Animal behaviour suggestions

[stratvox]

I was asked to cross post this comment I made back in February to the Wishlist.

Here's the link to the original post: 

Quote

I've given this some thought. Bears are pretty much all right AFAIC; they're not really interested in you per se, but they don't like you to be too close. This is very much like real life bear behaviour, actually. One quibble I have had with bears in the game is that in the wild most bears eat little to no meat; they are usually living off berries, roots, insects, etc, so for any given bear I think the game should have a possibility that meat from that bear carries a parasite risk (unlike, say, wolves, where it should be always there). Anyway...

I would love to see pack behaviour from wolves. It would make them far more challenging. Given wolf range sizes (can range from 15 to 200 square kilometers) I'd think you'd really be looking at one pack per major region, with maybe a solo wolf or maybe two in the transition regions (Ravine, Crumbling Highway, Winding River). Pack size could bounce around a bit (and be replenished) with larger areas like Pleasant Valley having 10-12 in the pack, Broken Railroad having five or six, and Forlorn Muskeg, Mystery Lake, Coastal Highway, and Timberwolf Mountain having eight to ten (numbers should be both ranged and adjustable according to difficulty setting). When a wolf spots you, its barking will draw other wolves in the region towards the wolf that barked, meaning that it becomes very important to deal with the wolf and make tracks pronto to avoid a wolf donnybrook taking you down a half hour or so later. Even cooler would be to have a solo wolf a long way away from other wolves bark and then start to track you, barking from time to time to redirect the coming wolves in the other pack. Hearing the series of barks while the wolf keeps its distance would definitely crank up the paranoia factor. Things like lighting a fire could deter the tracking wolf causing it to yip and flee... if you manage to deter the tracking wolf in some way the yipping will cause the other wolves to return towards their patrol areas. If you kill the wolf you'll want to book out and put some distance, to return the next day after the wolves have come, mourned their fallen comrade, and returned to their patrol areas, to harvest the carcass. If you end up killing a wolf without alerting it to your presence and therefore having no alarm barks to draw other wolves to its location, then you're good to go right away.

Deer should herd, so there'd need to be an attractor to help get deer to start to gather together, with the idea being having several deer herds form per region. Careful planning of ranges for both wolves and deer to give the deer a chance given the wolves touch of death wrt deer... or alternatively set it up so that it takes several wolves to get a deer and having other deer attempt to interfere by rushing the wolves to give the target deer a chance to escape. Predators very much fear injury... if a deer "sprains its ankle" it can still eat while it's healing, but a predator that's been hobbled faces an almost certain death sentence via starvation. So the idea here would be a wolf spots a deer herd, barks, when say three of them are there together they try to split a deer off from the herd, with the other deer attempting to interfere by attacking the wolves (c.f. that famous online vid of a hunter getting taken down by a deer smackin' him upside the head with its hooves). If they succeed in causing one wolf to flee yipping the wolves withdraw, the other wolves stop coming, and the deer herd returns to grazing.

Wolves and bears are both attracted to carrion (i.e. dead deers) so from time to time a deer should keel over (starvation? old age? disease? whatever). This should cause the local herd of deer to start migrating away from the carcass as they know that bears and wolves will be coming to ravage the carcass. If a bear finds it first, then when the wolves get close enough to see the bear working over ol' stiff-legs they'll run away yipping causing the other wolves to lose interest... if a wolf arrives first it'll bark to bring the other wolves directly to the carcass and start to feed... then when mister bear shows up having an algorithm decide whether the bear wants to try to scare away the wolves or decide to seek easier pickings elsewhere.

Similarly with moose, but as moose are significantly stronger have the wolves wait until the whole wolf pack shows up before trying to take it down. The way wolves hunt is actually quite similar to human primeval hunting... chase the animal to exhaustion before trying to take it down so as to minimise the risk of injury. So, when a few wolves gather around a deer herd they'll start trying to isolate an animal, but for a moose they'll wait until there are significantly more of them there before starting to harass the moose... and give the moose a chance to spot them and scare them off before too many of them gather, causing the hunt to be called off (I've seen moose up close and personal... they are intimidating AF).

Lastly, I'd like to see two more animals... foxes (with craftable head gear! I get why the game doesn't have it, but damn...). Foxes will do things like raid your outdoor food store, be snarable/trappable but with a much lower chance than rabbits, require bait for the snare, a different snare than a rabbit snare, and have a fairly high chance of stealing the bait and ruining the snare, and predate upon the rabbits, where when they get a rabbit they pick it up and book out to their den (which would be very cool to observe!) Finally, mountain lions... though I think that there should be maybe two or three across the whole game. They should be masters at stealth, and stalk silently. They'd be predating upon deer and rabbits and foxes and maybe wolves (and of course, the player or players) that are isolated and only in the right terrain where there are trees our high rock outcroppings to leap from, and be willing to spend a fair amount of time getting hunting conditions just right to attack. This can be quite time consuming for the mountain lion but they're definitely quite willing to do this in nature so it makes sense. If the player is the person being stalked a mechanism needs to be worked out that the player can notice to see that they are being stalked... don't really know what that would be quite yet, but there's got to be something. If the player does not notice or has the bad fortune to just walk right into the worst possible spot then a lion struggle ensues. This should be close to deadly (lose 75-90% condition so if you're already down it's bye bye time) and have very serious clothing damage with very real possibilities of losing multiple items of clothing, and a long recovery time because of lacerations, bleeding, etc (multiple bandages required, and some affliction that means the player needs to rest for a good five days before being back to top shape... heh, cat scratch fever maybe curable with a course of antibiotics or reishi tea taken over multiple days). Unlike all other animals, mountain lions should be able to migrate between regions too.

