From instancing to worldy games
Brad McQuaid has written a rather good article on instancing and the tradeoffs and choices inherent in it. In it, he divides up possible game types that use instancing into these types:
- A “pen and paper game, online” — essentially,a game that tries to recapture the charm and enjoyment of tabletop RPGing. The massiveness of such a project is the part to call into question.
- An “open big” game, which is apparently the way he thinks of WoW. In this model, you intentionally work with the business trend of the last few years, which is that games acquire the majority of their customers at launch, rather than going through a steady growth period that perhaps (because of network effects) develops into a “J-curve” of rising subscriptions. (Of current games, Eve, Second Life, and Runescape all show strong evidence of J-curve growth).
The difficulty with this sort of model has always been that if you fail to open big, you’re dead; this is the boat that boxed product is in now, of course. Big openings don’t stick, acquisition falls off, so you plan to recoup purely off the opening. In other words, retention becomes a less important factor — instead of being the thing that provides the most revenue, it becomes gravy. The game is designed to “end,” so to speak, and perhaps be resubbed to periodically when new content becomes available.
There’s attractions to this model, chief amongst them the avoidance of player fatigue. The downside, of course, is that opening big demands a huge investment.
- A hub-and-instances world, which he calls a hybrid of the pen-and-paper game and the MMO. This is, to my mind, what Guild Wars is, and it’s a popular meme these days. I tend to think of these as being very like Diablo, because the amount of gameplay in the “lobby” is limited. It’s like Quake if the server selection screen looked like a Quake map. They are, fundamentally, session-based games with a degree of persistence slapped on top.
To my mind, these games lie at the edge of what might be considered a virtual world in the first place. A lot depends on what degree of persistence is offered. HoloMUD, which I have cited before, was rather of this model in many ways, and the way in which I tended to regard it was to evaluate the lobby alone by my criteria of whether it was a virtual world.
Brad states that the folks who want to make this sort of game are
…mostly people who got into online gaming very early, in the old pay by the hour days, working on commercial games (not MUDs) and what attracts them to online isn’t necessarily what attracts the conventional or modern MMOG player. There’s not necessarily a yearning for a vast, shared, persistent virtual world, a complex economy, or any other cool or esoteric Kosterian theory or mechanic. No, that’s not necessarily what they’re looking for; rather, they just want to play with some people, a smaller group, and to have fun, likely in a more linear or scripted manner… they want to play D&D or an old school single player RPG, but with their friends. And they want it to last. (And then there are those who claim to want it all, to have their cake and eat it too – not sure what to say to that crowd). And there’s nothing wrong with all of this, outside of, IMHO, three things:
1. They keep calling themselves, marketing themselves, as Massively Multiplayer Games, which I think is misleading to the consumer.
2. They have some very serious design hurdles to overcome in order to create the amount of varied and interesting and preferably not-repeatable content I think they’re looking for.
3. And lastly, and this is a small subset of them, but it seems like the more vocal proponents of this sort of online game often times actively resent traditional MMOGs and their players – ‘catasses’ is what they call us.. I’ve seen things said like ‘we were first to make money online’ and ‘MMOG developers are just MUD kids lucky enough to actually be paid’. I’ve also seen things posted and said about how it’s only masochistic people who play traditional MMOGs because of their tendency to grind at times, to incorporate ‘ground hog day’, and the fact that MMOG players are willing to put up with griefers of any sort, to any degree. I say we call a truce as soon as their marketing folk stop calling their games MMOGs – call ‘em whatever you want, but something not MMOG. Then this catass will be cool with you all.
My sense is that this isn’t entirely accurate; a lot of the folks that I have seen developing this sort of game are what I would call single-player game developers. Yes, the “old guard” of online game developers includes a hefty dose of people who worked on session-based games, but folks like Kelton Flinn, John Taylor, Jessica Mulligan, and others have also worked at length in persistent worlds as well. The person who most resembles Brad’s remark is probably Jonathan Baron aka Bluebaron (his handle on Kesmai), whom I know to have some distaste for the “cumulative character” model of online gaming.
But more typically, I see this sort of game coming from people weaned on single-player gaming. Richard Garriott has talked about making online games this way for years and years, for example. Guild Wars draws its heritage from Diablo, not MUD1. It’s made explicit int he marketing materials of D&D Online.
