The Big Annoucement of EverQuest Next

For those that don’t know, even if I dislike people, I’m a big fan of MMOs (some more, some less). Currently, I’m stuck with Guild Wars 2 and our relationship just improves every day.

Except that, last week, SOE made their big announcement of the features of EverQuest Next. And that shook my relationship with GW2.


Watch live video from sonyonlinetv on TwitchTV

Now, there are a lot of good tidbits there. Although what everybody is talking about is the destructible/constructable environment (if those are even real words), what caught my attention is the “emergent AI” part. Mobs that learn the type of players to avoid; mobs that have intentions and not scripts; those are things that really can change the way the game plays. Everything else is not so big, if you ask me.

But, as a developer (application developer, not game developer, let’s make that clear), there are things that make me wonder if they will be able to push that — specially in the emergent AI thing.

Imagine this: you have your servers; they are running the code to learn things; there are 500 players in a zone, all attacking different mobs and all those will have to be “taught” which of their attacks work and which don’t; on top of that, the server had to do the usual player movement/attack checking, to avoid hacks and do their RNG thingy.

Now… How much processing power would those servers require? Specially if SOE really uses machine learning (genetic algorithm, neural networks) instead of a single tally (“Kills by mages: 2 = avoid mages”). Sure, sure, the complexity of such system is as large as the number of variables in the machine learning but still.

Maybe suddenly, because of all that math, you won’t find 500 players in a zone, but only 100 ’cause the servers can’t handle that much accounting for more than that.

Unless they don’t really learn anything and just have a RNG to select a place for a scripted event.

Edit: There is one thing I forgot to mention: Not only you have the server load, but you have the client load to take in consideration. I can’t stop wondering how many computers today can run something like they are planning, with clothes and hair dynamics (not calling “physics here yet”) and all those particles floating around from the destructible environment. Sure, the game is a few years away and computer power tends to grow over time, but with the current stagnation of sales and such, one has to wonder how much more the IT world will improve from now on. We are not in the 90s anymore, when a single year mean almost double the processing power, double the available memory and almost 4 times more GPU power.

On top of that, you have the human factor. This is something I learnt when I was following ArenaNet and their development of Guild Wars 2. In the very beginning, there were no “renown hearts” of any kind, only dynamic events. But (and I clearly remember Eric Flannum telling this story in one interview) people would go around a lake with green smoke and looking green and everything, with a NPC asking for help and they would simply not stop because they were expecting to find a yellow “!” around (and that’s why now Guild Wars 2 have renown hearts that work mostly like a classic quest with some overlaying dynamic events happening from time to time). Sure, the MMO landscape is different now than what it was one year ago and maybe people are more receptive for events that don’t have “!” around.

But still, how long till people find how to mess with the AI? How long till a guild forces orcs to simply run in circles or to go kamikaze over tanks? That is, again, if SOE is planning on use a real AI algorithm and not a tally or event with a RNG location.

Thing is, if SOE manages to deliver a game with all the features they are promising, they can have my money. But, till then, I’ll keep my “skeptical” alignment.

Edit: On thing I didn’t mention, mostly ’cause I’m unsure about it, is the monetization of the game. They mentioned that you can use EverQuest Next Landmark to create structures and then sell them on the market for real money; if people use your structure, you still get some “royalties” from it. So there a way to earn real money in the game, much like you can earn money with Diablo 3. But I’m not sure if this is really useful in a way or it is simply a way to catch people in a “I’ll make money playing games!” fashion.