The problem with randomized content is that it gets to feel generic after a while. When your brain has said "wait a second, I've seen this before" enough times it will get bored. Once you've broken down the algorithm or how the puzzle pieces fit together, it starts to feel pretty mundane.
If there is a lot of complexity and variety to randomized content, it might work. I don't know though.
Enter a whole new realm of challenge and adventure.
Originally posted by nilden Oh you mean like missions in Anarchy Online?
Or LDoN from EQ1? Wasnt completely random of course, but definitely had some major random elements.
There's an enormous difference between seeing somewhat random content, and seeing far more random content that is actually done well. We haven't seen the latter.
And what makes you think that randomly generated content would have to be instanced? If you can randomly generate a dungeon instance, then you can randomly generate an outdoor region. And if you can randomly generate one outdoor region, then you can randomly generate a million outdoor regions. And then you can have a planet-sized planet for your game world. But you'd better have fast travel or players will get sick of week-long travel times.
There are some things you'd lose by going this route, such as quest text and voiceovers, as you really can't randomly generate those.
Why hasn't it been done? Because it's hard. You'd need a programmer who was really good at probability to have a chance of pulling it off. I have the impression that while some game programmers might be pretty good at the relatively introductory mathematics that a comp sci major might see (linear algebra, multivariable calculus, etc.), they tend to be completely clueless about the more advanced mathematics that you'd need in order to do a randomly generated world properly.
Even if you did have a programmer capable of doing the probability work, you'd still need to get management to green-light the project. There are a handful of MMORPG companies like FunCom or ArenaNet that are willing to take substantial risks, but most would be too scared to try.
And really, if they can't do much easier stuff with probability like randomly generated trees, there's no sense in trying something harder.
Originally posted by nilden Oh you mean like missions in Anarchy Online?
Or LDoN from EQ1? Wasnt completely random of course, but definitely had some major random elements.
There's an enormous difference between seeing somewhat random content, and seeing far more random content that is actually done well. We haven't seen the latter.
And what makes you think that randomly generated content would have to be instanced? If you can randomly generate a dungeon instance, then you can randomly generate an outdoor region. And if you can randomly generate one outdoor region, then you can randomly generate a million outdoor regions. And then you can have a planet-sized planet for your game world. But you'd better have fast travel or players will get sick of week-long travel times.
There are some things you'd lose by going this route, such as quest text and voiceovers, as you really can't randomly generate those.
Why hasn't it been done? Because it's hard. You'd need a programmer who was really good at probability to have a chance of pulling it off. I have the impression that while some game programmers might be pretty good at the relatively introductory mathematics that a comp sci major might see (linear algebra, multivariable calculus, etc.), they tend to be completely clueless about the more advanced mathematics that you'd need in order to do a randomly generated world properly.
Even if you did have a programmer capable of doing the probability work, you'd still need to get management to green-light the project. There are a handful of MMORPG companies like FunCom or ArenaNet that are willing to take substantial risks, but most would be too scared to try.
And really, if they can't do much easier stuff with probability like randomly generated trees, there's no sense in trying something harder.
I actually wrote a program to procedurally write random quests/plots then randomly generating a dungeon from it. It's doable but still looks/feels random AND more importantly, it wasn't good enough. You need real people behind it.
Kyleran: "Now there's the real trick, learning to accept and enjoy a game for what
it offers rather than pass on what might be a great playing experience
because it lacks a few features you prefer."
John Henry Newman: "A man would do nothing if he waited until he could do it so well that no one could find fault."
FreddyNoNose: "A good game needs no defense; a bad game has no defense." "Easily digested content is just as easily forgotten."
LacedOpium: "So the question that begs to be asked is, if you are not interested in
the game mechanics that define the MMORPG genre, then why are you
playing an MMORPG?"
I play a lot of ARPG, games like torchlight, Diablo2 and Path of Exile. These games don't have much content. Maybe 20-30 hours. Thing is the content they have is randomized.
Map layouts, mob type, placement and abilities, chest placement, boss placement all different everytime you play. This keeps the games replayability really high.
Early CoH.
