2 years later... the almost sad state of MMOs in the new era

Some new video devoted to the sad state of MMOs.


(the only real MMO launched in 2020 is... a 10-year old game now released internationally)

Neranja said:

But most of them are basically other games (FPS mostly) with MMO and/or RPG concepts bolted on. This was all the rage when everyone saw what a cash cow WoW was, and everyone wanted a piece of that pie.

Also, I don't get MMOByte's "we have no new games to play in 2020" whine. MMOs were always games who take the longest to make and a lot of cash, and the genre has always been slow to innovate. Also, Ideally MMOs are games you play over a long time.

Click to expand...

The problem is that most of the recent MMOs were short-term ones. It's all asian cash grabs that look completely copy-pasted: action combat, flashy effects, knockdowns (seriously, every game I see features the characters being knocked on the ground as soon as they don't dodge the attack), with gender-locked anime basic classes.

A long term MMO that keep players in? That's GW2 (old as fuck), FF14 (older than fucker), WoW (we all know how that goes), EQ (on survival mode from TLPs) and that's pretty much it. All other western MMOs are dead, so we're left with ancient as fuck games. People had hopes for New World, and we all know where it's headed now.

Kirun said:

I think this is likely the direction you'll see for a while, until procedural generation/MMO "tools" become cheap as fuck (we're getting there) and someone wows the genre as an indie. I hate to go back to the example, but almost the exact same thing happened with the 4X genre. You pretty much only had Civilization keeping the genre "alive" (parallels to WoW here) until a bunch of indie studios came along and revitalized it in a big way.

Click to expand...

Yea. One of my old dreams designs for MMO would be the "single server non instanced MMO". Use next-gen procedural generation to create a continent with 50 times the size of current WoW, and like 8000 static dungeons. You use a GW2-style event system rather than an on-rail story-driven quest narrative: you are in a world of Adventurers, not Heroes. You're in for the loot and power, not to save the world. Factions galore, achievements everywhere.

Then, since you have that much estate, and people like to level, use the "I'm bored, let's make an alt" as a gameplay loop. At max level, you can open up the "Wheel of Fate" at any time, which lets you reroll to level 1, but with your old gear and a choice of different classes or races that unlock based on your progression. You start as human, but you played a lot in the north, in mountains and stuff? Then you can remort as a Northman (EQ barb). Or a Yeti (snow furry). Or an Ice Dwarf. And you've use more elemental spells than magic spells, so instead of a Sorcerer, you can be an Elementalist. Then, you can head to a different starter location and level all over again, but with an additional racial ability, better starting stats and a slightly better profession, and see new dungeons and stuff. That makes rerolling an alt part of the progression. Or, instead, you can stay max level for a few months, and go raid to get better gear (and better achievements, which unlock other classes).

The designers would focus on class designs rather than trying to make some "compelling story" or a new currency system. The hook is that you start as a basic Human Sorcerer, and work your way to Ancient Eldritch Lich Arcane Warlock Magus. If you want to. If you want to be an Empowered Ethereal Manifestation Void Elementalist Grandmaster instead, you might be (I hear those liches tend to be KOS to most factions). Expansions simply introduce more racial variants and class specializations.

But all of that DOES require a powerful procedural generation to make a compelling, yet shared world. Which means it's not quite there yet, I think.

Neranja said:

This has some challenges from a design and technical standpoint:

• People don't want more of the same, but individualized content that doesn't "feel" samey. This is predominantly clear with WoW itemization past a certain point (aka spreadsheet loot), and players hate it. But this is also visible in things like Diablo-style ARPG generated levels. The key word is "interesting". While the WoW cloned microdungeons (basically noninstanced hole-in-the-ground) won't be there with procedural content, humans are really good at pattern matching. This is going to be a problem when the players start to see "the patterns".
• Players (at least a subset of them) are really in there for the history and lore of the world. If the place doesn't make "sense" to them, they lose interest. Some Youtubers have big channels on the WoW lore, even with how nonsensical and retconned it has become.
• Big world implies big spread, and multiple start points. You can't have the world feel empty, and at the same time you need to accommodate players who want to start their adventure together.

I think it is possible, but only with some advanced AI support. Especially the lore part. Storybricks (from what we heard) was an interesting idea that turned out to be basically a "fund our AI research" vaporware.

Click to expand...

That's why I said "advanced" procedural generation. We can actually generate stuff that feels different - the problem is that the generators at one point make weird/surreal stuff that screams "random generator" because it stops making sense to the human mind. You need to couple procedural generation with AI categorizers who can recognize when something is too outlandish to be useful and eliminate it. And AI categorizers of that kind require lots of training.

It would probably make a kick-ass middleware for MMOs though, once you have the AI trained. Click, get a procedural world a couple hours/days later, twiddle parameters a bit until you're happy, then start plugging in your mechanics and if you insist, stories. And voila. Indy AAA MMOs.

Lore is currently close to an impossible task for AI. You can make backgrounds, but not stories. It's usually good stuff - you don't make stories, people who spot coincidences make their own stories. But if you want spoon-fed lore, AI fails even harder than procedural generators.


The risk of big worlds is known. You need a minimum and maximum density of players. If you fail to get it, you crash. If you get too many, the complains about overcamping start streaming to your forums. That's why the server/shard/megaserver model is seducive: if there's too many people, it spawns additional copies, if too little, it culls them. But once you go instances, you don't go back (see the discussion in Project_N)

Neranja said:

The problem with current AI tech at the moment however, is that the models have to be trained, and the end result derives from the training data. So in a way we need a lot of input to train the AI, and the end result will be "more of the same." What I see in the near future however is a form of AI-assisted design, where creative people throw together requirements and constraints, and the AI generated a detailed model that can be further parametrized or fixed up by humans.

Click to expand...

Procedural generation "works". You can feed it rules, like, say "an inn is based with a large common room with bar+stools, plus storage, plus cooking room, then optionally..." and you get functional inns.

But once the procedural generator explores the space of "functional Inns", it WILL stumble upon things that are nominally Inns, yet when seen by a human will make him go "what did the builder smoke?". You can look at the procedural design of spare parts where the things made seem to come out of a mad engineer's workshop... but are absolutely functional (and usually lighter and stronger than the one machinists made)

That's the point where either the human designers step in (for smaller worlds) and say "no", or you need an "aesthetic" categorizer than can point them where the procedural generation failed (or even outright snip the offending inn and replace it with the next generation)

I think high level, world-level lore is quire a strange concept for AI, but low level town NPCs could be generated via AI. This is also somewhat a scaling issue: at the moment the towns in MMOs don't feel that alive, because no one wants to walk through a real scaled town that really houses a thousand or more NPCs. So what we currently have is a form of "make-believe", a gamified version of a city where only the interesting parts (from a player perspective) are visible.

Click to expand...

At one point, you realize that Stormwind is basically a small village, not the capital of the largest kingdom of a continent.

It would be interesting to see how a "real-sized" town or village plays out, and how players react to it.

Click to expand...

If you want a real town, you need to devote... well, all of Elwynn and possibly Duskwood next door to it.

Although, once you do that, you can safely host a thousand players going on their daily business before heading out to adventure, I think.