How Low Can You Go?

When World of Warcraft launched in Europe, 80 servers were installed to support the 290,000 players who signed up for its first weekend. That's 3,625 players per server.

What if they'd had 800 servers instead of 80?

Modern commercial virtual worlds generally handle between 3,000 and 5,000 players per server. Players may say that this is what they want, but given half a chance they'll flee to instanced sub-world bubble environments so as to avoid contact with the seething masses. If the virtual worlds themselves had more shards, each with fewer players, there wouldn't be this kind of pressure to have pocket universes for friends-only play: the whole virtual world would be exactly that, only persistent instead of temporary.

Having more servers with fewer players has other advantages, too. It reduces the effects of farming (try selling gold from server #401 to someone playing on server #682), it makes server recycling easier (merging low-use servers and reopen released ones for newbies), it makes getting away from personal griefers a real possibility (how long would it take them to find where you'd gone?), it allows for multiple server types (PvP, commodified, PD, whatever), and it gives players a sense that they can have a real effect on their virtual world.

How few players could a server support for players to feel that they didn't need instances yet for their virtual world to remain viable? Never mind whether instances are a good thing or a bad thing: how many players is the minimum necessary for a server to remain attractivee, and is this more than the number of players that makes instancing a requirement?

At GDC, there was a panel on Persistent versus Instantiated Spaces, and I asked the panellists the above question. Mark Jacobs was of the opinion that there was perhaps a sweet spot where it could happen, but he didn't say where he thought that might be. Raph Koster pointed out that text MUDs can feel like a world with as few as 10 players, but didn't say what the low-end figure might be for commercial graphical MMORPGs.

So what do you think? Is the critical mass required for a full sense of community greater than or less than that required for relatively hassle-free play? Indeed, are the two figures related?

Richard

Comments

[ren]

1.

Not sure about a specific number but I know that with just over 1500 people in ATITD things could feel very lonely as travel times were long and the virtual real estate was pretty large – so we probably need to take those two variables into account too.

[Richard Bartle]

2.

Ren>travel times were long and the virtual real estate was pretty large

So is it that the pressure to instance is related to population density rather than absolute numbers of players? If all the players in ATITD were to hang out in the same place and only go off into the wilderness for specific reasons, would that work?

Richard

[ren]

3.

Ren>travel times were long and the virtual real estate was pretty large

Richard > So is it that the pressure to instance is related to population density rather than absolute numbers of players? If all the players in ATITD were to hang out in the same place and only go off into the wilderness for specific reasons, would that work?

I’m not sure what the pressure to instance is based on. It may very well be independent of population density or sense of community (which is think is dependent on pop density type factors), given the option it might just be something that groups of a given type want to do. The urge to create a ‘private’ club seems strong generally.

[Jez]

4.

I'm no expert but I do play a lot of WoW; sometimes it can feel empty enough as it is, without having even fewer players. Because it's such a huge universe (IMO) and chat is restricted to your local area only, having fewer players on a server would make gaming a very lonely pursuit unless you already had a sizeable guild of friends to play with. I don't, I know few WoW players in real life. I recently created a character on a high-population PvP server because I was bored of late-night solo questing on my PvE server.. "Private" adventuring is all well and good if you already have friends but it makes it much harder to find friends. Then there's the fact that half of the players can't talk to the other half - Horde and Alliance might be on separate servers most of the time - and travelling around is pretty slow work too...

[RedWolf]

5.

I think what WoW really needs to balance out the population between servers is a character transfer system that lets character freely travel between servers of the type they play (RP, PvP, Normal). I'm sure I've heard of such a system before in other games, I think it might have been Shadowbane, where you can travel to another server and the name of the server of your birth is appended to your character name. A system like this will both allow for balancing the load (through queues, lockouts etc), increasing population during off-peak times and increasing the size of the community as a whole.

[bruce boston]

6.

"it allows for multiple server types (PvP, commodified, PD, whatever)"

We might also consider 'population cap' as a server variable. No reason we can't have both 10k player servers and 1k player servers in the same game.

-bruce

[eric]

7.

when i played daoc, anything less than 1200 or so felt really empty (and that's already pretty small, split into 3 realms), though i do understand that going into certain dungeons or places like darkness falls got really crowded.

on the other hand, it could get hard wrangling together enough high level people to do a big df crawl or take on epic mobs, so a higher overall population is useful in that you have more people of your level to interact with.

finally, especially in daoc, you could not fight against other realms effectively if your server was small. you couldn't find people to fight, or if the realms were not balanced you couldn't compete. In daoc at least, a high-pop server was more a blessing than a curse.