Self-pity imprisons us in the walls of our own self-absorption. The whole world shrinks down to the size of our problem, and the more we dwell on it, the smaller we are and the larger the problem seems to grow.
I play a lot of ARPG, games like torchlight, Diablo2 and Path of Exile. These games don't have much content. Maybe 20-30 hours. Thing is the content they have is randomized.
Map layouts, mob type, placement and abilities, chest placement, boss placement all different everytime you play. This keeps the games replayability really high.
When I come to a MMORPG forum I read that people are devouring content. Content that's damn near ten times of what ARPGs offer. Thing is all of that content is static. It never changes, you run through it and then thats it.
What if content was randomized? Randomized questing, randomized dungeon layouts and mob placement? Maybe people would play MMOs more without becoming bored and complaining about a lack of content?
Would something like that work in a MMO? Would you play a MMO that offered that type of content?
Basically when it comes to dungeons loot and bosses inside the dungeons everything SHOULD be fuly randomised every time u enter one. That way even if u play the same dungeon many times u can never guess what to expect .Much more intresting that way. Same goes for da loot.
Im curious how it is possible big dev companies so far didnt do that yet.Only old diablo series and poe now.Wierd..
I keep getting this one confused with Eye of the Beholder (they used the same engine didn't they?) This was an example of what I call grey goo randomness ... although it was random levels with two random monster species per level, the randomness didn't really add much to the experience of play.
I found this did not compare well to the ascii roguelike games it was attempting to ape (eg: angband, nethack, ancient domains of mystery).
Randomized content can definitely work, and if done well can add a lot of replayability to boot, as well. The general world probably would be highly problematic to randomize in terms of layout, but other things such as quests and monsters could vary over time.
Instanced content (dungeons) would be absolutely perfect for randomized layouts as well as opposition. Particularly in a game like WoW where you're running the same dungeons loads and loads of times, such a feature could be a tremendous benefit.
I keep getting this one confused with Eye of the Beholder (they used the same engine didn't they?) This was an example of what I call grey goo randomness ... although it was random levels with two random monster species per level, the randomness didn't really add much to the experience of play.
I found this did not compare well to the ascii roguelike games it was attempting to ape (eg: angband, nethack, ancient domains of mystery).
It was 1993. There were definite technological limitations at the time. One can see the improvements in 2001 with AO and in 2004 with CoH - although they lost the -no pun intended- depth of DH.
I'm not sure if you partook in the beta for TSW or played it; but I've wondered what it would be like to have DH remade with that level of tech - the environment, lighting, sound, etc. Not the TSW world, etc, but something that offered those kind of graphics in a dungeon dwelving kind of game.
It would be a tough one in regard to what folks typically think of a MMORPG. I mean, why would the dungeon always be different? As a CORPG though, I picture being able to put together a party of 4-6 and headig off...
...for me, it would beat having done a dungeon so many times with the same map that after a wipe - it would not be a case of being able to run back to my body without having to watch the screen because that map's permanently imprinted in my brain...
I just picture the not knowing where you're going - the actual exploration factor - the random encounters, not always knowing what's around the corner - etc, etc, etc.
It's not a task then, I'm not doing something akin to tying my shoes.
It works for a more modern game, in a sense, because the layout potentials can vary that much. But that gets into door missions, and while some folks love to run the same dungeon over and over - for some reason, they hate door missions (even though their dungeon's basically a door - where the layout and mobs are the same each time).
Even beyond the layout of the map itself though - there's also the mobs. How many times are folks going to go into the same dungeon to kill the same guy so they can get the same weapon? I hate named weapons in that sense. It always leaves me wondering if the boss sends an assistant to Costco or the like to pick them up in bulk, eh? Even if the weapons just sported generic names - it would be less troublesome. Or if the bosses had random names and the weapons reflected that...
...but that all gets away from the nature of the grind of dungeon running and raiding. It's more like Black Friday shopping than it is any sort of adventure.
It's funny though, just watched a video of DH over on YouTube - so hard to believe how many hours I put into that game...lol, egads!
I miss the MMORPG genre. Will a developer ever make one again?
Kyleran: "Now there's the real trick, learning to accept and enjoy a game for what
it offers rather than pass on what might be a great playing experience
because it lacks a few features you prefer."