- An MMO that dabbles with instancing is the last type Brad cites — essentially, working within the mud tradition, but making use of instancing as a tool. Brad advocates doing this with fully in-context pocket worlds, such as a holodeck or an X-Men-style “Danger Room.” He also cynically points out that the more common reasons are because there wasn’t enough time or budget to develop sufficient content to keep spawn points from being contested or overcrowded.
My own take on instancing is that it’s merely a form of embedded game. It’s ideally suited to embed session-based activities within a virtual world. I see no mechanical difference between having people go into a Holodeck to do something, and having people go into a pocket zone “chess world” when they sit down at a chess board, or go into a 2d shoot-’em-up when they sit down at an arcade machine in a virtual pool hall.
The experiential difference, of course, is significant. Current instances are designed to feel seamless with the larger world experience. Achievements carry through, the interfaces are identical to the main game, and so on. But this is, to my mind, an underutilization of what instances can actually accomplish. It’s approaching instances as a way to rectify perceived shortcomings with shared virtual worlds, rather than approaching instances as a manifestation of the power of shared virtual worlds.
Similar interfaces, achievements carrying between embedded games and the “main world,” and so on, are all details. The real question is whether you are designing a virtual world, and embedding some small-scale games within it, or are designing small-scale games and putting a lobby outside them to connect them. The mindset is best expressed with a diagram. To know what sort of game you are making, you need to know whether you are drawing a bunch of boxes with a hub connecting them, or whether you are drawing one box and putting a lot of little boxes within it.
To my mind, virtual worlds with relatively few embedded games, such as the Diku-style games, are one big box with only a few smaller boxes within them. The largest of the embedded boxes is the combat-and-loot game. And the “worldy” games are just ones with a lot of embedded boxes in them.
There’s been a lot of talk about whether the day of the “worldy” games is over. The above is why I think it isn’t. The trend over time is still, even with World of Warcraft out there, to have more and more embedded boxes in our virtual worlds. We may see that the quality level of each box keeps rising, but I have little doubt that over time, users will demand more “rides” in their “theme parks,” and not just more rollercoasters but more sorts of rides. The rollercoaster-only theme park fails if it doesn’t have at least a few food stands, and while the rollercoaster may always be the main attraction, the whole package includes everything from parades on Main Street to shops to concerts to convention hotels to go-karts.
By that light, calling the “gamey” games “theme parks” points towards the way they will ultimately evolve: towards worldy games.
The challenge is the budgets. 🙂
Some minor notes: Brad suspects that I might have the history of instancing at my fingertips. He’s wrong. 🙂 Muds didn’t go for instancing much. But the original Might & Magic Online, which was intended to be the follow-up for Meridian 59 at 3DO, was reputedly based entirely on it. It never saw the light of day.
You could also thinking of instancing as just one technique (out of many) for fracturing reality. http://www.mxac.com.au/drt/FracturedReality.htm
World-like vs. game-like – Another difference is that game-like worlds (could) give the players goals (“You should kill the evil overlord”), while world-like worlds let players achieve their own goals (“I want to be the evil overlord” or “I want to be a famous pastry chef”).
You mentioned on my blog “Why didn’t you comment on anything *I* said?” Well, um, I agreed with most of it? There was one thing I’ll bone-pick with you on, though:
>> My own take on instancing is that it’s
>> merely a form of embedded game.
(OK, so that’s a big thing.) I think that’s a really limited way to look at it. You can look at instancing more as load balancing, both in a crude technical sense and as a design tool. For example, if you see your game as an amusement park (see: WoW) you can use it to manage the flow of people among the attractions. If you have too many people for one copy of Space Mountain, hey! Let’s make 2 more dynamically, if load demands! I would hazard a guess that WoW really wishes that they had the ability to launch instances of Ironforge right about now, just as a random example. Sure, you don’t get the feeling of everyone on your server being in Ironforge at once. At the same time, playing WoW you realize this feeling is not entirely a GOOD thing.
I like the way EQ2 and COH will launch instances of larger global areas based on load, but allow the user to choose between them; they can still choose to congregate at a given point (say for a player-driven event) but the population of the areas are still large enough to drive random meetings between people.
Of course, if you come down on the side of “Instancing over-used, kills community” then yes, you could just use it to launch backgammon games!
