Sign in with
Sign up | Sign in
Your question

Reasons for random world generation

Last response: in Video Games
Share
Anonymous
May 19, 2005 11:05:00 AM

Archived from groups: rec.games.roguelike.development (More info?)

>From some of the previous topics, I've been thinking... what's the
point to fully generating a random world? I see the point to randomly
generated dungeons - replayability. Not knowing where the exit and
items are - though I don't think that translates as well to a full
world-map.

It's something that the developer in me finds quite facinating and
appealing still, though I can really say -why-.

That question has two sides though. First, the personal side: I think
a lot of it is based on hopes, and on finding complex (and emergant?)
systems to be interesting. Perhaps hopes that a random world will be
created that's a surprise to myself, and something I find engaging.

Secondly, Lately I've been wondering about the actual game-play
additions that a full random world adds. With a typical roguelike,
such as angband/nethack/tome, I don't think I see much benifit. The
towns and world map are used as utilities, and somewhat like a vending
machine - finding it provides little enjoyment for myself. If they
were simply "randomly" scattered, it'd be even less interesting.

So. I'm left with wondering what point there is?
I come to the conclusion that the game needs to be designed
specifically for a fixed-duration play, and that the actual traveling
to different parts of the map, and finding different places, must be
important. These aspects being randomly generated must be part of the
gameplay to make it interesting.

Here are some ways this is doable I think:
* A plot/scenerio/quest is generated along with the world-map that
involves traveling (collect items, kill bad guy, make national peace,
etc)
* The different towns have some knowledge of surrouding areas, as far
as goals go ("Oh! To the west is an icy cave! You should bring
fire!")
* Character traveling ability grows as time goes - character gets
stronger, gets more gear, can afford special
equipment/guards/bribes/etc.
* Different areas with specialties that are randomized
* Different laws or contraban
* Affiliations, that make traveling more difficult based on where you
are coming from, or going to (probably combined with the travel ability
growing).

Basically, adding any sort of dependancy's graph, that has a few
different solutions, seems to make it much more interesting and
meaningful.

Also - one other little thing - I think I prefer the the phrase
"Randomly guided" as opposed to "generated". Pure randomness in a game
like this doesn't seem like a good thing to me.

What are other people's thoughts on this? I know a lot of us have a
goal of making a neat random-generated world, but has anyone stopped to
think of ways to make it actually improve game play? It seems that if
the gameplay-aspects are defined (and narrow in focus), it makes
focusing on what aspects to randomly generate to be less overwhelming,
and more attainable.
Anonymous
May 19, 2005 12:11:02 PM

Archived from groups: rec.games.roguelike.development (More info?)

Glen Wheeler wrote:
> It's the same with a random world. Replayability. If you can make

> generate a world with half as much interest as a hand-crafted one
then
> that's a huge amount of replayability.
>
> What do you call Crawl? I think that's a randomly generated world.
It's
> just a very simple world :) .

I guess I agree it helps with replayability. I guess my thoughts
lately have been on how deep the generation has to be before it adds
benifit, otherwise it is just "randomness" that just changes where
things are, which I don't see as really making things "interesting"
when replaying.

> Sorry I didn't respond to all of this. I think my approach to
generating
> my world is unpopular, because it is seen as a waste of valuable dev
time.
> I think it adds a huge amount to the value of my game. Replayability
is
> one, but there are more reasons:

I'm only vaguely familiar with your generation techniques, from other
threads. Do you have a site or a demo out yet? I think it sounds
interesting, personally. For this thread though, I was more interested
in the gameplay reprocusions then the actual techiques that generate
the random world. What are the differences someone encounters each
game? How do the differences effect gameplay choices? Is a city's
history something that effects a gameplay, or there to provide a
backdrop (which in itself is game play related IMHO)?

I've been finding myself agreeing a bit more with Kornel lately (in
regard to narrow focus on quick gameplay), which scares me a bit...
:-) (Just kidding of course! There's lots of wisdom in Kornel's
posts!)