John Henry Newman: "A man would do nothing if he waited until he could do it so well that no one could find fault."
FreddyNoNose: "A good game needs no defense; a bad game has no defense." "Easily digested content is just as easily forgotten."
LacedOpium: "So the question that begs to be asked is, if you are not interested in
the game mechanics that define the MMORPG genre, then why are you
playing an MMORPG?"
I am thinking that this idea is not far around the corner. The world changes everyday and in video games we strive to make our digital lives as close to our real lives as we can. I wouldn't brush this idea away, it could become larger than we think.
Don't think of randomly generated maps only in terms of having a bunch of custom done rooms, picking some of them at random, and then connecting them. That mild bit of randomness is far less than it's possible to do.
Computers aren't really any more capable of generating random content today than they were a decade ago. They can do the same thing faster today, but that doesn't matter, as it would have been a fraction of a second a decade ago, anyway. What's missing is a combination of game programmers with the proper competence in probability and management willing to let them use it. It might not even be both of them missing today, but you can't tell if you have one until you have both.
Randomized traps and the free-movement ability to respond to them are something that I think should be looked into. For example, DDO's traps are very nice in the 'early game'. Having a rogue who can disable them is very very useful, and in some places almost required (Gwylan's Stand comes to mind, or the hallway in VoN 5). But eventually you can overpower them with levels, hit points, and so on. Also, they're static. Imagine having the -entire pool- of traps available in the entire game randomly determined for each and every room as you enter it. So you could (very very rarely) come across a -real- death-trap of a room...or more commonly, there will be no traps at all. No way to memorize their location or even what they'll do or look like. This type of gameplay makes the 'mechanic rogue' archetype or 'engineer' archetype incredibly valuable without it just being 'another melee' or 'another ranged' and offers an alternate choice of playstyle for a game player. Viewing each room on a radial or spherical scale could assist in the placement of these random traps (for example: the RNG randomly determines that it is a spike trap, and then radially determines it's placement as 90 degrees +-0 Z ( chest level). Hence, the right wall has a spike trap that will lash out at anyone who walks too closely).
I don't think such a game would ever have 'mainstream' appeal, though, and that may be the biggest reason we never see one (provided, of course, that we don't).
Imagine having the -entire pool- of traps available in the entire game randomly determined for each and every room as you enter it. So you could (very very rarely) come across a -real- death-trap of a room...or more commonly, there will be no traps at all.
This is a pretty huge problem with games that go beyond the event horizon of excessive randomness.
One of the ways that games appeal to gamers is that they offer a sort of enforced fairness. Every obstacle that you're given can be beaten somehow. There are no obstacles that can't be overcome unless the challenge is to figure out that you can't overcome it and you need to find another way. You're never going to get to World 8-2 of Super Mario Bros and find a huge wall with a large pit in front of it, making it impossible for you to ever get to the end of the level and the end of the game. It's never going to happen.
That's one of the ways that video games comfort us where real life can't. Later today, a bridge you're driving across might collapse and there will be no action you can take to save yourself. Tonight, you might wake up in a burning building and there will be no path you can take that will get you out alive. Life is unfair like that, and games aren't.
Even when people talk about really really difficult games or difficult sections of games, they're always talking about something that can be beaten. A section where nothing you do can possibly get you past the obstacle isn't hard, it's stupid. (Again, not counting obstacles that you overcome by making sure you don't encounter them.)
If a game kills you in a situation where there's no action you could have possibly taken to survive, it's generally recognized as a very bad game. A room that randomly ends up with a certain-death combination of traps is one example. Another example would be a monster that does a random attack every 4 seconds, and one of the attacks is a highly-damaging 4 second AOE stun. If there truly is no way to avoid or cancel the stun, there's a small chance that the monster will use the stun a few times in a row in wipe the whole group.
So one of the dangers of randomizing a world, dungeon, or encounter in too many ways is that you lose control over whether it is fair or not. And gamers go into games with the understanding that there is fairness on some level... that the game will not present them with an unwinnable challenge unless it's a challenge that is unwinnable by design for a reason.