Hmm, my sense is that most players rather dislike the “multiple copies of public spaces” thing — I regularly see gripes about it, anyways.
Anyway, yeah, obviously it can be used for load balancing. But that use is basically a lot like sharding. And these days, I kinda like Runescape’s version of that — have shards, but let people load balance themselves. Space Mountain is a bad example, because it’s more like the embedded game example that I was referencing.
Hola,
I first read Raph’s entry, then Scott’s take on it, followed by Jason Booth’s, and finally I trudged through the source document. My comments?
I agree with some of the ideas, confused by others, and disagree with others. I think an issue around instancing or sharding is on how it affects the parent world.
Some believe that by instancing content, that it leads to a “ground hog day” aspect where the player replays the same content over and over game, achieving the same experience (player-audience or in-game). But unless the initial state of instance is indentical, as well as the play of the instance, the result will always be different. Of course getting that result should be fairly easy if the instance is a close environment, and the play is macro’d for easy duplication. Of course this sort of play would result in same level interactivity that watching a 30min Episode of on repeat. But I haven’t seen this sort of Instance active in a VW or any game, so really each instance is different each time.
How different the instancing is on the parent world is up to the designers. The variables that can be changed while in the instance, and carried over to parent world might affect the initial variables of the instance, or of other instances. The play of the instance is determined by both designer (AI, etc.) and the player. The end result of this is that I find instancing just as Raph describes it: an embedded game.
On another note I am baffled on how instancing can destroy immersion of a game. I see little difference between an embedded instance in a virtual world, and say playing Elder Scrolls 3: Morrowind and chatting about my experience at the local gamer pub. I think one approach to instancing relies on not hiding it’s existance as an embedded game. The player will recognize the instance for what it is, without effecting the immersion of the parent world.
Just curious but would the various video arcade machines you find in GTA: San Andreas be considered an Instanced (Embedded) game? I don’t believe they affect the parent world at all, nor do they break the immersion (in fact I would say they increase it).
-Nathan J.
I aplogize if the above is beyond understanding… it’s 3am where I am, and I just finished reading McQuaid’s manifesto, so I’m a litte frazzled.
I demand wordly games now! I would still be playing WoW if there were more boxes, say playing cards in the tavern. I get bored with the hack and slash stuff. Sometimes I just want a place to relax. I think once someone comes out with a true virtual world, where there are so many boxes you can’t even keep track of them all, then I will be happy.
Raph wrote – I kinda like Runescape’s version of that — have shards, but let people load balance themselves.
I think this is a flawed model because at some point MMORPGs will actually use intelligent AI that remembers information about PCs. At such a point, allowing a PC to move from world to world is extremely hairy since all the NPC memories need to be transferred over along with the PC.
Mike, storing that data on the player rather than the NPC is how I’d do it anyway… so it’s pretty portable anyhow.
My intuition tells me that storing all of a NPCs knowledge about a PC within the PC database is problematical, particularly as NPC AI gets more intelligent.
I’ve tried to come up with a specific example, but at best can only show that storing the NPC’s knowledge on a per-PC basis becomes awkward in some situations, not impossible. In particular, my mind keeps wandering to Chris Crawford’s use of rumors.
I’ve tried doing it both ways, and have ended up thinking that storing the memory on the PC is sufficient for any case where you only want to have the NPC react to the player in a specific way. Odds are you have more NPCs than players, and storage for the memories on the NPCs can get expensive — and that’s if you persist the NPCs at all, which you may not as an optimization; and if they don’t get killed! You end up having to create systems that “age” the data out, and given how many players an NPC might interact with, that might lead to pointlessly short retention times for data.
In the case of rumors, I agree, it gets a bit more problematic. My basic use-case would be something like an NPC shopkeeper getting to know his regular customers and greeting them differently. If the data is stored on the player, you can even pull tricks like having the new shopkeeper recognize the player even if the old one is dead — “Ah yes, my father told me about you, you always came to buy bread and cheese. Have I mentioned he was foully murdered?”
There is so much that can be done with incredibly basic NPC behaviors that isn’t right now…
Just for the record, reading what Raph said above, I agree with so much of it, though he states it from his own perspective, it would be nitpicking to even try to find the points we didn’t agree on… and then, given numerous conversations with him (we’re friends and have been for some time) I bet we’d end up just realizing that we’re phrasing things differently.