[Andres Ferraro]

8.

It will all depend on how heavy the interaction is given the game mechanics. Instatiated spaces will work with 10 players if we're all intearacting among each other. They won't work if the mechanics make it so that we never even see each other. I don't think there is a 'sweet spot' you can point at like that. There may be a 'sweet spot' in terms of the amount and type of interactions players want to have. But I don't see a 'server population sweet spot' unless you defiine what the game is.

[Thabor]

9.

The ideal number of players on a shard to remove the need for instancing would be somewhere between a full group and a single guild.. I think the drive to instance in only nominally driven by population.

For players I see the drive being uninterrupted access to content, and quality of experience. Instancing maintains immersion, because the dungeon hasn't been flattened by a horde of people. You don't have to wait for BossX to re-pop. The state isn't dependant on interaction by people outside your group, so difficulty is more consistent. None of those are directly related to population.

For designers the reasons are fairly parallel. You have realtively linear well defined place, where you can tell a story without having to deal with the usualy conventions of MMOGs. You have the ability to allow persistance for the duration of a session, instead of having to reset the stage every five minutes.

I think the number for a working community on a server is much higher typically than the cap you would need to approximate the same benefits without instancing.

[Yak]

10.

What if you could combine the two systems in a transparent way?

What does it take to make a world feel populated? Seeing people around me fighting/questing/speaking. Other than by text chat, do I really need to interact with them? Not at all. In face, if they couldn't "steal my kills" or "fight my mob", I'd feel even better (which is the beauty of instancing). I don't trade with them, I don't fight with them, I maybe /salute and answer "the slime mold with Freta's bag is over thata way. /point"

What if you had a shadowing process where players from other servers are temporarily overlaid... shadowed... onto your server? What if, while out killing slime molds in the empty Badlands, my server (#243) realizes that there are less than 10 players/square kilometer. In response, it combines the graphical inputs on a few neighboring servers that also have not enough density in that area (#819 and #104). (Of course, that's the non-techie description). The person I see might be ghosted or semi-transparent or have a green tinge so that I know I can't trade with them, but I can chat all I want and interact with them in space. If a few parties come in on my server at that area, then the population desity goes above the minimum level. When the ghosted dude/dudette leaves my screen (and I, correspondingly leave theirs), we both disappear from each other's displays.

During the entire process, I can't /tell him things, /trade, or do any other interactions other than straight spatial ones. You might see them killing mobs, but you can't help or kill steal, or anything.

The only downsides is that you might be able to use this to spy on people via PvP if you can predict what server you're going to shadow on and tell your friends... but you can do that anyway by having a similar-factioned toon sending messages to the enemy.

[Richard Bartle]

11.

Bruce Boston>We might also consider 'population cap' as a server variable. No reason we can't have both 10k player servers and 1k player servers in the same game

Yes, good idea, although of course the same world built for 10K players could seem empty to 1K players (or, if built for 1K players, could seem overcrowded to 10K).

Richard

[Abalieno]

12.

Modern commercial virtual worlds generally handle between 3,000 and 5,000 players per server. Players may say that this is what they want, but given half a chance they'll flee to instanced sub-world bubble environments so as to avoid contact with the seething masses.

Completely wrong premise. The players crowd the most popular servers where it will be easier to find groups for PvE and where it will be easier to find action in PvP. This is a *fact* and also the reason why World of Warcraft has HUGE unbalance problems where a few servers are packed and another big block completely empty.

The players move to small servers only when there are technical problems or when a brand new server is launched where everyone starts the "taxi to victory" from level 1 again.

The core of the question is about how the gameplay is structured and there's no definite general law to draw from this.

[Byron Ellacott]

13.

If it helps, the 'pressure to instance' for me is about eliminating competition for scarce resources. Content tends to exist in small regions, and without instancing also tends to be bitterly contested. Who wants to try competing with unemployed kids for the privilege of consuming the game's content?

[Sam Kelly]

14.

I find that for me, community comes exclusively out of social interactions, not spatial ones - other players in the same game area are an annoyance at best, unless they in trouble I can help with, or happen to be personal friends anyway. (Disclaimer: I don't do PvP, so my perspective is skewed that way. My experience is also mostly with large MUDs rather than even small MMOs.)

On the other hand, the people I see in the public chat channels and instanced club/guild/clan/whatchamacallit channels, I do feel a sense of community with - partly because I've chosen their company & a shared enterprise, and because I can leave the social space (or mute the channel) when I want to.