> - Emergent behaviour. I may make a random generator which has many
of my
> own ideas in it. But the resultant world is more than just the sum
of its
> parts. Many times (through my much-maligned world aging process)
things
> have happenned which I did not expect, but can make sense of once I
work out
> what's going on. Which is fun.

Could you give some examples? This sounds pretty interesting.

> Anyway, you do what you like. Don't even include a world (or a
dungeon)
> if you don't want to.

Oh, I'm pro world-generation .. even if I can't find a use for it, I
still find the idea interesting. :-) I'm sure I'll continue to work
on it. I'm just trying to find out the essence of what use it can be
put towards, to let the desired outcome (from a game-play perspective)
drive the techique, instead of starting from the other side.

I'm coding most of this more for the fun of it, then for a goal of
release. I may someday have something releasable, but the reward is in
the development.
Anonymous
May 19, 2005 8:22:43 PM

Archived from groups: rec.games.roguelike.development (More info?)

Arakon wrote::

> What are other people's thoughts on this? I know a lot of us have a
> goal of making a neat random-generated world, but has anyone stopped to
> think of ways to make it actually improve game play?

Yes I think so. I saw it was good in other games, and so I just used it
in my games, too. I didn't really think about the reasons.

> It seems that if
> the gameplay-aspects are defined (and narrow in focus), it makes
> focusing on what aspects to randomly generate to be less overwhelming,
> and more attainable.

I finally ended up with a mixture of random and predefine elements in my
project.

There are three levels:

1) The world layout. Which areas exist and how they are interlinked.
This is predefined in my project (actually one definition per game
module, so different modules can have different world layouts).

2) The level layout. My project support a number of random level
generation methods, also predefined levels

3) Structures, e.g. furnished rooms, parts of buildings and such. They
are again predefined, but randomly selected and combined.

So my overall solution uses randomness to create variety and
upredictability, and it uses ppredefined structures in places where
automated algorithms seemed to produce inferiour results or just were
too work-intense to be implemented.

--
c.u. Hajo
http://h-world.simugraph.com
Related resources
Anonymous
May 20, 2005 1:05:26 AM

Archived from groups: rec.games.roguelike.development (More info?)

Arakon wrote:
> So. I'm left with wondering what point there is?

Simple. People think that generating random worlds will immidately give
them access to random-plots. Which is so untrue.

> I come to the conclusion that the game needs to be designed
> specifically for a fixed-duration play, and that the actual traveling
> to different parts of the map, and finding different places, must be
> important. These aspects being randomly generated must be part of the
> gameplay to make it interesting.

The best as usual is a compromise. Completely prewritten worlds get
boring quite quickly -- how many times you can do the same things, and
visit the same places until it get's boring? They lack replayability.
Completely random worlds on the other hand also lack a few things -- a
sense of purpose for the player is one thing. Another is the feeling of
being completely lost. And also it reduces the discussions around the
game -- everyone has a different universe, so what could they talk about
together? Of course there's a golden middle.

In GenRogue the highest-scale overland map is fixed. The kingdoms are
preprepared, so are the landmasses and general geography. To add some
sense of wonder and replayability tough, the lower-scale maps are
random. Imagine it as a seeded Plasma-fractal -- you havce the general
outline, but the detais as you go "deeper" are generated. That way the
player knows that he's eg. in Cerdania, or in the Province of
Ankhardrim. And he may talk about it with other players. What exactly he
will meet there is up to the RNG tough. I will provide one "official"
seed for the world, so people may discuss an exact given situation. Or
add the possibility to feed the generator the desired seed? Then people
will be able to say, "Lets all play this month on world #173864 and
we'll share experiences... This will be the major feature of GenRogue
;-). In every character file there will be the seed of the world, so a
player may try his luck again in the same world, or share it with his
friends.

Major city names and their general feel is also prepared. But the layout
of the streets, and the townfolk are not. This way there will be a
fammiliar Carsultyar, which the players may learn to like/hate but also
each time it will provide new challenges. This way we still give a
benefit to people who know the game world ("The best blades are forged
from Carlutyar steel") but yet don't make him bored with the game itself.

