One of the fun things about the world in Farcry 4 is that events don’t feel as fake as they do in MMOs. For instance, when PQs were first introduced in Warhammer, they were new and fun. By the time Guild Wars 2 used the idea, it felt tired, old, and fake (especially when compared to that disaster of a manifesto). The events being marked on your map, the by-the-numbers repetition, and the odd insistency that these ‘random’ events need to be heavily scripted all leads to them being far short of what they should be.
To return to the Farcry example, seemingly at random Karma events will happen, whether it’s a fight between the two factions, a supply truck that you need to steal/destroy, or a messenger speeding on an ATM you can intercept. These aren’t major pieces of content, and if you want you can fully ignore them, but they also don’t feel as staged or as forced as the MMO examples above.
For instance, chasing down a supply truck can be lengthy if you have bad aim, and during that chase you might run into other Karma events or just a random enemy patrol, who will join in the fight and start shooting/chasing you. Or maybe the truck will drive towards randomly spawned allies, and they in turn will shoot and possibly kill the driver, or get into a fight with whatever enemies might also be chasing you. The major highlight is that these things aren’t scripted, so sometimes the Karma event is short and easy, and other times turns into a bigger deal, and those aspects are somewhat player-controlled (first shot killing the driver basically stops the event, while again bad shooting/chasing can seriously extend and snowball it).
All of this could still be taken further, especially in an MMO. While the Karma events are fun, imagine if those supply trucks didn’t spawn at random but instead because a faction actually needed supplies moved from a real point A to a real point B? If a messenger wasn’t random but not only had a reason to go someplace, but his actual message was real as well? If outposts traded hands to really push a war in one direction or another, rather than just falling because the story said it should fall now?
Way back at the start of the MMO genre, Ultima Online tried something like this with its living ecosystem. It never made it out of beta because the players killed everything, the chains fell apart, and it just didn’t work in terms of a fun, playable game. In terms of lost potential, I’d rank UO’s abandoned living ecosystem as one of, if not the greatest, losses to the genre (which should also tell you how sad and devoid of advancement the genre has been when its first big title is also one of the most innovative). But that was in 1997; its 2015 now, and we certainly have the technology and hardware to make what was impossible in 1997 very doable.
I also don’t believe what UO tried to do is actually impossible from a player-behavior perspective. Yes, in a game where killing stuff is needed to gain skills and loot, with zero clear negatives for such killing, people are going to kill stuff. But look at Skyrim for example; does everyone kill every NPC? No, because the game rules suggest (but don’t outright prevent) that you don’t do that, and so unless you are playing a very certain style (which has its own challenges, ie guard agro), you don’t. An MMO could be designed in a similar way, leading to a more living, working virtual world. If you discourage but also account for the outliers, and create a system that not only handles them, but actively supports them, it can work.
That nothing on the horizon is even attempting to do so is disappointing, but disappointment has been the hot MMO trend for at least the last few years, now hasn’t it?
Hardcore Casual 
I’m pretty sure Star Citizen talked about having unique NPCs that can be permanently killed (presumably in raid boss fashion).
But, I dunno. I don’t see the appeal of “living ecosystems” in MMOs. You say that no one kills all the Skyrim NPCs, but I guarantee that out of the tens (hundreds?) of thousands of people playing the hypothetical MMO, someone will get bored/trollish enough to do so. Just look at the sort of casual griefing that goes on in WoW when people kill the flight masters, trapping lowbies in beginning zones. Times that by a thousand when quest givers are permanently dead, or the zone overrun by high-level mobs that bored players spawn by killing everything.
Scripted everything probably isn’t the only answer it there, but whatever the alternative is, it will need to stand up to the wrath of a hundred XxArthasDKlolxX.
“You say that no one kills all the Skyrim NPCs,”
I said the exact opposite.
“does everyone kill every NPC? No, because the game rules suggest (but don’t outright prevent) that you don’t do that, and so unless you are playing a very certain style (which has its own challenges, ie guard agro), you don’t. An MMO could be designed in a similar way, leading to a more living, working virtual world. If you discourage but also account for the outliers, and create a system that not only handles them, but actively supports them, it can work.”
