Evolved loot

JemyM

Okay, now roll sanity.
Joined
October 26, 2006
Messages
6,027
I just argued for how simple rules can create complex results in one thread and the problem with how random loot nullifies exploration in another. Then it struck me, why have there been no attempts at introducing evolution yet?

Before the game begins, the game will simulate evolution using the following rules;
1. More loot than necessary is in the game
2. Due to the nature in the game, loot will get dragged around, gain and lose attributes
3. Loot end up in areas it's adapted to.

Nature includes factors like;
1. The carrier (attributes of the carrier)
2. Closeness to someone who can acquire the item (seclusion means an item doesn't move around)
3. A couple of ways an item can be acquired (kill, buy, steal)

Now what will happen is that loot is no longer "random". It will be decided by it's environment. Rich and protected and secluded strong monsters are more likely to have absurdly attractive weapons than poor and unprotected weak monsters living in the forest just outside the village, yet if it would happen it would be a rare (and perhaps fun) event, especially if each loot carry a backlog so you may see how the item ended up in the hands of that goblin.
 
Joined
Oct 26, 2006
Messages
6,027
I have already suggested this in another thread, it was not only for loot but for the entire game. There was a fairly long discussion about it.

In the end one conclusion was that because of the survival of the fittest principle the end result might be a few super monsters owning all the items.

But if put into some bounds and with certain rules it could be good.
 
Joined
Oct 25, 2006
Messages
6,292
I have already suggested this in another thread, it was not only for loot but for the entire game. There was a fairly long discussion about it.

In the end one conclusion was that because of the survival of the fittest principle the end result might be a few super monsters owning all the items.

But if put into some bounds and with certain rules it could be good.

This may happen if you only have one attribute (strength), remove landscape (neccessary for natural selection and creating different cultures), and exclude strength in numbers.

Add competing powers (strength, wealth, numbers), add a landscape that divide populations (a place without dragons is a place where the dragon do not get all the loot) and increase teamwork relative to strength (meaning a lone dragon attacking a village is a dead dragon) and you solved that problem.
 
Joined
Oct 26, 2006
Messages
6,027
Yes, that was in the discussion too, I suppose dwarf fortress is the closest you can get to this kind of system.

It is extremely complex both in terms of computing power and in terms of player immersion on top of that it is very complicated from a programming point of view, that is probably why none really attempt it.

That said I think gaming would really benefit from increased complexity in other aspects aside from graphics, graphics, graphics, I mean the other aspects has mostly been going backwards instead of forward. Developers can spend a year on developing a monster shader with extremly complicated math to make water look real. Imagine if the time was spent on good math for the loot system instead?
 
Joined
Oct 25, 2006
Messages
6,292
Yes, I most definitely agree on that one. What we have seen in games is more counterevolution than evolution. Far more systems have died than grown in the past years.
 
Joined
Oct 26, 2006
Messages
6,027
Yes, I most definitely agree on that one. What we have seen in games is more counterevolution than evolution. Far more systems have died than grown in the past years.

It does seem like shinier, but often uglier, graphics, are all the advances we're getting from mainstream developers. Mapping systems are worse, inventory management is worse, stats and numbers are scary and should be hidden whenever possible, and controls are constantly being "simplified" to the point that soon we might be able to break out the Atari 2600 joystick again. :p

Imagine if they embraced more procedural generation of worlds, monsters, quests, etc. instead of mindlessly copying what the latest biggest seller is doing. A friend of mine and I toyed around with that a few years ago, but real life took too much of our time.
 
Joined
Jul 3, 2011
Messages
1,147
Location
Madness
A similar system is actually in effect when you generate a new world in Dwarf Fortress. It first generates the terrain (so that it makes sense), then the game scatters different populations around the game world (including 1000 "megabeasts" (the very big nasties that can cause large amounts of destruction, these are by the way also randomly generated, so they tend to be slightly shogoth-like), then it lets the world age, civilizations spreads and flourishes/gets crushed. It is entirely possible to start a world where goblins are the dominant faction (they seem to breed really fast) and where the megabeast population is low because of their encounters with civilizations, but it is also entirely possible to get a world where goblins are struggling (although this does not seem to be very common, the goblin civilizations tend to be rather successful, unlike kobolds). While the world does not deal with loot, I guess it would not be an unthinkable addition, after all, the game has a rogue like mode.

*edit* Darn, this is why you should not start cooking while still in the process of typing out a post. GothicGothicness beat me to it.
 
Joined
Jun 2, 2011
Messages
1,756
Location
Stockholm, Sweden
Very intriguing. I'm curious about how well it really works out. I still think worlds designed and populated by designers have more character / history / depth / and are more fun to explore.
 
Joined
Aug 18, 2008
Messages
15,682
Location
Studio City, CA
Sounds more like an interesting experiment than sensible game design.
 
It works really well for Dwarf Fortress. You have to remember that Dwarf Fortress is not a story driven game though, but more of a huge sandbox for geeks. And part of the reason why the game works so well is because of the randomly generated world, which creates an endless amount of replay value.

Due to the way the world is randomized, you also get a rather realistic spread of the different civilizations. This is more obvious in the rogue like mode than when you build a fortress.