I also hope nobody bashes Raph, or his response. It was well thought out and he is one of the smartest MMOG designers and visionaries I know.
’nuff said
Brad, I don’t know why everyone seems to think we were disagreeing. I thought I was going off on a tangent, myself. *shrug*
As you know, I tend to think instancing is a tool that can have bad consequences if overused. I’m not much a fan of the “hub and instances” kinds of games — they’re not where my interest lies personally, and I don’t generally find them to have the communities that the online worlds have. Some of them I actually exclude from my definition of online world…
Raph: exactly, and while some seemed to think I was bashing on WoW, which I was not, I was much more stating what you are: they can be overused, when you do use them be careful, etc. And if I was voicing any skepticism or worry, it would be about those games totally based on Instancing, not MMOGs incorporating Instancing.
As suspected, we’re on the same page 🙂
You all do realize that World of Warcraft uses instances like crazy. Every raid dungeon is instanced, and there’s a lot of them. The game is designed to be played this way – there is an expectation that your party can tackle higher end challenges without worrying about crowding some other raid group or vice versa.
As a mathematician I would say that you’re always in an instance – the only question is how many instances there are and how they are spawned.
It is a misuse of instances, in my view, to use them for main public areas. Buy a faster disk array or tweak your architecture. The commons should be the commons.
The History of Instancing…
Let me point you to a very early game that used instanced dungeons. “The Realm” by Sierra Online. I left The Realm to join Everquest when it was released, so they have been using instanced dungeons since before EQ came out.
By the way, excellent article Brad, and I will be trying Vanguard on the day of release. Stick to The Vision!
And actually, The Realm instanced every fight, basically. The “combat cloud”! 🙂
Very good point. I had forgotten the combat cloud.
Curious… but couldn’t all the combat encounters in the early RPGs be considered random “instances”? For example, in the first Final Fantasy all the combat encounters took place in an off-world area, and once the combat was resolved you returned to the back to the parent world.
Is there an MMO that extended this concept over, so that combat NPCs only exist in an instanced shared, which is triggered randomly?
-Nathan J.
We’re working on a Hub-and-Instances game, and I find Brad’s characterization of people who make such games quite humerous, and bearing no actual resemblance to reality, if my company is any indication. I love virtual worlds, but the hub-and-instances thing was just the right structure for our game, when it came down to it.
Nathan, in a single-player game I don’t know that the word instance really applies. I mean, yes, the experience of the player is the same, be it the fights in Ultima III or in FF VII or in The Realm. But of those three, only in The Realm is there actually a separate “room” being created that duplicates existing real estate, permitting a fight to occur independent of other activity that is going on. In the standalone games, everything else just froze.
Anyway, the answer to your question is yeah, The Realm did that. 🙂 There was a loud group lobbying us to do it in UO as well, but we declined, since we came from a mud backgroun, and I think it was very much a standalone RPG tradition.
I believe The Realm was created by Steve Nichols, and I think I read somewhere that he works for Microsoft now. Not really sure in which capacity though.
Replying to Sanzibal,
IIRC Steve Nichols had a number of co-creators. Myself among them for The Realm’s beta and 1.0 versions. Dunno what happened to him after Sierra’s Oakhurst facility got closed.
Thinking about Brad’s article and the replies to it:
Brad talks about how competing with other players for an item will make winning it sweeter.
” By making items easier to get, human nature dictates that at least a lot (most?) of people will find they value these items less, that their sense of accomplishment and attachment to a virtual character or item is diminished. People tend to value things they had to work for more than things they obtained more easily, or for no real effort.”
No argument, but…
Whatever quest or trial the player goes through should be challenging, right? Wouldn’t that trial generate perceived value for the player? If it doesn’t, there’s something wrong with the quest or trial. A player will not automatically undervalue an item simply because he didn’t have to compete against other players.
Moreover, the method by which players compete for an item should be interesting.
Nothing destoys a game’s charm faster than tedium. Remember that line.
Playing repeated rounds of “who clicks on the item first wins” is tedious, especially if it is preceeded by an arduous or annoying trial. Brad acknowledges this implicitly, but in his zeal to promote “opportunity to experience content” (and along with it some form of player competition to limit who has access to that content), I’m not sure he “gets it.”
The Real World is *nothing* if not tedious. There’s only so “real” we ever want to be.