildings were designed to have looping paths from interconnected rooms and hallways, to make you feel like someone actually works/lives in the place you are breaking into. You can’t replace the experiences of FEAR or Thief with lazily made, procedurally generated maps.

[@NijiharaKaito0]

Since you mentioned a good use case for procedural generation but didn't give an example, I'll give you a recent one: Deep Rock Galactic. Every cave is different and even though you can find similar "rooms", they are never the same on the inside or on their connections. It gives the game a good exploration -> escape loop to every dive because you always have to put some effort in looking for the objective and running to the pod for extraction. Of course, when leaving some work to an algorithm there is a line that should never be crossed. I don't know where this line is but when a game does it, it's visible and there is no lack of examples, sadly.

[@TrueMentorGuidingMoonlight]

Kaito Nijihara You have a good example. Although I always wondered: how exactly do devs reduce the probability of a crappy map? Like say, an unreachable objective? Or what about when, like say in Minecraft, you have all the emerald ore blocks in a convenient location, thus making the game too easy from the start?

[@NijiharaKaito0]

@@TrueMentorGuidingMoonlight Colin Velius they don't, traversal is a big part of the fun, I think. All characters have a pickaxe, a different mobility tool, and the entire map is digable. There can be ore deposits high up in a wall and you need to figure out how to get to it using your tools (in single-player) or the synergies of different tools (in multi-player). As for not making things too easy, there's never enough ore in a single room to complete the quota and ore patches have a limited size so you can't get it from a single spot in a room (the further you go from spawn the more patches you find per room, if the objective isn't mining an ore, it will be some minutes of exploring away) , there are multiple ores to be mined for different reasons (weapon and gear upgrades and ammo resupply), all are important but only one is the main objective, and, finally, the extraction point is never too close to where you asked for it, so there is always a hectic escape.

[@SomeRandomJackAss]

@@NijiharaKaito0 That all needs to be coded/scripted. Where does one draw the line when designing a game where the rules for procedural generation become so complex that it's just easier to make levels? Sure, there's the novelty of procedural generation producing somewhat unique maps, but everything lies in finding a balance.

[@Secondhelix]

I play games for story, and see them as a storytelling medium, but not exclusively so. Mario kart, nidhogg, tetris: all fantastic games, but no story. Procedural generation does not have to decrease story impact, but can be used either in lieu of story, or as an accent. In binding of Isaac, the procedural generation helps give a "lost" feeling as you learn about how his mother treated him, for example.

[@utisti4976]

Same.

[@thomastakesatollforthedark2231]

And a game like Dead Cells which tells a pretty good story with it's random generation and even stops the generation after you kill the king.

[@SwevenHannibal]

Jack always makes the feeblest of arguments and defends them in the weakest way possible. He is *always* talking out of his ass. I love him.

[@barrybend7189]

Well you can do both like the Pokemon Mystery Dungeon games.

[@pokemonmanic3595]

Yessss

[@cat11112222]

@@pokemonmanic3595 your name checks out lmao but explores of sky is on a whole other level of greatness

[@ritenac6770]

God I love those games. Fun to play and surprisingly emotionally investing with the stories

[@jonlangthorne7924]

Good point well made, I loved mystery dungeon! Surprisingly engaging and you get invested.

[@captainmystery4123]

@pokemonmanic3595 where have I heard your name before?

[@chrisjones5949]

Yahtzee already won this argument with the various comments he's made in ZP over the years denigrating an overreliance on procedural generation. It has its place, but it can never truly be as good as a well-crafted experience. His point about shuffling the chapters in a book was stated better when he previously brought it up: basically, you have to choose between nothing of any real consequence happening in each chapter so as to maintain the status quo, or the book not making any sense. ProcGen is a tool, not a design document.

[@Sylfa]

"can never truly be as good" Never is a very long time, AI is getting better by the month. No one can ever win against a human in chess/go/etc has been a point that was claimed in the past. For current time though, there's plenty good with a good well made linear or multi-branching story line. But some games are all about the game play, and not the story. Those in particular do very well with well balanced procedural content. See Binding of Isaac, Minecraft, Terraria, FTL, Diablo 2, All 4X games.. The list is looong. You could even add any and all card games as they essentially work the same as procedural content does, defined rules with randomisation. And as for the future, I have no doubt that you won't be able to tell in 20-100 years time if the story in the game your playing was made by a human or an AI.

[@jeice13]

It does depend on what is supposed to be enjoyed about a game. If it is supposed to be a challenge where you are adapting and planning as things come up scripting wont work but if you are trying to create a specific story or feeling then it is probably easier to script than to exhaustively constrain the generator until it cant tell any other story

[@chrisjones5949]

You guys do make some excellent points, but this is also crossing over into whether story or gameplay is more important, which is a totally different episode of SCW. ;) As for AI being able to design a game - well, that's not quite the same thing, is it? The smartest AI in the world could make a game whose levels are entirely ProcGen, and they would still feel kind of samey and repetitive and inconsequential because they were made that way. You'd need to give it the idea to make a game with at least a somewhat structured and linear experience in mind, no matter how smart the AI was.

[@Timmy_T]

Ah, blowing Raspberries... That term surely can't take a wrong turn...

[@gokusondbz]

Lol.

[@prospero4183]

I'm sure a presidential candidate who blow raspberries will win arguments

[@daishoryujin95]

@johnTaylor delete this

[@watergod321]

Argument for, REPLAY VALUE! C'mon...it is the single strongest argument for procedural generation being as good as scripted

[@geminialchemist2441]

While I know many would use procedural generation as a huge replay factor, I myself would argue the opposite. I find that the games I return to the most, I do so for the story, or the characters, or the level design, or a combination of the three. Because of that, I find that once I'm done with a procedurally generated game, I have very little reason other than the gameplay to return to it. They just don't have what I value most in a game that makes me go back to it in the future.