I suppose what I'm saying is that while it's entirely possible to get both a good sense of community and hassle-free play (anecdotally, it happens quite a lot), that sweet spot is going to be in a different place for every combination of personal playing style and world design (content density, amount of information about, & interaction with, other players, and spatial size, I'm thinking of particularly).

I'm just wondering whether it's possible to observe & cluster the sweet spots for various groups, and then make sure they have enough information to give them a fair chance at a rational decision about which realm to head to.

[SavageX]

15.

honestly I havent seena "comfortable" play experience with more that 2000 to 2500 players in a world. It seems if you go over that, the lag gets bad and the player quality gets exponentially worse.

[Mike Rozak]

16.

Is there a sweet spot?

If you think of a MMORPG as a treadmill that takes 400 hours to get through all the content, from level 1 to level 60 (or max), and you have 4000 players logged into the world at any given time, then statistically, each player would be 6 minutes apart on the treadmill. (WoW doesn't quite conform to this assumption since it's new, and most players are lower levels. Plus, players tend to group into parties.)

If each player is 6 minutes apart, then any quest that requires them to spend more than 6 minutes in one spot will cause them to realize the treadmill nature of the world. Players don't want to consciously admit that they're on a theme park ride, so 6 minutes may be a bit short. (In WoW it feels more like 15 minutes separates players in a quest. I suppose someone could sit at a quest site, like a harpy den, and count the player frequency.)

With only 400 players, characters are stastically 60 minutes apart on the treadmill, so they will almost never have a quest in common. (Quest grouping being the main impetus for talking to strange players in WoW.) At 60 minutes, players will only see each other passing on the road, or if they intentionally hang out in public places, like inns.

WoW has inns, but no one seems to congregate there... probably because socially and game-wise an inn is no different than anyplace else in the world. Chat is global (to the zone), making "meeting spots" pointless. PCs don't get any game advantage from sitting at an inn either. Plus, there's no reason to chat to stangers unless you need to quest with them, but then they're already at the same quest sight as you. (The trading house are more crowded, but I could care less whether I have the best gear, so I don't bother.)

The result is a lot of PCs running around. Most strangers don't chat unless they happen to be doing the same quest, and half of the time players just invite you into a party without saying a word. (Or at least that's my cynical impression.)

If a world has 60 minutes treadmill between players, is this signficantly different than a private-dungeon world? I suspect the experiences would be almost identical.

So, for WoW, the sweet spot is someplace between 400 and 4000. If the world were more player-vs-player or player-with-player, the sweet spot would be a higher player density.

[Richard Bartle]

17.

Abalieno>The players crowd the most popular servers where it will be easier to find groups for PvE and where it will be easier to find action in PvP.

But once they've found the groups, they then don't want to play with anyone except those groups?

Hmm, maybe the thing to do would be to take Bruce's idea of different population sizes. You go to the souped-up 10K server to meet your friends, then decamp to a 1K server to play with them. The anti-farming properties could be maintained by allowed players to move characters from a high-population "seed" world to a low-population "play" world.

Richard

[Richard Bartle]

18.

Mike Rozak>If a world has 60 minutes treadmill between players, is this signficantly different than a private-dungeon world? I suspect the experiences would be almost identical.

In terms of time-between-quests, yes, it would be similar. In other respects, though, it would be different. The main thing about instanced content is that it is temporary rather than persistent, so it's like you're no longer playing a massively multiplayer game, you're just playing a multiplayer game. It's like everyone got aboard the cruise ship, hooked up with a few people, then spent the rest of the cruise going on speedboat rides. The virtual world becomes a glorified lobby for meeting potential playmates. While there's nothing wrong with that per se, it does mean that the very special and particular experience that can only come from a MMP game is cast aside in favour of a MP experience that is never going to be as good as in games specifically written to be MP (rather than MMP).

Hmm, I'm in danger here of breaking my own rule of not focusing on the pros and cons of instancing and non-instancing, so I'd better stop!

Richard

[will dieterich]

19.

It seems to me that people tend to look more for a middle populated server. They want a low population server so that camping and log in problems are not around, but also want a high population so thier are plenty of people to group with. Then once grouped they want to be off on thier own so they don't have to worry about other groups messing with thier play.

However we will get to see an example of the smaller, huge amounts of servers later this year with D&D online. IIRC they are planning to have a central city for each server, and each server will consists of just a few hunderd people. They say they are tring to get back to the feelings of MUDs. Will be interesting to see if they can make it work, and if people will accept it.

