« State of Play 2: Law Teaching on the Screen Workshop | Main | Accelerating Change 2004 »

Nov 08, 2004

Comments

1.

I wonder what Patricia Pizer and Michael Steele think of this, I didn't get to see their AGC presentation, but everyone said it was the best stuff on social dynamics in MMOGs that had been done in years.

2.

The conclusion of their paper starts with "Through a complex combination of features, SWG is one of the first attempts at encouraging social interaction in specific game locations."

This can't be right...

I also wish author could be more precise with the terms "player" and "character". Maybe that is just details, but it makes you wonder about the stuff you cannot check out. (As a player may have many accounts I doubt they measured "players"...)

3.

Ola> This can't be right...

Well, yeah -- that is wrong. You would need to ignore MUDs and MUSHes entirely, ignore social virtual worlds (e.g. TSO, AW) entirely, and perhaps even ignore many MMORPGs under 100K. In many of the 100K+ MMORPGs, though, I think they're -sort of- right. There are certainly social zones in DAoC, EQ, E&B (RIP), and others. There's the bazaar on Luclin. But SWG does really try to push player-to-player socialization in the cantinas as not just a place to do trades, but as part of designed game/role play.

I'd be interested in hearing where/if people think that there are designer-pushed social event spaces in other big MMORPGs.

4.

Wayfarer camps in EQ.

5.

greglas> I'd be interested in hearing where/if people think that there are designer-pushed social event spaces in other big MMORPGs.

Why this interest in big?

6.

Ola> Why this interest in big?

Yes, that probably does need some explaining. The short answer is that we've always been kind of inordinately keen on scale here -- see the 100K lists and Ted's numbers, so it's kind of our thing to care more about "massivaely multiplayer."

But that begs your question of my question. For me, personally, the "big" MMOGs are a little more interesting, even when in shards, due to: 1) major financial backing, 2) large player cultures with low owner/player ratios, 3) more certainty of diverse play styles/player types, and 4) more likelihood of a critical "readership."

7.

Jeff Cole> Wayfarer camps in EQ.

Yeah, I guess -- but the point there is just to group, not to socialize per se, right? (I never played with the LD expansion.)

8.

Greg,

I dunno. I've done quite a bit of socialization as the group forms, waits for others to show up, and waits for the mission. There's also the split and /random of loot at the end.

JC

9.

Yummy paper. This makes me wish we could datamine our own games this thoroughly!

Mike and Patricia's stuff is basically going way way forward from where I ended with the Small Worlds presentation. They actually datamined several MMOs and got a lot of interesting data.

Ola, as regards the social interaction stuff, the interest in big would be a fairly natural link to make--after all, small spaces do not tend to have the sorts of problems that the design elements discussed in the paper attempt to address. Most text games wouldn't bother with something like architecting social spaces, when the effective time to travel across the entire game even in the biggest games is typically on the order of seconds for an experienced player.

10.

greglas> For me, personally, the "big" MMOGs are a little more interesting, even when in shards, due to: 1) major financial backing, 2) large player cultures with low owner/player ratios, 3) more certainty of diverse play styles/player types, and 4) more likelihood of a critical "readership."

Hm... Well, I see the need for a large user base if you do quantitative analysis of e-baying. I don't see why you have to go 100K+ when it comes to solid financial backing, owner/player ratios or diversity in play styles... Perhaps if you want a representative sample of the general public in-game, but then you probably have to go way beyond 100K anyway.

As for critical readership... I stumble over inaccuracies or questionable claims in writings about the-big-ones all the time, even the ones terranova points to, so I am not sure if it matters...

11.

Raph> Ola, as regards the social interaction stuff, the interest in big would be a fairly natural link to make

Big games have many instances, smaller games have fewer instances, but perhaps even more players logged into the same instance. So unless you do analysis across instances (which would be intersting) it shouldn't matter at all.

12.