[@wtrbns]

@@geminialchemist2441 It seems you prefer story-driven games, which is fine and you'll probably get less out of procedural generation. On the other hand, gameplay, strategy and a sense of victory can be greatly enhanced when you don't know what's next and much of the predictability is removed by random generation. It also makes victories feel more "real" when you're less guided by the invisible hand of the developer.

[@SobiDani]

It feels strange to me that they were talking about games but didn't bring gameplay up at all.

[@sarafontanini7051]

well its not easy to say what procedural generation adds to gameplay since procedural generation isn't ACTUALLY directly tied to gameplay unless you want either a totally boring game or an incredibly unblanced game that could randomly spike or lower in difficulty willy nilly

[@Sylfa]

​@@sarafontanini7051 4X games would like a word. Also, "totally boring" and "incredibly unbalanced" are a result of poor designer skills, not of Procedural Generation. You can't just throw in random table lookups and think your doing Procedural Generation, you need to have a Procedure to it, it's in the name. Bad Procedure = unbalanced or boring, don't do that.

[@andregon4366]

Or ambience and build up. Two things procedural generation can't do properly.

[@andrewhickinbottom1051]

Another great discussion! This also applies to open world games, but i always prefer finely crafted, linear and memorable games over open world or procedural ones. Sure, they can keep you playing for much longer (possibly endlessly - ive racked up hundreds of hours on both American Truck Simulator and Star Citizen), and you kinda make your own story with the events that are conjured up in the game, but my fondest gaming memories are always things like Metal Gear Solid, Ace Combat 6, Mirror's Edge, Silent Hill 2 etc - linear games with finely crafted levels and cutscenes.

[@thomastakesatollforthedark2231]

It's what I like about Breath of the Wild. It's open world has a story to tell which, while can be told in many orders, always makes sense and keeps you engaged because of it

[@Sylfa]

Both type of games are fun in their own way. The discussion was a bit one-sided though, it didn't seem like they thought of many reasons *WHY* Procedural Generation is good in some type of games. Limiting it to one genre wasn't necessarily a good thing for the discussion either. 4X games would like to raise a point...

[@freekbertens4729]

All i hear is warframe being chanted from the other end of the cave

[@aravindpallippara1577]

It's just tilesets being stiched together in randomised order, I sometimes think it would have been better if DE actually sat down and designed good levels - it all just blends in after the first few dozen hours to be honest. On the other hand deep rock Galactic have a more truer form of procedural map generation, that's a really nice system they have got there.

[@graemevaughey7432]

I love how the title of the video is not "can procedural generation ever be as good as scripting?", but just straight up "why [it isn't]". Slightly leading the witness there. I mean, how is this even a debate?

[@StudioAREshorts]

4:55 Yeah, but Yahtzee, not ALL games are stories, some games are just arcadey fun. Or other games make perfectly good use of semi scripted instances combined with generation to slow down how fast a player "Depletes" a game. (Enter the Gungeon as an example, it has a story of sorts, but nigh infinite replayability)

[@madmanslime3188]

Yeah, found it odd that he stated it as fact. He is really is writing of an entire heap of games. In the story vs. gameplay podcast, Jack confronts Yahtzee on instances like this and Yahtzee says something along the lines of "The story is the story you make yourself" (ie. the story of the run in which you got a highscore). A pretty weak defense (seeing as no one plays games like tetris for an intreguing story) but it's better than nothing.

[@MazeFrame]

I got a bad idea: 1) Take a programming language that can alter itself (or at the least create and run a programm in itself) 2) Let that programming language create code to create graphics/sounds from inputs, all randomly assigned by some loose parameters 3) Have the game alter itself based on player input (which is pseudo random) and the random number functions of the CPU 4) Kickfarter/IndieNoGo 5) Put on Epic first (for extra profit) 6) Profit

[@d3line]

MazeFrame Indeed, this is a bad idea. If your game can generate correct non-trivial code and graphics from random inputs what you got yourself here is a self-modifying AI. That thing is ether solving all world’s problems or (more likely) destroying it, so putting that tech in the game store is really small thinking;)

[@judeclark397]

This argument should have just ended by Jack just saying "Not every game is driven by linear story." A lot of games are just about enjoying the game play and being able to play it many times due to randomness provided by procedural generation. Think about it this way, if the Binding of Issac was linear one of it's core mechanics, the acquiring of many random different items through different means, wouldn't work. Without this mechanic it's not even the same game anymore and will most likely kill replay value because once you've gone through it once you've seen everything the game has to show you. Yahtzee in this case is suggesting you just scrap an entire genre because it doesn't have a linear story even though that was never the focus to begin with.

[@SomeRandomJackAss]

Technically, The Binding of Isaac has a linear story. It has procedurally generated gameplay. Procedurally generated story would be like a text adventure game written with an AI or something.

[@imageez]

@@SomeRandomJackAss yeah it has a background story you can wiki and debate on online threads but you dont have to know the details of Isaac's familial issue to enjoy shooting tears to demons.

[@swaghettimemeballs4420]

You just gave me the sickest idea. What if Binding of Isaac DID have a story mode? I can't stop imagining one big ass level that consists of about 1000 rooms ranging from big to small, where you have multiple ways of making it through the level. They could make a new one every month, and it would encourage people to keep playing the game!