[LEKO]

20.

I play WoW since day 1 and I never played an MMOrpg before (except 1 month of DoaC and it was painful). From my personnal experience, I find that WoW is pleasantly crowded. I usually find at least 1 or 2 people to quest with. Of course, sometimes it takes 10-15 minutes to find people, but usually I end up playing with someone and I do really enjoy WoW teamplay.

The most obvious downer for me in terms of server population is the fact that where all those Characters lives? Where are their families? Tu sustain that huge population of fighters, the world should have a much bigger NPC population. I would like to see more farms, houses, marketplace, etc... With a lot of NPC's. Imagine you enter Thelsamar and have 3-4 children that starts running after you and salute you... Or a few NPC starts to gather around a level 50+ warrior in Goldshire because they don't see them often. It would be great to see NPCs offering you food or potion once in a while. Another good thing would be to have to train yourself in a new ability. You want to learn that special move... Ok, pay me and I'll show you how to use it... The NPC would move with you and would ask you to bash 5-10 times that "doll" to learn the ability or would ask you to cast a spell on a couple of critter roaming around... This would make the world more alive!

Another good thing to make sure people gather and group, would be to have some quest that would require 2+ people to accomplish. NPC would tell you to come back with 2 partner to go and kill that ELITE MOB and the mob would drop 3 different items that would complement each other and the quest would be considered done when all the 3 players would return to the NPC. This would promote team play and gives a good challenge to party and this would encourage player to stick together instead of join/disband all the time.

This kind of stuff would make the world more immersive and would partly resolve the too much / not enough players issues.

[Arnold Hendrick]

21.

Most of us know that instanced content came out of the huge problems with "camping" spawns, in dungeons or otherwise, in EQ. Even when far friendlier loot-sharing systems and anti-camping provisions were pioneered in DAoC, there was still a problem with rich spots (like Darkness Falls) being overcrowded, which led to many frustrating experiences.

However, crowded hunting spots also made it easier to find groups and make acquaintances, some of which became friends.

If you want to experiment with low-density servers, I suggest visiting the lowest population servers of some fading game and see what it feels like. Horizons or AC2 (prior to the new updates appearing) might be good choices.

You might also consider trying out EQ2, since outside of two central cities, every single zone is instanced in an attempt to keep population low. As a result, meeting a casual acquaintance more than once or twice is unlikely, since even if you stay at the same level, you're unlikely to end up in the same instance very often. This means that you must actively pursue friendships since casual contact no longer will suffice. Since EQ2 strongly rewards group over solo play, this is a strong incentive to make contact.

WoW is peculiar because it has little instanced content, and rewards solo over group play until the later stages of the game. I've been tracking WoW server loading, and also observed L2 server loading in some detail. It's clear that PvPers like to congregate, probably because nobody wants to give up hard-won levels and go to a new server where they're more likely to be gank-bait. However, in PvE and RP, people prefer lower density servers because they feel there is less chance of the choicest spots being camped by someone else.

The sweet spot in all this will vary with each gamer's tastes and preferences. From a business standpoint, a good MMORPG product today probably wants to maximize populations on as few servers as possible to reduce complexity (especially in support), help PvP play, and minimize the need to shut down servers later (and all the associated problems of land rights held by people who lost their server). The problem of camping and crowding for PvE play is best addressed with instances, which must be carefully strategized to minimize the problem of feeling like you're playing a multi-player game instead of a massively multiplayer game. Instances that are "raid sized" rather than "team sized" might go a long way to solving this problem. ((Note: while currently WoW supports raids, the sharing logic is so discouraging that nobody in their right mind organizes a raid unless they absolutely must do so.)

[Darniaq]

22.

Richard Bartle wrote: You go to the souped-up 10K server to meet your friends, then decamp to a 1K server to play with them

This is essentially the experiential core of instancing. You gather your group in the persistent part of the world and then go off and hunt in the instantiated one. Guild Wars will feature this most overtly, but both WoW and CoH push adventuring in private made-to-order content. And, EQ2 instances almost the entire game except for city zones.

I don't know that this needs to be formalized by server architecture, except if it saves the company money. It allows for better server load balancing, but the game play experience is roughly the same as controlled instancing.

Whether instancing itself is a good thing is based on the implementation. Most of them are, as you note, mirrors of the persistent content. Instances are as repeatable (for the most part).