It should by the way be noted that your own dwarf civilization is affected by this as well. You can actually start at war with one or more other civilization. This is rare, but it has happened to me more than once (it always seem to be the elves that I'm at war with).
 
Joined
Jun 2, 2011
Messages
1,756
Location
Stockholm, Sweden
Something slightly similar has been used by Item Randomiser mod for Baldur´s Gate 1&2.
For example, items from Shadows of Amn portion are divided into 7 tiers based on their power/usefulness, locations/enemies are divided into 7 tiers as well and in the beginning of the game items are randomly distributed into places with matching tier. Iirc, some locations get assigned more than one slot, some slots may remain empty after the randomisation and some story-related items don´t get randomised.
There´s also a possibility to make some percentage of items disappear from the game entirely.
This is of course a bit rigid since afaik items from one tier have no chance to appear in locations of different tier and characteristics of items are set.

Still, I think it´s a pretty solid compromise between hand-crafted (set item characteristics) and generated (randomisation) elements and it does quite wonders for replay value of these games.
 
Last edited:
Joined
Apr 4, 2008
Messages
2,437
Location
Prague
An interesting idea but I can't say I'd be a fan. Such a system would still have generic items, albeit generic items created on the fly for the purposes of one's playthrough.

I prefer more unique items with a story behind their effects, in the vein of BG and Morrowind itemization. I guess some sort of hybrid could exist... but it would require a lot of effort on the part of a developer.
 
Joined
Nov 10, 2008
Messages
5,980
Location
Florida, USA
Very intriguing. I'm curious about how well it really works out. I still think worlds designed and populated by designers have more character / history / depth / and are more fun to explore.

Might be possible to have both. Bethesda's latest have a random system taking care of the wilderness and designed cities. It is a shame for me to say it but with all the companions I use in Fallout NV I would actually like some scaled spawns.
 
Joined
Oct 26, 2006
Messages
6,027
Sounds more like an interesting experiment than sensible game design.

Even game design evolve. Good stuff must begin somewhere. All sensible game design begins with experimentation. Todays big business is unfortunately about not experimenting at all so there is nothing to evolve.
 
Joined
Oct 26, 2006
Messages
6,027
Even game design evolve. Good stuff must begin somewhere. All sensible game design begins with experimentation. Todays big business is unfortunately about not experimenting at all so there is nothing to evolve.

I agree.

I'm not saying the experiment has no merit. I just don't see games improving because of it.

Then again, I'm pretty confident I can design a fantastic loot system without going through such a simulated environment.

I think what I'm trying to say is that it's "going the long way around". Seems you're more interested in the meta-game aspect of the experiment, than the resulting game.

Loot should be handled according to the game you're trying to make. There are much easier ways to handle loot than this, in all the games I can imagine. Even games trying to simulate a real environment. Few games have managed to do so, though, as it's the kind of thing modern developers largely ignore as vital.

Did you try dwarf fortress DArt ?

Nope. The concept doesn't really appeal to me.
 
I agree.

I'm not saying the experiment has no merit. I just don't see games improving because of it.

Then again, I'm pretty confident I can design a fantastic loot system without going through such a simulated environment.

I think what I'm trying to say is that it's "going the long way around". Seems you're more interested in the meta-game aspect of the experiment, than the resulting game.

Loot should be handled according to the game you're trying to make. There are much easier ways to handle loot than this, in all the games I can imagine. Even games trying to simulate a real environment. Few games have managed to do so, though, as it's the kind of thing modern developers largely ignore as vital.



Nope. The concept doesn't really appeal to me.

Other than hand placed and random with some limits, what other ways are there?
 
Joined
Oct 8, 2009
Messages
4,425
Location
UK
Other than hand placed and random with some limits, what other ways are there?

That's a lot broader than you make it sound :)

"Random with some limits" is arguably what we're talking about here. At least, it's just as much generated loot as other systems with a set of parameters. You can generate loot in similar ways, but with a simplified system behind it. To go through the effort to reflect evolution just for the sake of loot seems pretty counter-productive to me. You'll most likely end up with a less than optimal entertainment experience, because "reality" isn't a good provider of entertainment.

I mean, Jemy is talking about "simple rules" - but you can't make this kind of system in a convincing way without reasonably complex rules. At least, I can't think of a good way of simulating evolution without it becoming incredibly "systemic" and "fake". If you REALLY want a plausible exchange of loot to lend a kind of "history" to what you find, you can't do it in a simple way. It won't work.

I'd certainly have to see it to believe it.

Generally, I think combining hand-placed loot with intelligently generated loot is very sufficient in most games.

It depends on what you're trying to do. Hand-placed loot is often preferable, when feasible. But there are games where the unpredictable nature of randomised (or generated) loot provides the optimal experience.

This kind of "evolution-based" loot is, to me, superfluous. You can simulate realism quite sufficiently by making a simplified generator. Realism for realism is not something I consider appropriate - unless it's truly worth the effort.

But I'd rather talk about specifics. In any specific game, there will be nuances to consider for optimal loot distribution - and it's hard to quantify in general terms.

I'm just saying that, off-hand, I can't really imagine a game where such an effort to simulate "realistic" loot evolution would be pragmatic or sensible as part of the design.
 
Back
Top Bottom