[@MrWhygodwhy]

It is an increase in replayability at the sacrifice of interesting content. Procedural Generation operates on the same logic grinding games do: Quantity of hours > Quality of hours.

[@sarafontanini7051]

@@imageez except this is untrue. Isaac's story is told pretty blatantly in cutscenes and what cna be debated is the meaning behind certain endings but its not like it was just written on a wiki, the story is IN THE DAMN GAME. Same for Enter the Gungeon, which has the story of its different main characters as the reward for you finding the things needed to get their endings and backstories and fight their unique final bosses. and there are OTHER elements to these games that make them work outside the procedural generation, in fact the procedural generation is probably the least important aspect but what IS generated and how its managed and there are too many rogulikes and other games that use procedural generation poorly and lazily which is why its become maligned.

[@Xfushion2]

The issue here is that procedural generation is a scripted system by itself. it must be crated first by taking into account the experience the game is trying to portray. Yahtzee points out that because procedural games are _random_ elements like the difficulty curve go away and it's not necessarily true, sure a bad Procedural system will just throw random game elements without limitation, But a good procedural system like Spelunky gradually increases it's own difficulty by limiting the elements it can generate because that system takes into account variables like player time and cleared levels to generate levels according to the Player skill gap. And that's what make spelunky a compelling game because even though is procedurally generated a run in the earliest stages of the game is no way close in difficulty and experience to a late stage. Same goes to games like Slay the spire.

[@megan_alnico]

All automation removes a human from a part of the process of creating something. Eventually though, people understand that the automation is just another tool. At one point of time a tool like Photoshop was not considered a 'valid' from of art. It was a cheat, too easy, and removed the true artistic merit of the work. If you go back far enough Photography got the same treatment. So what we have here with procedural generation of content is the early days before people understand what it's good at, what it's bad at and how to use the tool. We will at some point reach the stage where it's used everywhere and badly, like autotune in music. Eventually people figured out how much was the right amount of autotune, and the same can be said for procedural generation in games. Getting it right, that's the art.

[@vladtheinhaler93]

Spot on, before the camera, the primary work of a painter was portraiture, after the camera, paintings became more impressionistic, surreal, or abstract, in other words; less figurative and more imaginative. That is the point of technology and automation; minimising the grunt-work so human minds can focus their attentions on 'higher' tiers of productivity. It's a fault of the economic system that people rely on that grunt-work for their livelihoods (and that 'creative passion' can be so easily exploited), not the technology..

[@voidstrider801]

A few games have got it right in a matter of speaking. Nothing might ever replace a well crafted world or set of levels for the most part, but offering procedural generated content along side it works well. Example: Daggerfall uses a mix of procedural generation and pre-built areas, many dungeons are randomly generated, but not all, the over world, towns, cities etc are always the same. As far as using just procedural generation well, the Diablo trilogy clearly nailed that well, never really heard anyone complain about it in that series, things are always familiar but different, rearranged and shuffled, but never in a way that looks or feels off or completely random, there is method to the madness.

[@toperishtwice]

I don't really like these. You guys should either agree with your argument so you're invested in giving them the consideration and representation they deserve. There are so many points that you guys miss when you don't actually agree with your point of view.

[@OdaSwifteye]

I'm pretty sure they argue for the sake of the format. They probably agree with the subjects they argue about.

[@alangreenway6695]

You know it’s got real with a ‘Sonny Jim’.

[@devonreid8719]

Like your mum

[@CWargh63]

And then the "So there."

[@but_in_space_though2919]

I knew this was going to be a great episode, with how Yahtzee often talks about procedural generation in ZP.

[@Mas-cm6cy]

I Love how they are not really fighting, but are more like "oh i got this, lets have a chat about it then"

[@Eyeball117]

I'm impressed they got through a whole debate about procedural generation without mentioning minecraft

[@sarmavlada]

"Well it isn't so there"... damn that's some counter argument.

[@dwaynezilla]

Oh wow, this is so good. The points they both make, and the delivery. I'm really pleased with this pair-up, and putting them in a Slightly Civil War set-up is a great idea. Great work, you guys. And great work The Escapist for getting them together!

[@slothfulcobra]

This one got weird. I feel like Yahtzee tries harder to make Jack break and laugh than he does to debate.

[@Gisso94]

Scripted games tell a story. Procedural games let the player make his own story.

[@mystbunnygaming1449]

Damn I had some fun with Mario RPG randomizer lately. It's a fucking awesome game on its own, but shuffle characters and items around, and I found myself coming up with a new story for the game. I found Bowser at Mushroom Way getting cornered by a goomba, as if there was a rebellion, and the entire game became the story of crushing the rebels and putting Bowser back on the throne. It was gloriously satisfying. And also Mallow getting married to pogo-shy-guy, and Boomer bursting in to object to the union. So we killed him.

[@bernardoheusi6146]

Yes. Undertale

[@biker9476]

Kinda? Procedural games gives you a random set of assets. You can fill in the gaps to *make* a story, but you’re not the one telling it.

[@SomeRandomJackAss]

Emergent storytelling is a thing in scripted games, too.