Microsoft's cancelled Mythica project was the first and only instantiated game I know of that actually treated the instances as persistent (or at least, that was their goal): You break a bridge on your first time through and that bridge was still broke the second time (the game was using Havok physics, which make these sorts of dynamic modifications a bit easier I imagine). They hadn't finally implemented all of the nuances though before the project was cancelled, but this was an intriguing idea I wish others would have picked up on.

[Capt_Poco]

23.

Yes, but the lower pop worlds are still places where meeting
casual people and teaming up with others to kill a powerful boss and
etc occurs. It' s still a MMORPG and even though
it' s a lower population one. I think that people
should be able to choose what instance of a dungeon they want to go
to. Like Diablo. You would be able to "
host" an instance with your group and or you could
join another instance. Prehaps attach some in-game cost to
hosting an instance. That way and people could find the big
instance groups more easily and if they wanted to and could still
isolate themselves from all the ninja-looters.

[Russell Conner]

24.

On the other Really Small end, is ROE.

Primary servers are about 100 to 150 players.

The smaller RP server runs 15 to 25. Amazingly I know just about everyone on the RP server.

We have entire server raids... where most if not everyone on the server shows up to take on a boss. Everyone is invited regardless of level.

[Bruce Woodcock]

25.

I would say the minimal number is directly related to the size of the world. Remember that the initial UO map was designed for much fewer people on at one time, so instead of wall-to-wall housing you'd have a much more open feel.

The other issue is one of timing. When you first launch, you're going to have X number of people jamming your server with peak ratios maybe half your total subs, and your newbie areas are overcrowded. Within a few months you settle down to a more reasonable ratio, say 20% of total subs, and then 3 years later you may have lost half your subscriber base and you're down to 10%.

So you can design your world to handle 3,000 simultaneous, but it has to handle it when its only 300. But if your launch goes poorly, you could be seeing half those numbers again, and that's almost impossible to plan for.

Anyway, like I said, it really depends on the size of your world and the size of the zones (if you have them) therein. The key is not simply how many other people are online (which matters for things like text chat, auctions, etc.), but how likely one character is to run into another in the graphical world. If your world is small enough, I'd say somewhere in the 100-300 range.

There's some number in psychology/sociology that tells you how many people the average person can maintain contact and good knowledge of at once but I forget what it is -- didn't Raph give a talk where he talked about that, relating tribes to guilds and noting you can't get above a certain number without more structure? Wasn't it 200-something? That's probably related.

I know during my days on TinyMUDs that after about 100 people or so the public "hangout" areas really started to get crowded (which meant more than, like, 10-20 people at a time), and by the time you neared 200 people were already fragmenting off into their own subgroups and areas. (Those existed before, of course, but previously subsumed within the larger crowd.)

Bruce

[Ola Fosheim Grøstad]

26.

Arnold Hendrick> Most of us know that instanced content came out of the huge problems with "camping" spawns, in dungeons or otherwise, in EQ.

Hmm... Actually I have reasons to believe that it came out of a desire to provide new gameplay (vision), then it was used to solve crowding issues (pragmatic).

[dmyers]

27.

There's some number in psychology/sociology that tells you how many people the average person can maintain contact and good knowledge of at once but I forget what it is

***

Roughly, 150. This appears to be a cognitive capacity rather than, strictly speaking, a physical/social one. I.e., seems to also be about the maximum number of *fictional* characters capable of being juggled about within various mythologies and/or within any more contemporary, live and ongoing (soapish-like) drama. Similiar (in its cognitive origin) to the seven-plus-or-minus-two thing [http://www.well.com/user/smalin/miller.html].

In City of Heroes:
max pickup group size --> 8 (7+-2)
max supergroup (clan/guild) size --> 75 (sometimes considered too small, but seldom reaches that limit)
max supragroup (Hami raid) size --> ~150

*Average* group sizes are sigificantly smaller than max.

dunbar's law -->
http://www.bbsonline.org/documents/a/00/00/05/65/bbs00000565-00/bbs.dunbar.html

related and interesting -->
http://radio.weblogs.com/0114726/2003/02/12.html#a284

[greglas]

28.

This is a great thread. I just want to point out the notion of the imagined community would seem to me as relevant here as it is in the real world.

http://www.nationalismproject.org/what/anderson.htm

In short, shards aren't just about avoiding the lonely undead warlock problems. I might only know of / interact with 150 people on a shard, and past that point we've only got people I won't have any meaningful interaction with. But that doesn't mean that these larger groups are not significant. Just for starters, there's something aesthetically pleasing (or displeasing) about simply *knowing* that the possibilities of interaction in the shard extend to thousands of people. Beyond that, remember how the net served to bring together people (e.g. game researchers) who previously only spoke to each other, if at all, through publications in printed journals. Bigger shards = more niche groups.