That's not a problem of too much randomness. That's a problem of badly-designed randomness.
Everything in every game, and more broadly in the entire universe, is random. It's just a matter of, random with what probability distribution. This outcome with probability 1 and all others with probability 0 (what most games try to shoot for in most cases) is a perfectly legitimate probability distribution. So is a uniform random variable over some range, or n different outcomes each occurring with probability 1/n. Or a poisson random variable, which is what Lissyl probably had in mind, whether or not he knows that's what it's called. But for a lot of game purposes, the simple random variables that you'll see in an undergraduate probability course are simply the wrong ones to use.
You want maps, mob spawns, and so forth to be taken from the set that will provide an appropriate challenge to the player--neither unduly easy nor unduly hard. Or perhaps rather, you want to randomly choose from a small subset of this, if only because trying to choose from the entire set of fair challenges is too hard. But you want the subset of randomly generated content to be as large as possible while remaining a subset of the set of fair challenges. If the subset is too small, then it becomes too predictable.
You want maps, mob spawns, and so forth to be taken from the set that will provide an appropriate challenge to the player--neither unduly easy nor unduly hard. Or perhaps rather, you want to randomly choose from a small subset of this, if only because trying to choose from the entire set of fair challenges is too hard. But you want the subset of randomly generated content to be as large as possible while remaining a subset of the set of fair challenges. If the subset is too small, then it becomes too predictable.
Pretty much this - and this is what I think is more readily available with today's technology to provide something that would work. They would be able to work with larger sets of data to pull from - they would be able to render and present that data in 3D form, etc, etc, etc.
Layouts, rooms, scenery, lighting, sound, etc, etc, etc. A 4-6 player (or more) environment where you have everything generated so that it looks as good as something that was actually designed....
I miss the MMORPG genre. Will a developer ever make one again?
Computers aren't really any more capable of generating random content today than they were a decade ago. They can do the same thing faster today, but that doesn't matter, as it would have been a fraction of a second a decade ago, anyway. What's missing is a combination of game programmers with the proper competence in probability and management willing to let them use it. It might not even be both of them missing today, but you can't tell if you have one until you have both.
For trivial things like gear, stats, textures and basic maps where you're basically selecting from a finite set of options then yes, it's computationally trival. But for more complex things like entire levels or areas and depending on the techniques the process can be expensive. Many generation algorithms are based on trial and error. For example you end up wtth a dungeon that just has one room and a bunch of long halls - this is likely boring and not practical for effectively placing monsters so you throw it out and try again. Maybe you've built the level and there's not enough interesting treasure or the random environmental obstacles have blocked the only exit - throw it out and try again. Not only do you have to generate the level, but you also have to validate that it's meets certain qualities. Often the approaches that have the highest randomness are the most complex so you end up with less random approaches like creating 60 dungeon sections that fit together in a limited number of ways.
I think 3D is more challenging for content generation as well. You add pathing issues, size issues (dragon doesn't fit in the hall), players getting stuck or falling through cracks. MMORPGs also introduce more challenges as it's one thing to tie up a desktop's CPU/RAM/disk for a few seconds and another to put that load on one of your server resources. Then once you've generated the content you need to get it to the players that are trying to access it.
The big question is can you generate interesting content as opposed to a bunch of generic levels? I think the answer is hand crafted content will always be vastly superior and you apply randomness where it's easy or adds value.
The last approach I used was to generate a bunch of random levels and cherry pick the interesting ones to be improved. It saved me some time and added diversity.
Randomised content is used, but in a limited way. It could be a partial answer to end game problems but as yet has not helped much. Partly because the content does not really alter that much.
Comments
The problem with randomized content is that it gets to feel generic after a while. When your brain has said "wait a second, I've seen this before" enough times it will get bored. Once you've broken down the algorithm or how the puzzle pieces fit together, it starts to feel pretty mundane.
If there is a lot of complexity and variety to randomized content, it might work. I don't know though.
Enter a whole new realm of challenge and adventure.
There's an enormous difference between seeing somewhat random content, and seeing far more random content that is actually done well. We haven't seen the latter.
