What’s the ideal number of players to have online in the same instantiation of the game world at the same time in a multiplayer game? Obviously, this depends on what kind of multiplayer it is, but the spectrum of answers doesn’t really go from 2 to 10,000 the way you’d expect. Fundamentally, there are four different sizes a multiplayer game can be, and while I think the smallest size could possibly benefit from being split up into two or three different chunks, I also think the larger size might actually be a bad idea that was only adopted because it seemed like a good idea at the time.
Militaries across the world and history have had widely varying unit sizes, but once formation fighting began to decline and the number of guys you needed to stand in a row in order to block a mountain pass was no longer a deciding factor, everyone pretty quickly settled on a very similar optimum setup. Not only that, but you can see this optimum setup come up in ancient and medieval militaries whenever the concerns of formation fighting didn’t present a larger concern. This is because humans are hard-wired to work in groups of a certain size, which means the squad, the platoon, and the company show up over and over again throughout history (larger units have less to do with the fundamental psychology common to nearly all humanity and more to do with how the bureaucracy of your specific organization happens to be structured – in the case of translating this to MMOs, the bureaucracy is non-existent, so we don’t care). The squad of eight-ish, the platoon of forty-ish, and the company of 200-ish show up again and again, because eight-ish is the number of people who can stand around and have a conversation with each other, forty-ish is the number of people who can all work on the same thing without needing any layers of bureaucracy to stay coordinated, and 200-ish is the number of people who can maintain social bonds with one another.
Military organization gives us the most striking and easily observed example of this, because military organization explicitly splits people up into different groups of a specific size, which makes it easy to track how many people were in what units even thousands of years later, whereas it’s much harder to know after the fact what the size of cliques and parties was in the 18th century court of the Qing Dynasty or whatever. I use the military units as a shorthand, but this isn’t unique to military-themed games or even to games that involve combat. The basic pattern is a result of how humans socialized with one another and repeats everywhere.
You can have squad, platoon, or company-level games, and then you can also have games that are bigger than that, at which point automatic human socialization can no longer scale up to your total player population and you either have to impose a bureaucracy or just accept that the game is going to be a chaos of people not in any real communication with each other taking action with basically no concern for how anyone else will be impacted. Also, the company level can’t really be 200 people because your players have families and workplaces and stuff taking up some of the number of people they’re keeping track of. You can’t actually do a thing like real militaries do where you pluck someone out of society, put them in a barracks, and now the only people they interact with on a day to day or even month to month basis are other people in that barracks.
In a squad-level game, with 2-8 people in communication (whether that’s two squads of eight on opposing teams or an eight-way free-for-all with a voice chat or an eight-person co-op or what), everyone can always talk to everyone else, and once everyone knows each other’s voices, you can keep track of what everyone is saying to everyone else without any real organization. Every now and again there will be a multilogue where people are talking over each other, but people will automatically realize this is happening, stop, and you’ll even commonly hear statements like “Alice was saying something but I couldn’t hear it, what were you saying?” Even when people can’t hear over the multilogue, they can still hear and keep track of who’s talking, and you can untangle the knot in the aftermath. And we’re all so used to doing this kind of thing that we do it automatically.
I mentioned earlier that squad-level multiplayer games might be able to be broken down further, because I think there’s some visible gameplay differences between having two, four, or eight people on the same team. Not only does a one-on-one versus match feel very different to a four-way brawl, even a two-player and four-player co-op begin to take on very different feelings. As you get up to eight, it becomes much easier for a generally competent team to carry one or even two new players who don’t really know what they’re doing. Any number between two and eight-ish is enough to comfortably hold a conversation, but differentiating class roles and designing interesting levels seems to be much easier for four people than for eight, and it’s also a lot harder to wallflower when you’re one of only four people in a conversation as opposed to eight. You can have a conversation with eight people, but at that point the group has gotten big enough that usually you start having to fight for space a bit, speaking as soon as there’s a pause to get an idea out, and the conversation knots where multiple people are talking over each other and we have to untangle things afterwards start to become more common. We do sort of get a similar dynamic with higher levels of organization as well: 40-ish and 200-ish aren’t averages, but rather upper limits for the gameplay mood they deliver, and there will be a noticeable strain on the game as we reach them. Notably, D&D parties typically include 3-6 people plus a GM, which means they start at having four people in the conversation and stop one shy of having eight.
At the platoon level, the distinction between ten, twenty, or thirty people becomes less important, as what matters is that there are now too many people involved for everyone to be in the same conversation at the same time, but still few enough that everyone can keep track vaguely of what everyone else is doing right now and how they all feel about each other. You can’t have a forty-man raid team in the same voice chat without imposing some rules on who talks and when, but you can totally have a server with forty people on it wherein each person has a basic understanding of who does and doesn’t get along with who else in the server, and where you can usually find Alice or Bob when they’re in the game, and what those people are working on. Forty is really pushing it, and 20-30 is more ideal, but essentially this is the number of people who would get together and hunt a mammoth in the ancestral environment. It’s the size that classrooms gravitate towards.