Julian started an earlier thread, btw, on some of this stuff, suggesting shards were on the decline -- interesting that Richard seems to be pointing out the benefits of greater slicing and dicing?

http://terranova.blogs.com/terra_nova/2004/04/whats_in_a_shar.html

[Philip Tan]

29.

Hmm, I seem to recall talking about this with you on thursday, Richard...

Is a large number of users on a single server needed to generate a feeling of community? Some player-hosted multiplayer online games can already have thriving 'communities' of discussion and interaction, and in the case of some online CCGs, item exchanges. Yes, it's not exactly the same thing, but I hesitate to immediately conflate direct avatar-to-avatar contact with online communities coming together.

If there's some maximum number of people that players can possibly track, Greglas' post reminds me that you're never sure exactly which person in an MMO server you're going to add to your mental (or in-game) list of contacts. But I'm not sure if it's the stated goal of any MMO to hit players' saturation points. If I can only find 5 people that I'm going to enjoy interacting with (possibly asynchronously) on a server of 100, I may still end up enjoying the game enough to pay subscription fees. Indeed, it may be more emotionally difficult for me to ever leave a small clique of friends.

However, based on some of my attempts, I suggest a smaller population may require designers to come up with quests, roles and mechanics that allow average players to feel that they're having a much greater impact on their world than in most current MMOs, and this would result in more production time. I don't mind being a wheat farmer or random monster-killer if I'm a small part of an world of thousands. But if there are only 50 people on a server I'd expect to be one of the 50 most important people in that virtual world, and quest designs and mechanics have to reflect that.

[Mike Rozak]

30.

Richard Bartle wrote - The main thing about instanced content is that it is temporary rather than persistent, so it's like you're no longer playing a massively multiplayer game, you're just playing a multiplayer game.

How is a single-instance with monsters that respawn after 10 minutes any less temporary than instanced content?

A world with low population AND where strangers don't have a way of meeting up seems to turn into a single-player game too. If strangers can and are encouraged to meet up, then it's more like the cruise ship you described.(Below.)

It's like everyone got aboard the cruise ship, hooked up with a few people, then spent the rest of the cruise going on speedboat rides. The virtual world becomes a glorified lobby for meeting potential playmates. While there's nothing wrong with that per se, it does mean that the very special and particular experience that can only come from a MMP game is cast aside in favour of a MP experience that is never going to be as good as in games specifically written to be MP (rather than MMP).

I've been pondering this issue lately and have been thinking similar thoughts, although not with the cruise ship analogy. Both instancing and low-pop with rapid respawns weaken the potential of a massively multiplayer game. They turn the experience into a single-player game that's inferior to a real single-player game because your PC can't ever change the world; at least in NwN or Morrowind your character's actions are permanent and affect the world.

[Richard Bartle]

31.

Mike Rozak>How is a single-instance with monsters that respawn after 10 minutes any less temporary than instanced content?

I'm not a fan of unexplained respawning either, but at least it's always there to respawn and be fought over.

Richard

[Michelangiolo]

32.

Why do these games need to exist in worlds at all? I mean, sure, all of them start off with a few continents or main solar systems, or whatever the genre calls for, but why that big? There is a lot of adventure to be had in our solar system alone (heck, Saturn and its moons are interesting and numerous enough to carry a space-base colony/mining/fighting game). Genre-wise, look how extensive the Arthurian or Norse legends are, and they barely covered what would amount to counties, geographically. Not every setting needs snow, lava, water, desert, and plains levels. Seasons that change once a month or so, and a basic understanding of geography could offer most of those things.

I like the idea of a centralized city or meeting point where all players gather, based somewhat on the MUD / D&D idea. Everyone goes to Waterdeep because that’s where adventurers go. The wilderness surrounding it is just that, wilderness. Any pioneer / explorer / tracker knows it’s foolish to head off in one direction and keep going because there’s really not much out there to find. Maybe a few farms close to the city, but beyond that, only the occasional encounter and lots of trees and grass. Why so boring? Well, you’re a hero. Heroes do things for people and earn renown.