NPCs are treated like cities -- there are some pregenerated ones, but
most are random. You may still know and meet Dash Karrad, the Immortal
Librarian, but most of the people you meet will be fresh.

And last but not least -- plots. There are a few overarching
pre-prepared plots, out of which some may be choosen for the given game
(I wont give examples not to spoil the fun ;-) ), but generally the
quest and plots will be random generated.

Compromise -- the key to success.

> Also - one other little thing - I think I prefer the the phrase
> "Randomly guided" as opposed to "generated". Pure randomness in a game
> like this doesn't seem like a good thing to me.

Want a pure random roguelike? Pseudo-pascal:

procedure NextFrame(Move : Key);
var x,y : Byte;
begin
for x := 1 to 80 do
for x := 1 to 80 do
begin
RandomSeed(Move);
Screen[x,y] := Picture(Random(Color),Random(ASCII));
end;
end;

procedure RandomGame;
var Move : Key;
begin
repeat
Move := GetKey;
NextFrame(Move);
Draw(Screen);
until Bored;
end;

:-D :-D
(note that the game *depends* on what you press :-D)
(additions may be added like random_win_condition (a randomly picked
screen configuration)
--
At your service,
Kornel Kisielewicz (charonATmagma-net.pl) [http://chaos.magma-net.pl]
"Oh come on. We both know the truth about this game --
vapourware." -- Anathiel about GenRogue
Anonymous
May 20, 2005 1:08:17 AM

Archived from groups: rec.games.roguelike.development (More info?)

Arakon wrote:
> I've been finding myself agreeing a bit more with Kornel lately (in
> regard to narrow focus on quick gameplay), which scares me a bit...
> :-) (Just kidding of course! There's lots of wisdom in Kornel's
> posts!)

The funny thing about what you wrote is that generally I am considered
the developer of the greatest vapourware/featurecreep/overambitious game
around ;-).
--
At your service,
Kornel Kisielewicz (charonATmagma-net.pl) [http://chaos.magma-net.pl]
My opinions are my own. Share them at your own risk.
Anonymous
May 20, 2005 4:41:08 AM

Archived from groups: rec.games.roguelike.development (More info?)

"Arakon" <fabian.lemke@gmail.com> wrote in message
news:1116511500.938881.226240@g43g2000cwa.googlegroups.com...
> >From some of the previous topics, I've been thinking... what's the
> point to fully generating a random world? I see the point to randomly
> generated dungeons - replayability. Not knowing where the exit and
> items are - though I don't think that translates as well to a full
> world-map.
>

It's the same with a random world. Replayability. If you can make
generate a world with half as much interest as a hand-crafted one then
that's a huge amount of replayability.

What do you call Crawl? I think that's a randomly generated world. It's
just a very simple world :) .

> [...]

Sorry I didn't respond to all of this. I think my approach to generating
my world is unpopular, because it is seen as a waste of valuable dev time.
I think it adds a huge amount to the value of my game. Replayability is
one, but there are more reasons:

- I don't abuse the static world. When I play test the game I don't want
to already know the best or my favourite way to do things. I want to be
forced to think every time I start a new character.
- I don't want to be bored to tears. How many times do I playtest the
game? Zillions...static world would be boring.
- Emergent behaviour. I may make a random generator which has many of my
own ideas in it. But the resultant world is more than just the sum of its
parts. Many times (through my much-maligned world aging process) things
have happenned which I did not expect, but can make sense of once I work out
what's going on. Which is fun.

Anyway, you do what you like. Don't even include a world (or a dungeon)
if you don't want to.

--
Glen
L:p yt E+++ T-- R+ P+++ D+ G+ F:*band !RL RLA-
W:AF Q+++ AI++ GFX++ SFX-- RN++++ PO--- !Hp Re-- S+
Anonymous
May 20, 2005 5:39:42 AM

Archived from groups: rec.games.roguelike.development (More info?)