And what makes you think that randomly generated content would have to be instanced? If you can randomly generate a dungeon instance, then you can randomly generate an outdoor region. And if you can randomly generate one outdoor region, then you can randomly generate a million outdoor regions. And then you can have a planet-sized planet for your game world. But you'd better have fast travel or players will get sick of week-long travel times.
There are some things you'd lose by going this route, such as quest text and voiceovers, as you really can't randomly generate those.
Why hasn't it been done? Because it's hard. You'd need a programmer who was really good at probability to have a chance of pulling it off. I have the impression that while some game programmers might be pretty good at the relatively introductory mathematics that a comp sci major might see (linear algebra, multivariable calculus, etc.), they tend to be completely clueless about the more advanced mathematics that you'd need in order to do a randomly generated world properly.
Even if you did have a programmer capable of doing the probability work, you'd still need to get management to green-light the project. There are a handful of MMORPG companies like FunCom or ArenaNet that are willing to take substantial risks, but most would be too scared to try.
And really, if they can't do much easier stuff with probability like randomly generated trees, there's no sense in trying something harder.
I actually wrote a program to procedurally write random quests/plots then randomly generating a dungeon from it. It's doable but still looks/feels random AND more importantly, it wasn't good enough. You need real people behind it.
Epic Music: https://www.youtube.com/watch?v=vAigCvelkhQ&list=PLo9FRw1AkDuQLEz7Gvvaz3ideB2NpFtT1
https://archive.org/details/softwarelibrary_msdos?&sort=-downloads&page=1
Kyleran: "Now there's the real trick, learning to accept and enjoy a game for what it offers rather than pass on what might be a great playing experience because it lacks a few features you prefer."
John Henry Newman: "A man would do nothing if he waited until he could do it so well that no one could find fault."
FreddyNoNose: "A good game needs no defense; a bad game has no defense." "Easily digested content is just as easily forgotten."
LacedOpium: "So the question that begs to be asked is, if you are not interested in the game mechanics that define the MMORPG genre, then why are you playing an MMORPG?"
Self-pity imprisons us in the walls of our own self-absorption. The whole world shrinks down to the size of our problem, and the more we dwell on it, the smaller we are and the larger the problem seems to grow.
I miss the MMORPG genre. Will a developer ever make one again?
Explorer: 87%, Killer: 67%, Achiever: 27%, Socializer: 20%
Basically when it comes to dungeons loot and bosses inside the dungeons everything SHOULD be fuly randomised every time u enter one. That way even if u play the same dungeon many times u can never guess what to expect .Much more intresting that way. Same goes for da loot.
Im curious how it is possible big dev companies so far didnt do that yet.Only old diablo series and poe now.Wierd..
I keep getting this one confused with Eye of the Beholder (they used the same engine didn't they?) This was an example of what I call grey goo randomness ... although it was random levels with two random monster species per level, the randomness didn't really add much to the experience of play.
I found this did not compare well to the ascii roguelike games it was attempting to ape (eg: angband, nethack, ancient domains of mystery).
Randomized content can definitely work, and if done well can add a lot of replayability to boot, as well. The general world probably would be highly problematic to randomize in terms of layout, but other things such as quests and monsters could vary over time.
Instanced content (dungeons) would be absolutely perfect for randomized layouts as well as opposition. Particularly in a game like WoW where you're running the same dungeons loads and loads of times, such a feature could be a tremendous benefit.
It was 1993. There were definite technological limitations at the time. One can see the improvements in 2001 with AO and in 2004 with CoH - although they lost the -no pun intended- depth of DH.
I'm not sure if you partook in the beta for TSW or played it; but I've wondered what it would be like to have DH remade with that level of tech - the environment, lighting, sound, etc. Not the TSW world, etc, but something that offered those kind of graphics in a dungeon dwelving kind of game.
It would be a tough one in regard to what folks typically think of a MMORPG. I mean, why would the dungeon always be different? As a CORPG though, I picture being able to put together a party of 4-6 and headig off...
...for me, it would beat having done a dungeon so many times with the same map that after a wipe - it would not be a case of being able to run back to my body without having to watch the screen because that map's permanently imprinted in my brain...