So, the central city offers numerous missions for all classes, both inside, underneath, and outside the city. As you level, you gain access to different persons and locations within the cities (with fame comes access to the best pubs and people, after all). The key is, people recognize you as a ne’er-do-well, an adventurer, a mercenary, a hero, a legend. That’s what it is all about, isn’t? WoW, for one, boasts that it has 2000+ missions, but I’ll never find them all, nor would I be eligible for them all, but even if a few NPCs had to offer a few missions on their own, a city with 1000 ( a high and unnecessary amount in my mind) to 2000 NPCs shouldn’t be that hard to realize, and it would definitely make a city seem bustling.

That was a long lead-in, but think of this in terms of server load…the main city will always be populated because you have to both start and return to there, but everything else could be an instance. That isn’t to say that Joe Schmoe and his band if brigands couldn’t follow you into the wilderness to your dungeon, or down a given manhole, but unless they’re in your group, they won’t know what you’re there for. Sure, they kill the big troll before you get to him, but you only need his head as proof, you still get the mission rewards. Better yet, I’d love to see an NPC say, “Well, I grant you put in some effort (here’s some xp and money), but from what I hear, you were beaten to the punch, fortunately a similar situation’s arisen nearby; try to make sure you’re not followed this time.”

Let parts of the world come into existence as missions are taken (there are all sorts of ways to do this with manholes, sink holes, broken walls, ruins exposed by weather), and let them fade from history after the fact. This does tend to happen with City of Heroes more than others, but as players get higher up on the food chain, they spend less and less time in Atlas Park in places where lower level players will never see them. To what does that leave many to aspire?

Neatly put, gather everyone in one place, creating an atmosphere that begs socializing, instance missions that require teamwork from the outset and encourage people from multiple disparate levels to work together (i.e., as a high level player, you don’t *need* a lower level player to complete a mission, but it’ll take 30% less time if you bring one along…a squire for a knight, or a navigator for a pilot). Big city always (big load), countryside sometimes (low load), instances frequently (tiny load).

We have such a huge history of paper-based and text-based games to draw from, let alone literature and media, but so little of it is applied to actual game play. If I were making an MMO, the first thing I’d would be to collect all of the major gripes, everything that is boring, everything that frustrates, a give a real hard look at players’ suggestions for solutions. They’re certainly haven’t been many developers who have proven themselves right so far, and, for good or bad, you can’t make something massively multiplayer without at least somewhat catering to the masses.

Really, if Everquest could keep an audience for multiple (profitable) years, imagine what they could have done if it was truly fun instead of occasionally enjoyable (this goes for any MMO, before the outcry).

[Mike Rozak]

33.

Richard bartle wrote - I'm not a fan of unexplained respawning either, but at least it's always there to respawn and be fought over.

Sorry, I wasn't griping about respawning. I was just pointing out that respawning causes player changes to the world to be temporary, just as instancing does. Spawning doesn't isolate and protect players (from other players) in their own sub-world, while instancing does.

[Michael Hartman]

34.

Another factor that nudges people towards more populated servers is the horrible game design concept of "rare drops."

On a low population server, there are not enough people killing the right mobs to result in enough "rare drop" items making their way into the economy.

This is flawed game design from the beginning in my opinion, but flawed or not, it is a major part of games like WoW, EQ, etc.

When players know this, they know they better not go to a low population server or it might be weeks or months before something they really want will even become available.

[Thabor]

35.

While "rare drops" may be bad design, they are a largely player driven phenomena. Players want full and easy access to the "uberest" equipment while at the same time demending that it be difficult to obtain so that they remain unique.

[Michael Hartman]

36.

> While "rare drops" may be bad design,
> they are a largely player driven
> phenomena.

I think they are a laziness driven phenomenon.

What is easier:

A) Create 10 hours of content.

B) Create 1 hour of content, but force players to repeat it 10 times on average in order to "complete/win/find the item."

B is a lot easier.

As a result, EQ and EQ-like games go with B.

Disgusting, really.

[Michael Hartman]

37.

> demending that it be difficult to
> obtain so that they remain unique.

I forgot to comment on this.

Uniqueness through variety is infinitely superior to uniqueness through tedium.

When the only way players can be unique is to engage in a freakishly tedious endeavor (farming some mob for weeks or months on end), the game designer has utterly and horribly failed.

When there are so many choices available to the player that uniqueness can be obtained through creative choice, then you have a good system.

[Eric Random]

38.

I believe a "sweet spot" can be defined, somewhat, in general terminology; Not necessarily using an equation per se, but moreso as a reference of issues which are sensitive to population size. In that sense, it is a list of advantages and disadvantages of particular design criterion which have effects which vary with population size.