"Arakon" <fabian.lemke@gmail.com> wrote in message
news:1116515462.958427.291180@o13g2000cwo.googlegroups.com...
>
> Glen Wheeler wrote:
>> It's the same with a random world. Replayability. If you can make
>
>> generate a world with half as much interest as a hand-crafted one
> then
>> that's a huge amount of replayability.
>>
>> What do you call Crawl? I think that's a randomly generated world.
> It's
>> just a very simple world :) .
>
> I guess I agree it helps with replayability. I guess my thoughts
> lately have been on how deep the generation has to be before it adds
> benifit, otherwise it is just "randomness" that just changes where
> things are, which I don't see as really making things "interesting"
> when replaying.
>

I agree with you here; do it properly or don't do it at all :) .

>> Sorry I didn't respond to all of this. I think my approach to
> generating
>> my world is unpopular, because it is seen as a waste of valuable dev
> time.
>> I think it adds a huge amount to the value of my game. Replayability
> is
>> one, but there are more reasons:
>
> I'm only vaguely familiar with your generation techniques, from other
> threads. Do you have a site or a demo out yet?

There was a site, but then it became unmaintainable (damn RL) so I took it
down. Even with the snail-pace of my RLs evolution the site became
hideously out of date.
No demo, sorry. It's just not ready.

> I think it sounds
> interesting, personally. For this thread though, I was more interested
> in the gameplay reprocusions then the actual techiques that generate
> the random world. What are the differences someone encounters each
> game?

Well, pretty much everything. The races that you encounter are randomly
generated (they have are constructed by genome based on their evolution). I
explained it in more detail (yet still not ``in detail'') in the other world
generation thread. In one case the entire plot may revolve around one race
controlling all the iron/ore/whatever ore in the world and using it to
enforce military rule on the other races. In another world there may be a
cult of crazy farmers plotting to summon some god into the world.
Technically it's hard to achieve these things but fun and satisfying
nonetheless.

> How do the differences effect gameplay choices? Is a city's
> history something that effects a gameplay, or there to provide a
> backdrop (which in itself is game play related IMHO)?
>

Mostly backdrop, but there are significant gameplay considerations. Most
affected is the economy and population: who lives here and what are their
jobs, and how comfortable are they with surviving. Once a race has
surviving under their belt they tend to worry about other extravagancies,
like taking over the world :) .

Dungeons are entrances into the underworld. The underworld isn't that
different from the overworld except that people have evolved to make do
without light and are extremely prejudiced against the overworld people. So
if you are an underworldian, despite being a merman, dwarf, or whatever you
ended up evolving into, prepare for a fight up above. Moreover, prepare for
a fight *on the way up*. Go dungeons go.

> I've been finding myself agreeing a bit more with Kornel lately (in
> regard to narrow focus on quick gameplay), which scares me a bit...
> :-) (Just kidding of course! There's lots of wisdom in Kornel's
> posts!)
>

Kornel has proven himself, with DRL and DRL. That's more than myself...

>> - Emergent behaviour. I may make a random generator which has many
> of my
>> own ideas in it. But the resultant world is more than just the sum
> of its
>> parts. Many times (through my much-maligned world aging process)
> things
>> have happenned which I did not expect, but can make sense of once I
> work out
>> what's going on. Which is fun.
>
> Could you give some examples? This sounds pretty interesting.
>

The last time this happenned was when a stangely large (as in area)
foothill was created with a strangely large lake on top. The foothill was
just steep enough to make the water into waterfalls as it came down. One of
the simplifying aspects of my world-generation is that lakes, once created,
don't really dry up. If they flow too far they just kinda sit there and
stop. If the land is still angled downward they erode away until it all
settles down. So on top of this foothill there was also a settlement (since
settlements are likely to be created near water) with a customary Rex and
Regina. The water flowed off the foothill and around the base, until the
routine which cuts the flow stopped it and eroded away the area around that
place. The end result was a rather successful but completely isolated city
of mer-people who were rather more advanced than the rest of the people in
the world (because of the simplified manner in which I do teching, they had
no trouble being relatively unbothered and having a large area to expand
into).
I thought that was cool. If I had more of the game done, then it would be
interesting from a player's point of view to get up to that city and trade
with them for things which just plain do not exist anywhere else in the
world. The design doc for quests would have a lot to say about it too, but
shouldn't mention it since it's not coded *at all* :) .

