[M] monster evolution through genetics

[Kurpitsa]

Now that I think again this idea, this idea could be pretty good in dungeons. I don't know is there normal and hard mode, but dungeons are suppose to be challenges, they're not suppose to be ran easily trough. Something like this would be pretty awesome.

If there are normal and hard mode, I don't know would it be good to have this on the other, since you first play trough normal mode, it's quite confusing for the monsters to change so much in hard mode. More health, damage and maybe a couple skills more for some monsters is good for hard mode.

[Arghore]

realy like the idea, couple of thing i like to comment add too:

Constraints, to make populating area's easier i think it would make more sense to link some constraints to the area the monsters are placed ...
this means work can be divided up into someone who makes the monster builds and someone who designs the area/places the spawns.

Constraints Monster
profession= N
attribute lines={Blood Magic,Soul Reaping}
level= *area_level*
attribute points= 4*level (so lvl8 = 32)
max_skills= area_max
elite_skill= area_elite

This way a skale that is put in the starter area will be allot different then one set in an elite end game area...
--------------------------------------
Then i would raise a concern allready raised by Hallomik, if you were to assign all the bloodmagic spells to all the bloodmagic creatures, this would mean that over time, all the creatures that use bloodmagic will migate towards the same bloodmagic build ... making them sortof the same ... obviously we dont want all blood monsters running around with the same cookie cutter build ...

so i would propose to have a skill_pool for every creature, sortof combining your first idea, with your respons to kurpitsa ... Then this monster profile could be further astablished by setting 1 or 2 fixed skills, offering a bit more predictability, but more over, a more predefined recognisable monster profile (but you allready did this by setting lifesiphon, just wanted to put it out in words, with an argument why this (to me) would be desireable...

The monster constraints would then have this skill set in its constraints to pick from, meaning multiple blood creatures will be somewhat different:

As an example different touching monsters from blood magic, which can also migate into a ranged blood drinker depending on its evolement:

skill_pool1= vampiric gaze { vampiric bite, vampiric touch, blood drinker, vampiric swarm, jaundiced gaze, strip enchantment, blood renewal, barbed signet, signet of lost souls, unholy feast}

skill_pool2= life siphon { touch of agony, signet of agony, wallowsbite, shadowstrike, signet of agony, plague touch, mark of subversion, demonic flesh, blood bond, masochism (pvp)}

Skill pool 1 obviously has a vampiric profile, while pool2 has more of an agonising profile, both have touch and ranged skills, with this i didnt take into account wether they are actually ranged or melee creatures, with this i mean, melee could be combined with some order spells, ranged perhaps with some wells ... anyway, i think this illustrates enough how the monster can evolve allot though still keep some sort of monster profile that fits the creature... the above mentioned linking of constraints with the area means the map builder does not have to concern himself with also fully equiping the monsters... ofcourse some communication would be nice so that the monsters placed would both fit the area and has a skill pool that also fits that area...

[Fril Estelin]

Originally Posted by tmakinen

But this is all rather technical stuff ...

I don't want to disrupt your thread tmakinen, but the technical elements are A LOT more complex than what you quote. It is definitely not as simple as you say (just imagine if activating a skill while jumping was as easy as calling one function).

[Arghore]

it is mostlikely, though jumping and having spell animations at the same time ... thats a different ballgame...

[MadMesh]

I like this idea, this mechanism would keep PvE a lot more interresting. Most people steamroll through PvE in GW1 with SabWay or DiscordWay making general PvE even in Hard Mode a joke.

I dont think putting an "adaptive mob" system would automaticly make everything more difficult, it would actually make PvE as hard or as easy as desired.

The only problem I see with this is farming, farming would become hard if not impossible. Now I'm not a big farming fan but some people are and a system like this could kill/cripple farming as a type of gameplay.

[Fox]

That's a great idea, but I highly doubt the time will be taken to make such a dynamic PVE system.

[dakka dakka]

I like your idea, is very original and looks good on paper, though it may look like a code nightmare to programmers

@Nevin:

We don't even know if there IS going to be a hard mode, but having a system that the OP described doesn't make the game harder per se, just more flexible and forces the player to think on their toes.

I seriously have doubts of there being a hard mode unless it is an alternate "elite" server, but this can create a division between players which ANET doesn't want.

[Shadow Ritualist]

Yeah, this is nice, it'd be the best, if each build was semi-random, but still leaning towards specific skills. And they'd need to use skills that work well with each other, but overall this is a great idea. I don't think it'd be really practical unless only some of the stronger monsters had this in place, considering it won't matter as much for weaker monsters.

And this could also maybe be for specific smarter races. A skale wouldn't change the skills it used, but maybe a tengu would.

[TedTheShred]

Brilliant, genius idea. Should be implemented. No Question.

Have you considered the same system applying to groups of enemies?

The system would work as follows: The spawning of a group of enemies is determined by their fitness function. Each area, each spawn point has a set difficulty, and the builds of monsters are assigned a difficulty based on their recent fitness function.

Groups of monsters within a set range of sizes would be spawned with a random assortment of monsters of the same set (ie monk skale and mesmer skale, not monk skale and mesmer undead) whose combined difficulties add to the desired difficulty of the group. From this, the fitness function of the group is assessed the same way as individual monsters (via staying power and damage output).

This would allow monsters to develop support builds as well as damage builds. For example, a group of monsters with a monk would have better staying power over one without; or an interrupt ranger might thrive in a group of monsters and fail on it's own.

This would also allow for groups of monsters to synergize. A great Spiteful Spirit monster build might not thrive next to that same great interrupt ranger. The build with an IR would therefore suffer and it's fitness function would drop.