[@sarafontanini7051]

not entirely, it dpends on other factors in the game, not purely the procedural generation which

[@digitaldeathsquid3448]

I think Jake Swearingen of New York magazine made perhaps my favourite point against procedural generation when critiquing the initial state of No Man's Sky - "You can procedurally generate 18.6 quintillion unique planets, but you can't procedurally generate 18.6 quintillion unique things to do." Procedural generation of level design can, at times, mess with the primary gameplay loop, depending on what sequence of levels, rooms, enemies and objects the game throws at you. It can also lead to environments feeling same-y, as once you've seen everything in the sequence, that's kinda it. Structured games and levels, on the other hand, can slowly introduce you to new areas with new themes, keeping the environments feeling fresh - Subnautica, Metroid Prime 3 and Prince Of Persia: The Sands Of Time are games I would herald as quite good at this kind of structure and diversity in level design. Each has areas (with the first two mentioned having somewhat non-linear elements) with each environment, or set of environments, feeling unique despite taking place in the same general area (granted, the area in Metroid Prime 3 spans multiple planets and occasionally star systems)

[@HorriblePodcast]

I love story based games but I've been playing Dead Cells for almost a year and a half and I'm not tired of it yet. I haven't gotten this much replayability from a game in the last 20 years at least.

[@Daniel_Schmaniel]

Recently, I have been playing Remnant from the Ashes, a game which is often touted as procedural generation done right. It does have many qualities that make it a great game however I would not say the level design is one of them. I am not sure how much of the game is procedural but many dungeons are put together in a very confusing mess. Concourse of the Sun is a dungeon on the world of Rhom that really fits the bill for this. We could not find the exit for ages because the exit was on a different level than the rest of the map. I eventually got lucky and saw it pop up on the minimap while going down a flight of stairs. None of the levels feel right, like it was organically designed. It ends up making things needlessly confusing.

[@stevezpj]

Surprised there was no mention of Dwarf Fortress, especially Boatmurdered :D

[@denimchicken104]

Yahtzee banging on about creatives working hard because they are passionate is only partly true. He sounds like he’s speaking from the perspective of someone who is his own boss, which is a very different situation from the creatives who are part of a large team with a lot of corporate oversight. Exploitation in the game industry of creative passion is a very real problem.

[@sarcasmismyfavoriteemotion4180]

Having wrote a small (10ish pages) essay on this topic, I argue that both sides are wrong and right, depending on the situation. Procedural generation has its uses - replayability, adding organic moments, accessibility for small teams, etc. But, it also has many downfalls. A huge issue with it is that a computer cannot easily determine what is fun, or even playable. As of now, the only way to ensure that a procedurally generated level is playable, and more importantly FUN, is to lower the amount of procedurally generated content. Some practices are emerging that can overcome this hurdle (Machine Learning is even becoming prevalent, but that's its own can of worms), but are often complex and unwieldy to implement into game development. In the end, each had a type of gameplay it supports well, and a type of gameplay it isn't suited for. The situation you would use a hammer is significantly different than the situation you would use a scalpel for. I may love roguelikes, but I wouldn't expect a story centric game like the Last of Us to implement it.

[@Medytacjusz]

It's easy for Yahtzee to say that the artist wants to toil and there's no exploitation because he's an indie dev that works mostly alone and not in a big team and has total control over his work

[@TheRABIDdude]

This was really fascinating! Started off with quite practical arguments, then ended up arguing about the philosphy of art and interpretation. What a ride.

[@NekoiNemo]

Is this argument even valid? That's like comparing a hammer and a screwdriver - both have their uses, and both are better suited for certain kind of task. And, best of all, you can't really have a big impressive project without using both together. 4:40 Gotta disagree there. Books are storytelling mediums, movies are to some extent storytelling mediums, games are primarily an activity, and story is only a secondary concern, if one at all. Case and point - you can have game with no story (pong, most "arcade" type games), but you cannot have a game with only story and no gameplay (the so-called "kinetic visual novels", that have no branching and all "gameplay" consists on pressing "next" until the credits roll. Tehy are only classified as "game" by the same principle slideshow of book pages is technically classified as "movie"). 5:35 Ah, but you see, that's where you're wrong. Procedural book wouldn't be pages in random order - it would be a basic story structure (say, a Hero's Journey) with stuff like setting, characters and motivations being decided algorithmically. And didn't i just describe 80% of all books, movies and narrative games ever made? Are you familiar with "women's soft cover books" and jRPG games? You know, the entire genres of media that feel like they use the exact same structure, but just swap set dressing in and out? Do you really think you couldn't generate stuff of equal or better quality with even today's level procedural generation algorithms?

[@herald9204]

As a counterpoint, Xenoblade Chronicles is story based with a scripted world, which exists as a storytelling medium with gameplay. There is an entire genre around visual novels with a branching storyline, and procedural generation in a story based game can only get you as far as Dead Cells or The Binding of Isaac.

[@sarcasmismyfavoriteemotion4180]

I actually just wrote a small essay on this topic, and yeah. Each has its uses, when it is advantageous and disadvantageous to implement it. You wouldn't use a hammer when a scalpel is called for, nor the reverse.

[@thecynicpyro]

Wow you really took this seriously

[@NekoiNemo]

@@herald9204 I think you missed my point completely. Xenoblade Chronicles may have a lot of story, but it is still a gameplay driven game, not a story driven one. I'm not quite sure what was the point of bringing the VNs up, considering *I* specifically mentioned them already, but ok. As for procedural generation part... Boy. Are you familiar with a little indie game called Dwarf Fortress? You know, the one where procedural generation produces a detailed world the size of Eurasia, with couple thousands years of simulated history that isn't just lore fluff, but actually affects the political climate, racial culture, locations of artefacts, and many more things. Yep, that's all procedural. But once again, that was NEVER my point. My point, that you had missed, is that games are ACTIVITY (as in player interacting with the medium, not just passively absorbing it like with a book or a movie) based medium, and as such gameplay related content is what really matters, not narrative related one. And while procedurally generating narrative that is cohesive is indeed a bit harder, procedurally generating content for the gameplay is not, and was already done to the spectacular degrees (once again, case and point - aforementioned DF).