>> Anyway, you do what you like. Don't even include a world (or a
> dungeon)
>> if you don't want to.
>
> Oh, I'm pro world-generation .. even if I can't find a use for it, I
> still find the idea interesting. :-) I'm sure I'll continue to work
> on it. I'm just trying to find out the essence of what use it can be
> put towards, to let the desired outcome (from a game-play perspective)
> drive the techique, instead of starting from the other side.
>
> I'm coding most of this more for the fun of it, then for a goal of
> release. I may someday have something releasable, but the reward is in
> the development.
>

That's my attitude also :) .

Have fun!

--
Glen
L:p yt E+++ T-- R+ P+++ D+ G+ F:*band !RL RLA-
W:AF Q+++ AI++ GFX++ SFX-- RN++++ PO--- !Hp Re-- S+
Anonymous
May 20, 2005 1:06:40 PM

Archived from groups: rec.games.roguelike.development (More info?)

Glen Wheeler wrote:
>
> Well, pretty much everything. The races that you encounter are
randomly
> generated (they have are constructed by genome based on their
evolution). I
> explained it in more detail (yet still not ``in detail'') in the
other world
> generation thread. In one case the entire plot may revolve around
one race
> controlling all the iron/ore/whatever ore in the world and using it
to
> enforce military rule on the other races. In another world there may
be a
> cult of crazy farmers plotting to summon some god into the world.
> Technically it's hard to achieve these things but fun and
satisfying
> nonetheless.
>

It sounds interesting from the exploration point of view.

> [clip]
> I thought that was cool. If I had more of the game done, then it
would be
> interesting from a player's point of view to get up to that city and
trade
> with them for things which just plain do not exist anywhere else in
the
> world. The design doc for quests would have a lot to say about it
too, but
> shouldn't mention it since it's not coded *at all* :) .
>

I like the idea a lot. I was reading about ultima online, and how they
simluated everything at the begining. Eventually though, they found
that the cool effects simulation gave were something that no one ever
really noticed, so they started just doing things as events. I'm
starting to lean towards the same direction - it sounds like a good
idea for starting off the world, but I think for me, adding the
"flavor" after, and during, using pre-scripted events/conditions/etc
sounds like a better option.

It's easy enough to code up a quick event, it's a lot more time
consuming to balance out a simulation, and rare that it would come to a
specific point at the "right" time. For backdrop though, it still
looks pretty effective.
Anonymous
May 20, 2005 1:19:52 PM

Archived from groups: rec.games.roguelike.development (More info?)

Kornel Kisielewicz wrote:
>
> Simple. People think that generating random worlds will immidately
give
> them access to random-plots. Which is so untrue.

Very true. Random-plots is a completely different subject, and can
work just fine in a static, or semi-static world map. It's a lot
easier to start if there is a backdrop actually.

> [clip]
> being completely lost. And also it reduces the discussions around the

> game -- everyone has a different universe, so what could they talk
about
> together? Of course there's a golden middle.

Oh - Good point. Community building is important, especially to this
style of game.

> In GenRogue the highest-scale overland map is fixed. The kingdoms are

> preprepared, so are the landmasses and general geography. To add some

> sense of wonder and replayability tough, the lower-scale maps are
> random. Imagine it as a seeded Plasma-fractal -- you havce the
general
> [clip]

Ahh! Some tid-bits on GenRogue! Cool :) 
It sounds like a great hybrid. If only I could give you a development
team to finish it .. :-)

Having a semi-set setting is nice though. You wont' know what's
different, but know generally where to go for what. A nice idea,
that's for sure. It layers the generation algorthms nice too, so the
code can be nice looking.