The same system governing both the micro and the macro would better achieve a unified front of enemies that adapt to the collective threat of players than the system solely governing the evolution of individual monsters. Players would know what kind of monsters are there based on the region, but the builds of the monsters and monster groups would be an ever-changing landscape.

[tmakinen]

Yes, I've been playing with the idea but there are some issues related to group evolution.

• it is harder to define an objective fitness function for an entire mob
• it requires much more resources (there's at least an order of magnitude size difference between the set of monster types and the set of possible mob types)
• since GW2 will be solo friendly, it is to be expected that monsters are primarily solitary instead of in mobs as in GW1
• the advantages of mob level evolution are not that clear over monster level evolution

Let me elaborate the last point. Monster level evolution already considers mob composition. Let's say that one area is populated with mobs of blood necros only and the other with mixed mobs of blood necros and healers. Even if the blood necros in those two environments are made exactly the same, they will evolve quite differently. Blood necros without healers must have a self heal to be fit. Blood necros with healers can drop the self heal and even dip deep into sac skills and still be fit as judged by the exactly same criteria as the monoculture necros.

This also answers the question 'won't all monsters then evolve to the same builds'. While restricting the available pool of skills is one way to tackle it, an even more effective way is to adjust mob composition. If mob composition changes from monocultures towards balanced teams as the game progresses, the algorithm automatically turns starter monsters into relatively ineffective generalists and endgame monsters into highly effective specialists.

[Evangelionfan]

a randomized build though sounding nice to begin with would be nothing more than a nusance....MAYBE if it were say a monster has 4 skillslots and he has 5 skills he could use...but nothing more. people need to gear for areas not monster groupings.

I dont mind fighting mobs but it may seem cool for say pre searing ascalon where once youre lvl 5 it all becomes very easy, but In a high level situation like in UW i dont want to deal with monsters haveing a mixed variety of skills that we cant protect against.

[Kendil]

I've a question, how fast will the monsters adapt to new builds/teams? If they are supposed to counter farming, then they need to adapt pretty quickly. But in other posts I get the impression that it'll take like 6month for an effective build to spread, and that is way too long if it is to tackle farming. Or isn't it?

[tmakinen]

The overall evolutionary speed can be set by the developers. However, the speed of any particular monster population depends on their current fitness, respawning rate and the diversity of threats that they encounter. Monsters that are killed only on occasion in 'fair' fights and by a variety of random builds evolve very slowly, staying essentially the same all the time since there isn't any evolutionary pressure. Monsters that are excessively overfarmed by a single gimmick build that they can't counter feel a huge evolutionary pressure and may develop proper counters to that gimmick in a couple of weeks.

[Shadow Ritualist]

Another way they could be kept away from all migrating to the same builds, is have them meet up with other species and fight them, using their fitness in that as well, depending on the type of the enemy, some may evolve to be stronger towards that type.

[Kendil]

But, couldn't this be overcome by farmers? If you use a gimmick build to defeat (for example) scales. Then they evolve a counter to that. What if the farmers just creates a second gimmick build, and that builds counter is the first gimmick build strong against.

Is it a too small a chance? If not, is it a way to counter that?

[TedTheShred]

Another way they could be kept away from all migrating to the same builds, is have them meet up with other species and fight them, using their fitness in that as well, depending on the type of the enemy, some may evolve to be stronger towards that type.

Interesting proposition. I would like to see that. Periodical large-scale fights between different species of monsters would accelerate the speed of change (given somewhat frequent respawn times to lengthen the engagement) and force the two groups into different, conflicting niches. This way, when players encounter these battle-tested builds they will be different experiences and allow the monsters to evolve their builds against players from different starting points, likely decreasing homogeneity.

What if the farmers just creates a second gimmick build, and that builds counter is the first gimmick build strong against.

Not an at-all impossible scenario; selectively pressuring monsters into builds susceptible to certain farming-builds. I would point out, however, that this would take a great degree of cooperation from a large number of people to constantly pressure monsters into certain builds. The randomness of the system will help stop this from happening as well as the likelihood that over time if two alternating builds are susceptible to two gimmicky builds, their fitness factor will drop, and they will b selected less often.

[shinyglove]

what's wrong with the monsters atm? .

[Kendil]

Originally Posted by shinyglove

what's wrong with the monsters atm? .

They are easy to farm. They prove no challange once you've learnd their build. They are no different from what they was 4 years ago (which can be a bit boring for the ones that has played for that amount of time), which is not the case for PvP, as the meta changes now and then - which makes gaming more of a challange and more intresting over a longer period of time.

And to go a bit off topic, the monster AI is made to be stupid, they don't kite etc. as good as one should. With smarter monsters it'd be a lot more challenging. Add that with "PvE meta" (or whatever you want to call tmakinen's idea) and you've got a better game.

[TedTheShred]

Why is this thread so buried? It's the best idea ever.

I came across this article today:

Evolution of Adaptive Behaviour in Robots by Means of Darwinian Selection

I immediately remembered this suggestion.

Are there any other aspects of Monster AI that can be genetically formed? This article suggests that the outcomes could be quite robust.

Perhaps everything from skill bars to movement could be governed by the genetic development of traits. Alpha testers would encounter randomly wandering monsters. Selective pressures would then naturally transform erratic movements into tactical kiting.

[Arghore]

Tnx for ressurecting this thread ... it is indeed one of the best ideas on monster management for a(ny) game, hope Anet sees it and does something along these lines to make the experience of area's something different everytime...