[@Terrik240]

For: Procedural Generation (PG from here) can add replayability that can only stem from either a masterwork of manual authorship or the base level requirement for PG. Games like Minecraft and Terraria are entirely dependent on PG to make everyworld a unique adventure in that you won't be building your house in the same place or of the same material with every single adventure. Against: PG limits the type of game it can be. Games like Far Cry and Assassins Creed would be become senseless and mindless (well, sooner than they actually did) if everything to do was randomly generated, and where it was placed left to RNJesus. Games like Doom and Wolfenstein would be hard pressed to find cohesive gameplay among the random rooms and corridors that sometimes to lead to nowhere, something you don't want in a high octane thriller. Games like Outlast and Alien Isolation would be boring at best and unfair at worst, as the scares either fail to do anything because they happened where the entity could never possibly reach you, or you became trapped at the end of a tunnel at every turn.

[@templariclegion2826]

I didn't hear this brought up, but replayability is something to think about. And, when something crazy happens, it's random and is more impactful.

[@FrBrodyTime]

Guyssss the point of procedural generation isn't for more compelling stories, it's to add excitement and new experiences to inpossible roguelikes such as Nuclear Throne and Enter the Gungeon, because the games are immensely difficult and the player WILL keep dying over and over again, so the prodedural generating keeps things fresh and provides new challenge for the player, and in the case of Enter the Gungeon, it's even used to hide enemies and secrets that you can find the more you play. Love the show!

[@georgelmfao]

Is anybody else disappointed that we didn't hear the word "bullocks" once?

[@Mas-cm6cy]

YES

[@CWargh63]

I heard it in my brain when I read your sentence...Sandra Bullock's last name in the plural or possessive I do believe.

[@Bluesine_R]

*Bollocks

[@Magmafrost13]

9:01 Did Yahtzee seriously just use exactly the same excuse publishers use when justifying abusive crunch practices? Jim Sterling would be spinning in his --grave-- office chair

[@goldenphoenix527]

"What if we humans aren't the ones writing the story?" Then we get Botnik Studios and stuff like "Harry Potter and the Portrait of What Looked Like a Large Pile of Ash" happens. A hilarious gimmick the first time you see it, but horribly unintelligible.

[@RobertBarton86]

"Well they can do that by looking at the clouds. Or taking LSD...." "Two fine activities!" Fantastic!

[@mystbunnygaming1449]

There's a place for both. Take the popularity of randomizers lately. Link to the Past is an amazing game, but start shuffling items around, and suddenly it's like re-discovering how awesome the game truly can be. I've recently tried a Mario RPG randomizer as well, and I haven't sunk so many hours in a single day into this game since I played it for the very first time.

[@FamousWolfe]

I think it depends on the game itself. Personally, I love all the procedural generated levels in XCOM 2 as it made each mission feel unique and awesome (in comparison to XCOM EU where it felt like you were playing the same 12-15 maps on repeat).

[@ianfraser2572]

XCOM 2 was the first thing I thought of as well, randomized levels are easily one if the biggest improvements. Some of my favourite mods are ones that add extra map packets to create even more level variety.

[@sophiathekitty]

I'd argue that using both in the right balance for the game you're making. Bethesda worlds are hand sculpted but still have the samey feel of procedural generation. Algorithmically generating additional assets could reduce the "this is the same locker and same desk everywhere" And additionally in the fallout games the random encounters and the emergent gameplay they create makes the games better. And as a sometimes hobby game dev myself having procedural generation is how I make up for my inability to socialize with others to do collaborative projects. One idea I've had but not really made much progress on is building info simulations... Where you can define the locations in a world, and define different types of info about the world. And then simulate how info could spread through the world.... And have it generate all the dialog trees.... Which you could then polish and refine.

[@romanoforsale]

Probably the most dynamic debate up until this point.

[@StudioAREshorts]

Just FYI real quick Scripted generation is like you buy one of every Lego City set and arrange them so that they form a good looking, functional, thematic Lego City. Procedural generation is having Lego corporation ship a bunch of completely random sets to you so you can get the "Heartlake City" set as well as "Lego Power Miners" side by side.

[@alvaroherrero498]

I would say that David Lynch doesn’t simply do away to structure in his work, but rather, he carefully subverts the structure of conventional storytelling. To do that sort of thing you must have great understanding of said storytelling in the first place.

[@modelcitizen1977]

I say there's room for both to excel in different genres. Take a game like Combat! for the Atari 2600. There's no story being told there, it's just pure gameplay. Having a procedurally generate maps on the fly that you and your opponent have to develop strategies for off the cuff would be incredible. Trying to procedurally generate The Last of Us would obviously be a shit show.

[@sethstringer1670]

I believe that from a storytelling point of view, scripted environments are much better at engaging the player, but from a gameplay point of view, it varies. For example, a game like mario, randomly generating the positions of all the blocks and enemies would be shite, as seen with the lower quality mario maker levels that crop up, however, depending on the type of game you're playing, random generation can work. The binding of Isaac is a good example, but a better one in my opinion would be cadence of hyrule. The replayability on that game is amazing since it's entirely different each time through, and while what you're attempting to achieve is the same, the difference in the way you go about it makes it tons of fun. My favourite example though would be steamworld heist, where each mission is unique, but there are random variations of those missions each time you play them. It works very well for a tactics game because it keeps you on your toes, forcing you to pick just the right load out to tackle the mission, in which you won't know entirely what's coming, even if you've already played it

[@alebanj]

But.. I love binding of issac

[@TheOneGreat]

Love this series. It's realky thought-provoking. Keep it coming.