You can use platoon-level as just upscaled squad-level, with lots of squads pursuing lots of objectives as in the MMO raid example, but I don’t think this is a good idea. Sure, it feels very epic the first time you do it, but it wears off quickly and pretty soon it’s just a bunch of admin for squad-level gameplay. Where platoon-level shines is in a group’s ability to engage in organic, long-term planning by splitting off into different squad-level groups trying to achieve different things. This is the MineCraft or the multiplayer survival game model (although MineCraft can scale both up and down from this level), where there’s about twenty to forty people on the server, and they tend to split up into groups of two to four or so, each focusing on their own goal but either helping other groups or antagonizing other groups or both, depending on what the game mechanics make possible or encourage. In most MineCraft servers, everyone splits off to build different mega-projects, but will come by to visit and swap and/or gift building materials to one another. In Ark: Survival Evolved, people will split up into little groups and murder people from other groups to try and seize materials or establish dominance. Either way, there’s a platoon-level community that organically breaks into smaller groups, which then interact with one another in a way that’s very comprehensible to everyone involved in each group. If you’re in group A and you don’t know how groups B and C feel about each other, it’s not because it’s too much to keep track of, it’s because you don’t care.
Company-level is dictated by the total number of other human beings that the average person can keep track of at once. In a company level game, there are way too many people for you to keep track of their individual relationships with every other individual on the server, but still few enough that after you’ve spent a few months getting to know everyone, you can recognize every other individual on the server. In small high schools, people commonly know most or all of the people in higher or lower grades. In larger high schools, they only know people from other grade levels who are specifically their friends or who share some kind of extracurricular activity with them, and otherwise only know people in their own grade. This is why.
In theory, company-level can be up to 200 people. In practice, all of your players have other social groups already taking up space in their head, which means you can’t have more than maybe 100 people in a company-level game. We don’t see a whole lot of games using the 100-ish person level, except for battle royale games, which are really a series of squad-level encounters being organically strung together. I think the ideal scenario for a company-level server size is actually an MMORPG, though. Your standard theme park deal where you run around completing quests and gathering loot and such. We’re past the point where people can easily keep track of one another’s goals and relationships with everyone else, but still at the point where everyone can recognize everyone else. This is ideal for a video game that is more like a place you can visit than a match you can win or a world you can alter. With this many people running around, it’s difficult to keep track of who might aid or interfere with your long-term goals, but in an MMORPG you don’t really have to care so much. You have a checklist of (hopefully) cool stuff to do, you do as much stuff on that checklist as you like, and along the way you meet other people who are also doing cool stuff and become part of a community. It’s hard to keep track of how everyone feels about everyone else, but you aren’t trying to do anything complex or map-altering, so it’s not as important to keep track of whose toes you might be stepping on with your project.
That brings us to massively multiplayer level, where you have over a hundred people online on any given server at any given time, often thousands, forming a server community of tens of thousands of active players. And I’m not convinced this should even be a thing. The most common result seems to be that most people just check out and play like it was single-player. When communities do form, they tend to be of a couple dozen people anyway. Whenever a group organizing 100+ people does cohere, it’s usually by creating some kind of bureaucracy to help manage it, and, really, is that a good thing?
Now, there are occasionally things done in MMOs that require very large numbers of people, and which can’t be replicated by just having a very large number of people in a Discord or lobby who then split off into smaller servers to play together. Like, City of Heroes has costume contests that couldn’t plausibly attract the number of contestants they do if there were only 100-200 people per server. Guild Wars 2 has spontaneous world events triggered by lots of people just happening to be in the same place at the same time, which can only happen on servers large enough that you occasionally just get like forty people in the same area of the game without anyone planning it in advance. I really don’t think it’s worth it, though, especially since things like the CoH costume contest example can easily be replicated by just having a special server for costume contests. People sign up, import their costume, attend the contest, then go back to their regular server when it’s done. Most people do this already, just importing their character to whatever server the contest is being held on, the only difference would be having a dedicated contest server instead of picking one of the regular ones.
The last time I speculated on the optimal way to run an MMO, I pointed out that it was kind of moot, because MMOs have very large up front development costs and are no longer very popular. Existing MMOs can keep the lights on with a very small player count sustaining them, but making new ones is basically out of the question. So, talking about how MMOs would be better off with a hundred times as many servers a hundredth of the usual size rather than the current setup of a dozen-ish servers each capable of hosting like ten thousand players, that’s all of mainly academic interest anyway.
But for the record, I’d really like a 100-person MMO server, and if we could get that four-hour day/night cycle I was talking about earlier while we’re at it, that’d be great. And could it also have a good terrain traversal mechanic, where you have to climb and swim and glide and stuff? Y’know, since MMOs are really good at being a place that you go to, having the terrain itself be more interactable would be really nice. Oh, and do the Guild Wars 2 thing where quests give rise to other quests in a way that non-permanently affects the map in a cycle, so winning quest A leads to quest B, losing it leads to quest C, and it all eventually loops back in on itself. And have the more action-oriented combat, too, like, with dodge rolls and stuff, and where the angle you hit an enemy from makes a difference. And a system whereby mob factions will fight with each other and with town NPCs, spreading out across the map and capturing towns until players drive them out, which could integrate into those quest cycle things I mentioned earlier, so, like, if the NPC orcs go unchecked for too long they’ll eventually invade the local quest hub and take over, and then the only quests available in the region have to do with driving them back out until the quest hub is liberated and things go back to normal. And there should be a vampire-themed class, too, just ’cause vampires are the best, with minion mastery gameplay, like, maybe a vampire with undead minions, or who can summon demons or something. And actually, can we have, like, a really broad set of potential powers like City of Heroes has, where each class has a primary and a secondary power selection and there are like fifteen options for each, allowing you to create nearly any power combination you can think of? And pair that with a really good character creator, too, one with lots of different options. Oh, and let the player decrease their own level if they want, so that they can face challenges they passed by on their way to the cap at the level they were intended for. And also could a percentage of the game’s gross income just get mailed to me as a check, without me having to worry about profits or put up any money for development or anything? Awesome, thanks.