I just picture the not knowing where you're going - the actual exploration factor - the random encounters, not always knowing what's around the corner - etc, etc, etc.
It's not a task then, I'm not doing something akin to tying my shoes.
It works for a more modern game, in a sense, because the layout potentials can vary that much. But that gets into door missions, and while some folks love to run the same dungeon over and over - for some reason, they hate door missions (even though their dungeon's basically a door - where the layout and mobs are the same each time).
Even beyond the layout of the map itself though - there's also the mobs. How many times are folks going to go into the same dungeon to kill the same guy so they can get the same weapon? I hate named weapons in that sense. It always leaves me wondering if the boss sends an assistant to Costco or the like to pick them up in bulk, eh? Even if the weapons just sported generic names - it would be less troublesome. Or if the bosses had random names and the weapons reflected that...
...but that all gets away from the nature of the grind of dungeon running and raiding. It's more like Black Friday shopping than it is any sort of adventure.
It's funny though, just watched a video of DH over on YouTube - so hard to believe how many hours I put into that game...lol, egads!
I miss the MMORPG genre. Will a developer ever make one again?
Explorer: 87%, Killer: 67%, Achiever: 27%, Socializer: 20%
Brings back memories. GOOD OLD SSI games!
Epic Music: https://www.youtube.com/watch?v=vAigCvelkhQ&list=PLo9FRw1AkDuQLEz7Gvvaz3ideB2NpFtT1
https://archive.org/details/softwarelibrary_msdos?&sort=-downloads&page=1
Kyleran: "Now there's the real trick, learning to accept and enjoy a game for what it offers rather than pass on what might be a great playing experience because it lacks a few features you prefer."
John Henry Newman: "A man would do nothing if he waited until he could do it so well that no one could find fault."
FreddyNoNose: "A good game needs no defense; a bad game has no defense." "Easily digested content is just as easily forgotten."
LacedOpium: "So the question that begs to be asked is, if you are not interested in the game mechanics that define the MMORPG genre, then why are you playing an MMORPG?"
Come check out what I have to say on my blog http://civilgamer.com
Also check out http;//agonasylum.com for Darkfall player trading and stories forums
Don't think of randomly generated maps only in terms of having a bunch of custom done rooms, picking some of them at random, and then connecting them. That mild bit of randomness is far less than it's possible to do.
Computers aren't really any more capable of generating random content today than they were a decade ago. They can do the same thing faster today, but that doesn't matter, as it would have been a fraction of a second a decade ago, anyway. What's missing is a combination of game programmers with the proper competence in probability and management willing to let them use it. It might not even be both of them missing today, but you can't tell if you have one until you have both.
Randomized traps and the free-movement ability to respond to them are something that I think should be looked into. For example, DDO's traps are very nice in the 'early game'. Having a rogue who can disable them is very very useful, and in some places almost required (Gwylan's Stand comes to mind, or the hallway in VoN 5). But eventually you can overpower them with levels, hit points, and so on. Also, they're static. Imagine having the -entire pool- of traps available in the entire game randomly determined for each and every room as you enter it. So you could (very very rarely) come across a -real- death-trap of a room...or more commonly, there will be no traps at all. No way to memorize their location or even what they'll do or look like. This type of gameplay makes the 'mechanic rogue' archetype or 'engineer' archetype incredibly valuable without it just being 'another melee' or 'another ranged' and offers an alternate choice of playstyle for a game player. Viewing each room on a radial or spherical scale could assist in the placement of these random traps (for example: the RNG randomly determines that it is a spike trap, and then radially determines it's placement as 90 degrees +-0 Z ( chest level). Hence, the right wall has a spike trap that will lash out at anyone who walks too closely).
I don't think such a game would ever have 'mainstream' appeal, though, and that may be the biggest reason we never see one (provided, of course, that we don't).
This is a pretty huge problem with games that go beyond the event horizon of excessive randomness.
One of the ways that games appeal to gamers is that they offer a sort of enforced fairness. Every obstacle that you're given can be beaten somehow. There are no obstacles that can't be overcome unless the challenge is to figure out that you can't overcome it and you need to find another way. You're never going to get to World 8-2 of Super Mario Bros and find a huge wall with a large pit in front of it, making it impossible for you to ever get to the end of the level and the end of the game. It's never going to happen.