[@demigodace]

Can't help but feel like jack missed the easiest point- re-play-ability. I'm Against procedural generation on a wider scale but it does offer the potential for something different in the experience every time you play it. How many great well written games have we played that we have never gone back to? I really enjoyed playing through Disco Elysium recently but even though it has great writing and different playthroughs can offer different experiences i still haven't found the desire to go back because i don't think the major beats would change. Compare that to something like slay the spire that i've played through literally hundreds of runs. Different experiences to be sure but i got to spend maybe 40 hours in disco elysium and i've played slay the spire for nearly 200 hours. Some people can go back and play through scripted games many times but i think procedural generation does have some benefits, only really useful for certain games and types of experiences I guess.

[@ahmednasser8214]

Procedural game design: Gameplay in the driver's seat, replayability riding shotgun and picking the music, and story in the back seat. Scripted game design: May have story driving and gameplay riding shotgun the whole way. It may have then switch driving and riding shotgun. And may or may not take replayability with them.

[@TheNeXTGUI]

I played Mystery Dungeons Explorers of Time and the dungeons always felt predictable and routine specifically because of the Procedural Generation.

[@gavinislightning]

I feel like it's less of interpreting procedurally generated games, but instilling something that wasn't there in the first place. Like when someone puts a blank canvas in a museum and the art snobs go "what does it mean?". When it didn't mean anything.

[@silverwings9276]

It can work but only under specific circumstances, a la Isaac, where the repetitive nature of auto-generated rooms is mitigated by an extremely high variance in terms of gameplay, which makes going through similar rooms acceptable and maybe better even; auto-procedural generated elements will work only when the gameplay foundation can actually take advantage of it. It's a difficult balance to strike, since even good games like Dead Cells falls into the repetitive pit, Isaac has much more depth and detail that people realize, which is why it is so loved.

[@dragoniraflameblade]

I think there's a way to have the machine and artist work together. Take those book bots - feed all your work into it that you've done and polish up the output. It's not perfect, but I'd be interested in seeing if it's a new form of ghostwriting. Same with games.

[@cameranfrancis202]

Deep Rock Galactic is a perfect example of how great procedural generation can be. A large part of that game is about the exploration and the excitement of never knowing what awaits you at the end of a cave system. It simulates the random and natural formation of environments which adds to the tension. Sometimes games that are designed by people can feel too hand holdy and actually far more fake because you are having things forced into you.

[@zubiddydoodoopop]

Replayability alone is reason enough to warrant procedural generation. There's a reason Minecraft is one of the biggest IPs gaming will ever see. It's core is random seed generation where the adventure you experience is different every time opening up a seemingly infinite amount of content from a finite game with a finite file size. This creates a level of continuous player involvement even the greatest of fixed experiences couldn't even hope to achieve. Singleplayer games do give off higher highs when executed properly but those hits get weaker & weaker with each proceeding playthrough. The Half Life series will always be one of my favorite experiences that I'll cherish for the rest of my life but one thing I know for certain is that I'll never experience that euphoria, the fear, the confusion, or even the fun I once had when I first played those games half a lifetime ago. However when I go back to play games with these RnG elements more or less consuming the essence of the game some of that intrigue still remains meaning that I'll end up going back to play it despite the fact that I don't hold it in as high regard as some of my favorite singleplayer games. In fact Yahtzee, if I do remember correctly, your critique of either Portal 1 or 2 was the replayability (in terms of showing it off to new people) where that created tension between the experienced player and the newcomer where the experienced one was battling between just taking the controller and showing them the solution or goading them on in frustration to try new approaches. You know the solution because it's scripted and will never change. However if we flip over to Minecraft again and show this same new person how to go about surviving there's a level of novelty still there that won't inherently be fully laid out for you in how to pass on your knowledge to the next generation without activating creative mode giving you full access to the utilities thus removing that procedural element from the game. In fact I'll go as far as to say that's actually more natural than a fixed experience because the real world ebbs and flows in away that can't be entirely fixated into a singular corridor like experience without dramatically watering it down until you're back to a book like experience whereas on the opposite end of the spectrum you have complete chaos. The ideal is somewhere in the middle, balance is key. That's ultimately what videogames are, a balance of the all the arts that came before it both scripted & procedural, curation & creation, order & chaos. Posing one as unnecessary to the other would be like plucking out your non-dominant eye because it's not as useful to you as the other where you end up losing a significant element of perspective as a result (depth perception). To dive even deeper down this rabbit hole, gaming wouldn't even exist today without both of these elements existing in symbiosis. Let's look at the technical limitations of system's past, the old arcade cabinets, Ataris and cartridge based Nintendos. The depth of the games people could make at the time were severely limited in both storage and processing power. The best way to get around this (if your ultimate goal is to create a game with the greatest player time investment) is to have a simple established goal (high score/survive/etc.) with a bunch of RnG thrown in to ensure the experience won't be repeated the longer you end up playing. A prime example is Tetris. A simplistic game with a defined goal with randomly delivered pieces for you to dispatch or you'll game over. Without both existing you don't have a game. AI can be boiled down to the same terms since it's a set of pre defined parameters responding to the player's (essentially random) input meaning even within the tight definitions of that program's code the exact sequence of actions won't necessarily play out even if the player were to precisely retrace their steps in a given circumstance. Modern games operate under the same principles but with higher limits of what can be done meaning Tetris, Half Life, Minecraft, and everything else are all products of scripting and randomness. And if you REALLY want to get philosophical about things, humans, too, are both scripted and random where we have agency in how we deal with the hands we're dealt (DNA, genetic inheritance/dispositions/etc). So in conclusion you're both wrong, I win. *half hearted taunting* Procedural generation is a natural evolution and integral component of the gaming medium, a byproduct or circumventing limitations, and a tool for developers to drastically expand the scope of their creation that can be used to favorably compliment their scripted elements

[@TamTroll]

i think random generation is good in some certain cases. For example FTL: Faster then Light uses procedural generation to map out jump points and the events therin. Having pre-made maps instead would make things too predictable and boring. half the fun is the surprise and forced adaptation to whatever you're given. your plan to use fire to wipe out enemy attackers might be pushed aside when you're given a teleportation bay for example. Minecraft is also a good example, as one of the main draws is the random generation creating huge worlds you can explore, practically none of which are at all alike.