> [clip]
>
> And last but not least -- plots. There are a few overarching
> pre-prepared plots, out of which some may be choosen for the given
game
> (I wont give examples not to spoil the fun ;-) ), but generally the
> quest and plots will be random generated.
>
> [clip]

Yeah, the plots are the important parts. I think that more interesting
encounters could make for a good game as well. If a player irritated
an important NPC, other NPCs could jump them somewhere, or ambushs, or
running across NPCs being ambushed, etc. There's just lots of options
once plots are full-featured, especially with encounters.

I have a busy day, so I may add some more later.
An interesting topic though. Personally, I've come to the conclusion
that a fully random world is interesting, though may not add directly
to game play by itself. The world generation would have to include
game-play reasons during creation.
Anonymous
May 20, 2005 4:00:08 PM

Archived from groups: rec.games.roguelike.development (More info?)

Arakon <fabian.lemke@gmail.com> wrote:

> I like the idea a lot. I was reading about ultima online, and how they
> simluated everything at the begining. Eventually though, they found
> that the cool effects simulation gave were something that no one ever
> really noticed, so they started just doing things as events.

It's not that people didn't notice the simulation, it's that it didn't
make for a fun game.

<http://www.legendmud.org/raph/gaming/economy.html&gt;

--
Jim Strathmeyer
Anonymous
May 21, 2005 12:56:27 PM

Archived from groups: rec.games.roguelike.development (More info?)

Jim Strathmeyer wrote:
>
> It's not that people didn't notice the simulation, it's that it
didn't
> make for a fun game.
>
> <http://www.legendmud.org/raph/gaming/economy.html&gt;
>
> --
> Jim Strathmeyer

For economy, yes. There was other "emergent" behavior as far as
ecology went. The typical example is the dragon that runs out of food
bothering the village. All in all, it didn't happen very often, and
when it did, no one really took notice. You're right - it just wasn't
"fun". I think part of the fact is that no one really noticed the
"Why" of a lot of it.

I'd provide references, but the mud-dev archive was down last I checked.
Anonymous
May 23, 2005 8:17:11 PM

Archived from groups: rec.games.roguelike.development (More info?)

Arakon wrote:
> Kornel Kisielewicz wrote:
>>together? Of course there's a golden middle.
>
> Oh - Good point. Community building is important, especially to this
> style of game.

I learned the value of the community while working on DoomRL :-).

>>[clip]
>
> Ahh! Some tid-bits on GenRogue! Cool :) 
> It sounds like a great hybrid. If only I could give you a development
> team to finish it .. :-)

If only I could find a dev-team to work with :-/

--
At your service,
Kornel Kisielewicz (charonATmagma-net.pl) [http://chaos.magma-net.pl]
"From what I've read, a lot of people believe that GenRogue
exists and will be released some day" -- Arxenia Xentrophore
Anonymous
May 23, 2005 8:34:28 PM

Archived from groups: rec.games.roguelike.development (More info?)

Kornel Kisielewicz wrote:
>
> If only I could find a dev-team to work with :-/

If only you could be turned back to the good side ... I feel the
conflict within you ... let go of your Pascal ... Python is the path to
redemption!

--
http://www.gnu.org/philosophy/right-to-read.html
Palladium? Trusted Computing? DRM? Microsoft? Sauron.
"One ring to rule them all, one ring to find them
One ring to bring them all, and in the darkness bind them."
Anonymous
May 24, 2005 7:35:28 AM

Archived from groups: rec.games.roguelike.development (More info?)

Twisted One wrote:
> Kornel Kisielewicz wrote:
>
>>
>> If only I could find a dev-team to work with :-/
>
> If only you could be turned back to the good side ... I feel the
> conflict within you ... let go of your Pascal ... Python is the path to
> redemption!

You do not understand the power of the Dark Side...
--
At your service,
Kornel Kisielewicz (charonATmagma-net.pl) [http://chaos.magma-net.pl]
"A random group of people is useless for anything except
statistics" -- Anubis
!