That's one of the ways that video games comfort us where real life can't. Later today, a bridge you're driving across might collapse and there will be no action you can take to save yourself. Tonight, you might wake up in a burning building and there will be no path you can take that will get you out alive. Life is unfair like that, and games aren't.
Even when people talk about really really difficult games or difficult sections of games, they're always talking about something that can be beaten. A section where nothing you do can possibly get you past the obstacle isn't hard, it's stupid. (Again, not counting obstacles that you overcome by making sure you don't encounter them.)
If a game kills you in a situation where there's no action you could have possibly taken to survive, it's generally recognized as a very bad game. A room that randomly ends up with a certain-death combination of traps is one example. Another example would be a monster that does a random attack every 4 seconds, and one of the attacks is a highly-damaging 4 second AOE stun. If there truly is no way to avoid or cancel the stun, there's a small chance that the monster will use the stun a few times in a row in wipe the whole group.
So one of the dangers of randomizing a world, dungeon, or encounter in too many ways is that you lose control over whether it is fair or not. And gamers go into games with the understanding that there is fairness on some level... that the game will not present them with an unwinnable challenge unless it's a challenge that is unwinnable by design for a reason.
That's not a problem of too much randomness. That's a problem of badly-designed randomness.
Everything in every game, and more broadly in the entire universe, is random. It's just a matter of, random with what probability distribution. This outcome with probability 1 and all others with probability 0 (what most games try to shoot for in most cases) is a perfectly legitimate probability distribution. So is a uniform random variable over some range, or n different outcomes each occurring with probability 1/n. Or a poisson random variable, which is what Lissyl probably had in mind, whether or not he knows that's what it's called. But for a lot of game purposes, the simple random variables that you'll see in an undergraduate probability course are simply the wrong ones to use.
You want maps, mob spawns, and so forth to be taken from the set that will provide an appropriate challenge to the player--neither unduly easy nor unduly hard. Or perhaps rather, you want to randomly choose from a small subset of this, if only because trying to choose from the entire set of fair challenges is too hard. But you want the subset of randomly generated content to be as large as possible while remaining a subset of the set of fair challenges. If the subset is too small, then it becomes too predictable.
Pretty much this - and this is what I think is more readily available with today's technology to provide something that would work. They would be able to work with larger sets of data to pull from - they would be able to render and present that data in 3D form, etc, etc, etc.
Layouts, rooms, scenery, lighting, sound, etc, etc, etc. A 4-6 player (or more) environment where you have everything generated so that it looks as good as something that was actually designed....
I miss the MMORPG genre. Will a developer ever make one again?
Explorer: 87%, Killer: 67%, Achiever: 27%, Socializer: 20%
For trivial things like gear, stats, textures and basic maps where you're basically selecting from a finite set of options then yes, it's computationally trival. But for more complex things like entire levels or areas and depending on the techniques the process can be expensive. Many generation algorithms are based on trial and error. For example you end up wtth a dungeon that just has one room and a bunch of long halls - this is likely boring and not practical for effectively placing monsters so you throw it out and try again. Maybe you've built the level and there's not enough interesting treasure or the random environmental obstacles have blocked the only exit - throw it out and try again. Not only do you have to generate the level, but you also have to validate that it's meets certain qualities. Often the approaches that have the highest randomness are the most complex so you end up with less random approaches like creating 60 dungeon sections that fit together in a limited number of ways.
I think 3D is more challenging for content generation as well. You add pathing issues, size issues (dragon doesn't fit in the hall), players getting stuck or falling through cracks. MMORPGs also introduce more challenges as it's one thing to tie up a desktop's CPU/RAM/disk for a few seconds and another to put that load on one of your server resources. Then once you've generated the content you need to get it to the players that are trying to access it.
The big question is can you generate interesting content as opposed to a bunch of generic levels? I think the answer is hand crafted content will always be vastly superior and you apply randomness where it's easy or adds value.
The last approach I used was to generate a bunch of random levels and cherry pick the interesting ones to be improved. It saved me some time and added diversity.