[@wyattnolte]

Wow, Yahtzee coming all out for development crunch. He's confusing the architect for the builder and telling him to find another job if the construction site is too dangerous. Calling Jim Sterling.

[@steveschenker1380]

It was great seeing Jack on that Zoom reunion of the Community cast!

[@sarahreynolds1870]

X-Com is one of my favorite games because it uses procedurally-generated elements to make the game more replayable but puts them within a scripted/linear structure and difficulty curve that doesn't make the randomness feel like it's overwhleming the narrative.

[@Mernom]

Argument against 'random games can be easier to make than scripted games' - Making a good random game can be much harder than simply doing it yourself.

[@The_Hell_Lord]

Some games aren't meant to be enjoyed as a work of art but rather a fun experience akin to a sport or theme park ride (a casino is also a good example but that doesn't help my case as much) and there are only two ways of achieving such an experience: multiplayer and procedural generation (I've gathered that Yahtzee isn't a massive fan of either). Combining the two and balancing the extent to which they influence the experience can also work quite well if done right.

[@madskull13]

What feels better depends on what side of a game is more engaging. If the story is more emotional and engaging, then engaging a more human element is the way to go. However, if it's the gameplay loop that's more engaging then procedural generation can deliver far more experience through that loop than a limited amount of hand crafted levels.

[@rhaythe]

Two procedural games that probably deserved more attention than they got were Cryptark and Duskers. Both, I thought, worked quite well with the playground they set up for themselves.

[@acrobaticalpaca6675]

Hey Yahts, you should toats play noita. Its like a procedurally generated, 2d, side on, pixels with enjoyable phisics, dungeon crawler where you play as a witch that can craft wands that interact with the environment in interesting ways. You can really get the binding of isaac thing where some runs get laughably op but its as much about combining the stuff you find in clever ways to fuck with the physics as it is about finding things that are powerful on their own.

[@FamousWolfe]

Noita looks interesting but until it leaves early access that's a no-no for me. Nothing personal, I just don't buy early access or pre-order games anymore.

[@CDL_Gaming]

I think this argument depends heavily on what the player is wanting from the game. If the Player wants to play a game and continue playing it because they find it to be fun, procedural generation is a great way to expand gameplay. If the Player wants something story-oriented, then something more scripted and constructed is a great way to accomplish that. It could come down to the Player wanting to have their adventures be the story, or being a told a story through the eyes of an avatar. For instance, if you're looking for a sci-fi adventure you could go the route of procedural generation and play No Man's Sky, where everything is procedurally generated and no planet is ever truly the same thus the story of the game is your adventure and experiences therein. Alternatively you could go the route of Mass Effect and have a scripted experience (I know there's a good deal of player choice but it's all still part of a bigger arc that ends up in the same place) where the story takes front and center stage with you're actions moving the story along. The downsides to both of these examples being that procedural generation does get a bit predictable if the developer didn't put enough variety of assets into the game, with NMS specifically although there is theoretically an infinite number of randomly generated worlds/creatures/plants/etc desert planets all feel the same and ice planets all feel the same yadda yadda and there's not much emotional attachment after awhile, and a scripted experience, while it can be incredibly moving, is not really your experience, with Mass Effect you are following Commander Shepherd's journey and just choosing what he does along the way with everyone having played the game having roughly the same experiences. (Sorry for the run-on sentence.)

[@michaelmccarty1327]

It's not the same kind of thing as what they're arguing about, but there is some procedural generation in RTS games like Age of Empires. There are some universal constants in each game, for example, there will be gold, food, wood, etc near your home base, but the exact position of those items is left to the dice. It adds to the replay-ability of the same map.

[@nick1wasd]

I feel like procedural tells stories that just... happen, while scripting tells stories that happen because of reasons. Oxygen not Included, Rimworld, and Civilization all tell good stories, but the meaning is made by the player, reacting to what madness goes on in the world. The author, in a sense, is the player themselves, because by them reacting to what happens, what the computer does in kind changes, and so they shape the world, slowly but surely. I feel like giving people that power can be good, but not many people enjoy being given it, that's why it's such a niche genre of games compared to everything else

[@AutumnReel4444]

Oh man, I know it isn't indicative of their beliefs, but literally everything Yahtzee said I disagreed with other than the points they agreed on. Sheesh.

[@thisisacallforhelp3484]

In my opinion, one of the greatest strengths of procedural generation is unique challenges. While a linear game asks you to go through a linear series of events, which can be seen multiple times and tend to remain the same, procedurally generated challenges challenge a different kind of skills Mastery of the mechanics over memorization of a few select moments. While a run of Dark Souls tends to be the same, and therefore is the same and allows you to use the experience you already have, a run of Enter the Gungeon asks for pure mastery of the game’s mechanics, and thus I think is a more unique and different kind of challenge. That’s not to say linear games cannot test this at all, but procedural games tend to be better at it because it isn’t a set series of scenarios.