The problem with all of this is modelling this stuff can be computationally intensive... in programming terms you need to create an agent that represents each animal with behaviour and for each "tick" of the game clock each agent needs to be assessed and to behave as its programming demands, and furthermore those will have higher order effects as animals react to the presence of other animals as well as the player... and in order for this to work it'll have to happen for all the animals in the region the player is in for each tick of the game clock.

There's mention on the job postings at Hinterland of someone having "innovative ideas for coop gameplay". The way I could see this is with a client server model. Allow up to four players per game across all the regions, logging into a game server that maintains game state. Since the server is not doing any graphical work (that's all on the clients) then it has more resources to actually model all of this animal behaviour. The server should have some kind of voting system that allows the players to vote on when to suspend the game for later resumption when everyone is available to play. The big thing that's problematic is time compression... sleeping's not quite so bad because presumable that'll largely be happening at night but crafting and so on is a tougher nut to crack... it's a bit rough for someone to set their player to a five hour crafting session and then have to wait five hours of game time before they can really do anything because the other players are out doing stuff. Part of the idea with a coop game like this is that the players all start out in different places, with the idea being that they survive their initial phases, get together (have to find each other!), and then themselves can cooperate in activities like hunting, crafting, and so on. Safe houses with multiple beds become significantly more valuable (hunting lodge, camp office, mountaineer's hut, etc), and can result in the group splitting up to do things like send an expedition to TWM to raid cargo containers over a few weeks of game time while another group hunt bear in say ML with the goal of crafting a full set of bearskin bedrolls for all the members of the group.

The idea here is that a coop game is a long term investment on the part of the players requiring multiple sessions on the server. There are a number of thorny issues to crack (big one being time compression, but other ones are... what happens when someone dies? Do they respawn into the same game? I'd think that it'd be important to not have this be the case because if someone dies when they respawn they'll know exactly where to go to find their compatriots... and can get tricked right out in no time once they find them... this has obvious problems from a realism/gameplay perspective). Also, what do you do with players that get into games strictly to screw around with the other players (i.e. trolls)? PvP is not what you want with this game. Set up a multiverse where when you croak in one game you can turn around and log into a new one that's running that's now down a player? Have a player reputation system so that trolls can get marked as such and have existing players on a game instance reject their application to join? How do we manage this multiverse to allow this? How do we set up a network of game servers that will allow people to join? What about latency and how it affects gameplay? All these are tough questions that'll require serious development to deal with.

I've actually got some refinements to this idea but need to give it some more thought... plus I'm at work and if I hammer on this I'll be limiting my job survival prospects.

[Hotzn]

On 4.7.2018 at 5:19 PM, stratvox said:

I've actually got some refinements to this idea but need to give it some more thought... plus I'm at work and if I hammer on this I'll be limiting my job survival prospects.

Very good ideas, but my feeling is that it would be incredibly demanding to make all of them a reality. I am not a programmer, but I would think that simulating behaviour for animals will consume calculating power. And the more complex the behaviour is and the more animals are simulated simultaneously, the more the game is slowed down. I would love to see wolf behaviour being driven by different factors, like for example hunger, pain, aggression, fear, individual personality etc. So every wolf would act differently, according to its personality and the individual situation. High hunger and aggression would likely result in bold behaviour and ultimately attack. But by throwing the wolf a piece of meat, its priority would be to eat, and eating the meat the individual wolf's hunger would be lowered and the likeliness of an attack accordingly. Or you could shoot the rifle in the air. The sound of the gunshot would raise fear for all wolves in the area, and if fear exceeds a certain threshold, they would flee. And so on.

You would be able to "read" some of the wolf's intentions and react accordingly. Does it growl and head for you directly? Better drop some meat. Does it growl and hesitate? Maybe you can drive it off by throwing a well-aimed stone. And of course a wolf's parameters would be influenced by the presence of other wolves. So pack behaviour would come as a natural consequence. One member of the pack becomes fearful and runs off? All remaining members get more fearful. So depending on the situation, you might be able to drive away the whole pack by driving away one individual. Or... not.

Anyway, I digress. To sum things up, enhanced animal AI would be somewhere at the top of my wishlist as well. But I would think this must be introduced step by step and very carefully. Giving wolves improved AI could result in them becoming much more deadly overnight. Keep in mind not every player out there is an astute interlopist.

[StrayCat]

28 minutes ago, Hotzn said:

I would love to see wolf behaviour being driven by different factors, like for example hunger, pain, aggression, fear, individual personality etc. So every wolf would act differently, according to its personality and the individual situation.

THIS. 
Can't find the "overlike this superb post" button.