In EVE you can suicide gank basically anyone anywhere outside of a station. Has that resulted in the game being plagued by XxArthasKDlolxX kids making high-sec a complete mess? Nope, because the game accounts for those types, and makes life difficult for them, while still allowing for something like Burn Jita.
There are even better examples of people effectively setting up rules that prevent ganking, while always leaving it theoretically possible, than EVE. Real life, for one. The single greatest obstacle to this aspect of a virtual world is the ability of people to create a new alt/account, thereby getting around any countermeasures the developers enact. EVE does better than most games at treading that fine line, their greatest compromise is that the omnipotent concord enforcer isn’t very realistic. You have the marketing people wanting to lower the initial cost and sock it to the whales later, while incurring a monetary cost for making a new account after much bad behavior would require the opposite approach. It always seems to come down to virtual world long term design principles versus monetization, doesn’t it?
Much of the discussion around EQNext has been about the use of this sort of living ecosystem, with an underlying technology of the Storybricks AI. Some of their comments include the idea that their multiple servers will likely evolve in entirely different directions because of different player interactions.
I’m not sure how well this will interact with “and then the players killed everything” but certainly the sort of vision you describe is exactly what SoE has been talking about in regards to EQN.
I’m not sold on Storybricks. It’s nice that they are making an effort, but there is only so much they can do. There won’t be some revolutionary AI breakthrough, so it is really just going to be an interation on what we already have. That’s what I suspect anyway, I’d love to be wrong.
Problem with EQN is it’s being made by SOE, so even though the basic concept seems to be following my Sandbox MMO tab, SOE is going to be SOE and find (likely multiple) ways to screw it up.
At the very least, a game could use a “wallpaper bubbles” model of populations. Farm all the mobs in this zone, and while it might be barren for a while, spawns get pushed up in other zones, at least making the ravening mob move around.
Eve Online kind of does this with the way it spawns anomalies, which are sites which contain PvE content such as resources (mining, data/relic collection) as well as places where NPCs hang out for players to fight. Eventually one area has few, with other areas winding up having a lot of sites. The same goes for the asteroid belts. Mine a lot in one system and the belts begin to contain less ore to mine, with the asteroid belts in other systems growing in size.
I’m not sure I even want virtual worlds any more. It seemed like a great idea ten, fifteen years ago but even if all the myriad problems both of technical and human behavior were solved, would I really want to come home from work and spend hours in a genuinely unpredictable environment that accurately and convincingly operated as if it were really “there”?
It begins to sound worryingly like the video-game equivalent of experimental theater, where the audience and the actors all conspire together to create an experience. It’s a far cry (no pun intended) from mainstream entertainment.
It would be good for someone to make stuff like this but I’m not at all sure I’m the audience for it any more.
“It begins to sound worryingly like the video-game equivalent of experimental theater, where the audience and the actors all conspire together to create an experience. It’s a far cry (no pun intended) from mainstream entertainment.”
Experimental theater is the exact opposite of a virtual world. Normal theater is a preset narrative set in a virtual world, so is one step removed from a virtual world. Experimental theater is a loose narrative set in a virtual world, which is then set off on a highly meta journey through ridiculousness with feedback between different groups in a random manner, obeying no convincing or realistic rulesystems but operating in a feedback loop of irony and symbolism. It is 2 steps removed from a virtual world, plus it falls off a cliff and is thus 2 steps and a cliff-fall removed.
Elite:Dangerous is doing some of this living ecosystem thing (or claiming to). The various NPC factions have ships that actually travel around etc. But it is lacking virtually all of the mechanisms (guilds, grouping, most inter-player communication) which make an MMO.
I am excited by Elite:Dangerous. It seems to be have the solid foundations for a virtual universe. Social constructs like guilds and grouping are game-play mechanics that can be added later. It’s a promising start,.