[@skepticalbadger1695]

They're rare but like Dead Cells, sometimes it works brilliantly. Most of the time, it falls short.

[@TheSUGA1202]

There is this game called Hades that is the middle ground between procedural and scripting

[@joeyjojojrshabadoo7462]

I'd like to see procedurally generated maps in Call Of Duty so it's not just a matter of memorizing the best camping stops.

[@Kilmoran]

I know the point of this is to have a debate, meaning that they argue for a side whether they agree with it or not, but the truth is both are simultaneously useful and honestly can't be measured as better or not compared to each other because they essentially perform different things. They are not replacements or appropriate stand ins for each other (as is seen clearly when it goes bad).

[@grip7777]

The podcast is always the best part of this! Nice series overall.

[@dedmnwlkn5796]

0:40 the epitome of the dungeons and dragons term "meet or exceed"

[@MetalheadSean666]

I feel like these video prove that all these controversys and argument are ultimately silly and pointless and I love it

[@nosdregamon]

YT's algorithm, brought me here. I guess I don't like this format of a fake argument. I'd rather see some possible solutions to the various problems (various degrees pf replay-ability, lack of structure and emotional impact), or chances (say handcrafted stories inside procedural generated environments). I also think someone missed a point, painting procedural generation as being devoid of human input. At least at his point in time, procedural generation is still 100% human input, just accentuated & multiplied by some variables in a (again human crafted) framework.

[@fortunatesoul12]

The podcast after is fucking great, I urge everyone to hear it

[@TheCreepypro]

another great debate with a fundamentally flawed question cause at the end of the day it is entirely up to each individual to decide which they prefer then again I'm sure they discuss that in the podcast

[@emilybroderick2421]

Good debate! I think that procedural generation has the potential to be as good as scripting, but that might have to wait until we progress programming technology a bit more. The No Man's Sky model is flawed, but it could point the way toward making procedural generation into something consistently engaging. Or else, we might need to get creative about somehow incorporating a structured or mostly structured story into a randomized world.

[@KuraIthys]

What about... Procedurally assisted design? ;p We design bits and pieces of the world using a procedural system, but rather than bake that procedural system into the end product explicitly, we get someone to use the procedural tools to make things quicker. They looked through a bunch of procedural content and chose that which they liked best, then tweaked it in whatever places it didn't quite seem to make sense. Possibly combined the output of multiple procedurally generated sections... It's... Guided, somewhat planned design, but with the benefit of being able to generate large areas quickly as a starting point...

[@JonathanRodd]

I love how genuinely interested Jack seemed about the development of Yahtzee's bald spot :P

[@brycejohansen7114]

I think Jack is right on this... in the end, they are both opposite sides of the same coin. They're tools meant to enhance different things in a game. For example, Stardew Valley's dungeons are meant to be procedural because you are not meant to know where you are going, it's the big unknown while the town is a static environment because you need to know where each shop or house is.

[@tombonner2799]

You can tell when an argument is statement is poorly supported when the only person constructing the opposition is the person making the argument.

[@jaysonpreston31]

OMG, this is good (first time) I might head along to the podcast that comes after there should be a link in the description it's available on spotify soundcloud and all the usual places yes just search for thee escapist

[@EtCeteraTape]

There's a lot that can be said about not trying to fully generate your game procedurally (which will end up nowhere, of course, games are a work of art and should be hand-crafted as much as possible, that's what makes them feel alive), but using the systems sparingly, where they actually allow you to lower the amount of monkey job necessary and add variety and replayability to otherwise repetitive sequences, be it cave level design, dynamic battle systems or speech generation. Over-relying on automation leads to repetitivenes, yes, as different seeds in the same algo don't really produce too distinct results, but there's enough tasks that simply couldn't have been accomplished otherwise without some form of generation (take any space sim world space). And if the algo is complex enough - isn't the actual Universe with all its stories and complexity a result of such a process?..

[@LuciferonMinecraft]

I honestly think that the issue here is not whether procederal generation should be incorporated into games, which it obviously should, but whether games designed around procedural generation can be as good. Like, you might utilize automation to generate things for the game, such as a distribution of trees or forests, which you can then take and work with. That is utilizing procedural generation as a tool. But on the other hand, If you do everything via generation and leave it up to the player, you get something like Dwarf Fortress. And credit where its due, that game can have interesting moments, but they are all player driven moments. Similar with minecraft. The layout of any minecraft world is procedural, which lends itself to huge amounts of replayability, but nobody has ever accused minecrafts story elements such as they are of being the reason people keep playing it. Again, its all player driven. Skyrim, which has extensive story telling, could never have been made with randomization like minecraft or DF. Because at they end of the day, they are trying to do different things. Yahtsee has stated on numerous occasions, and here as well, that he regards video games as a medium of storytelling, and that the distinction he makes is between games which make you feel, and games which make you numb. It shouldn't be surprising that many, if not all of the games that he praises go out of their way to tell a good story, rather than the player having to make it up. Think Portal, Subnautica, Dark Souls. All games which could have been procedurally generated, but weren't, which could have chosen to forgo telling a story and make the player come up with a story, but didn't. And they were huge successes because of it. I suspect that Yahtsee would never change his mind, because his view of what makes a good game, or a game good at being a video game, is one which tells you the story and generates emotional moments. And while he has nothing against procedural games inherently, the very quality of being randomly generates works against a game and diminishes its ability to tell a story, especially if it is a game trying to make you numb.

[@m-2718]

Lets face it we are here just to understand the podcast

[@mattobrien725]

Yahtzee and Jack would make a good comedy duo