For example, in its simplest form: a server no longer allows players to log on when x amount of simultaneous player connections exist. This has no effect on a population much lower than x. Near or at x will cause players to search for less populated servers. Players like to play, and they will act in their own best interests. Intermittent ability to log onto a server is undesirable, and a player will search for a less populated server. This choice is affected by population size. Simple.

Players tend to enjoy mechanisms which allow them to group quickly with characters well-suited to their particular group dynamic. This group dynamic is largely dependent upon the dynamics of the challenges presented by the content and the limitations of the mechanics, which is highly variable across games. The player tendency exists, nonetheless. A player will have higher probability of locating well-suited characters quickly in larger populations. The greater the population, the more likely. The lower the population, the less likely.

Player-based economies are strongly affected by population size. The greater the population size, the more active and competitive the market. That is, the larger the population, the more goods and services are supplied and with greater variance, and the greater demand for goods and services offered. One may argue that supply and demand scales to population and thus balances out, but it is the loss of market variance and competition where operators in small markets suffer. Small economies tend to have limited offering of goods and services, respond slowly to market changes due to higher costs in marginal adjustments, and as such, small increases in demand can induce large increases in price as suppliers are less able to adjust output. Further, game mechanics may require a hierarchy of industry (one must have miners and mines before blacksmiths, and blacksmiths before armorers and weaponsmiths, and weaponsmiths before weapon retailers). Small fluctuations in the lower levels of industry can scale to drastic effects in small economies. Intermittent player involvement, a basic property of VW's, compounds the issue. Consider further that the game premise is the same regardless of population size, as are the challenges offered by the content. Immature and contracting economies may increase the difficulty of experiencing the content (not having access to needed equipment) and this affecting the time is takes to consume it. Variance across servers will have players shopping for more active economies, and perhaps even altogether opting out of immature and contracting economies, compounding the problem.

There are numerous other design decisions affected by population size, population variance, rates of change in population sizes, and population densities. Since persistence vs instantiation was raised, I'll focus on that. I don't believe that design decisions on whether one should use persistence or instantiation are based on population size. In it's simplest sense, it -is- based on a solution to content scarcity, which is affected by population size, but it's more than that, as scarcity can be scaled to population size. Most often this is accomplished through increasing the content. With instantiation, there are two variables at play, quantity and control. Quanity in the sense, there are 10 players wishing to explore caves alone, but only 5 caves. Simple, make 10 caves, which can be done either persistent (there are always 10 caves) , or instantiated (there are 10 caves when 10 caves are needed). Quantity is affected by population size. Control is not. Control tends to not be in persistent content, but tends to be in instantiated content. That is, once instantiated, access is managed by the party which instantiated it. This element is effective regardless of population size. Using persistent content without access control, a single additional player outside of the group negates control. The control element is important, as it provides both the player and the designer with greater control over the gameplay experience (ie. player narrative). Further, I reject the notion that instantiation limits the player's ability to affect the world, as that is an element of game design, and not a property of instantiation. The difficulty more lies in the ability to tie such a mechanic into the world fiction, but I find it more of a difficulty rather than an impossibility.

The range in server population size should be thoroughly considered during design and technology discussions. Although later testing may occur on server and client bandwidth, at both process-level and network-level, it doesn't -seem- as if such ranges are given considerate attention during game design. The main question for me is, really, just how high or low the population can be in direct relation to the overall expectations of a design, and not disconnected metrics such as server and network loads which can be adjusted to scale.

[Jim S.]

39.

Richard wrote:

>So is it that the pressure to instance is related to population density rather than absolute numbers of players?>

More important is the game's layout, specifically the "bottleneck" locations. How crowded those places are depends on the number of players that need access to that area at any given time. The playerbase size produces the traffic, but the layout determines how well that traffic is handled.

Another factor is the size of groups. If groups are expected to average at 5, obviously the area will be a fifth as crowded as if it were a solo game/area. There might be equal numbers of players in both cases, but the number of units of players is smaller with bigger groups.

[Kanaric]

40.

Hmm.... that guy Arnold Hendrick who posted. I wonder if that was the MIA former Microprose game designer with the same name who did Pirates Gold and Darklands? (plus a lot of other legendary games)

[Raph]

41.

Interesting writing on Dunbar's Number from Chris Allen, of Skotos, with attention on its applications in social software:

The Dunbar Number as a Limit to Group Sizes

and

Dunbar, Altruistic Punishment, and Meta-Moderation

[Tanner]

42.

I think that your site is good!

[Maudie]

43.

Great, continue like that!