And Raph! While you are here :-) (sorry for spamming)

Maybe I misread the paper, but I didn't find information on how SWG deals with public chat. Do you have proximity filtering? I.e. does it matter how far away you are from another character? If SWG does, then location within the locale might be an issue.

I also wonder if you included flexible macroing in order to prevent 3rd party software from spreading (an thus bringing the trojan into the field). My guess is that you did..?

13.

Big games have many instances, smaller games have fewer instances, but perhaps even more players logged into the same instance.

This is almost never the case; in the 100k club, each instance (by which I assume you mean server or "shard") typically has more people on it than a smaller game that has only one instance--in the low thousands of simultaneous users.

So I'd say that "big" is basically about examining the problems that occur when you have groups large enough to demand increased attention towards issues like space, partioning of the userbase while connected, multiple large-scale subgroups (on the order of 50+), and so on. Most of the most interesting issues simply won't arise in a world of say, 250 simultaneous at peak.

Do you have proximity filtering? I.e. does it matter how far away you are from another character? If SWG does, then location within the locale might be an issue.

Of course it matters. Various interactions have ranges at which they can be performed, including hearing normal chat (we do have "tell" as well, and global chat channels, and in-game email).

Chat is purely radial, and does not do line of hearing checks other than "inside/outside." So the cantinas are quite a babble.

I also wonder if you included flexible macroing in order to prevent 3rd party software from spreading (an thus bringing the trojan into the field). My guess is that you did..?

Hurm... sort of, but there were many other factors. The macros permit roleplayers to put together one-keystroke commands for more complex performances, for example. Musicians and dancers can arrange their shows. Repetitive tasks can be streamlined (and there will always be some repetitive tasks, no matter how you try to remove them).

That said, I agree that the macro system is too powerful--in particular, recursive macros were not supposed to be possible, and due to a quirk in the UI code, are what the folks in the article are using.

14.

Why noone has ever tried to link the shards and make the actual link a gameplay element?

The purpose can be about forming a real community that isn't enforced in a specific server, letting the players to move if they so choose, but always between "cloned" shards (so not offering something different from shard to shard).

In this way you can use the "portals" as a way to regulate the server load and manage the population unbalances plaguing (PvP) games like World of Warcraft or DAoC.

From a side you add rules useful for the design (balance, server's load), from the other side you link these portals into the rules of the game, perhaps letting the players open a portal when specific PvP results are achieved. Etc...

I believe that the potential of a similar mechanic is extremely interesting. There a lot more consequences for the game on top of those I pointed out.

15.

Raph> So I'd say that "big" is basically about examining the problems that occur when you have groups large enough to demand increased attention towards issues like space, partioning of the userbase while connected, multiple large-scale subgroups (on the order of 50+), and so on. Most of the most interesting issues simply won't arise in a world of say, 250 simultaneous at peak.

Yes -- and I would add that those synergies spread offline as well -- with larger scale you can get major fanboards, detailed walkthroughs, geographic (real-space geographic) communites, fanfests, etc. Interesting things start to happen with scale not just ingame, but out of game.

Ola> As for critical readership... I stumble over inaccuracies or questionable claims in writings about the-big-ones all the time, even the ones terranova points to, so I am not sure if it matters...

Oh -- I'm not saying scale increases communicative accuracy at all. But it does give you much more noise, and thus much more signal. It also provides a shared vocabulary for conversation. By analogy, two average readers of this blog today would probably find it easy to have a conversation about Shrek II. But they probably couldn't talk about an obscure documentary. (This doesn't express the view, though, that the documentary wouldn't be better than Shrek II or that the conversation about the documentary wouldn't be more worthhwile.)

16.

Raph > This is almost never the case; in the 100k club, each instance (by which I assume you mean server or "shard") typically has more people on it than a smaller game that has only one instance--in the low thousands of simultaneous users.

You'll find at least one game in the sub-100K club that claims peaks at 15K simultanous.

Also note that a game like EQ segregates it's population: General, PvP, Roleplay, Europeans, Elite... So Greg's argument about diversity of playstyles doesn't hold at all.

Raph> Of course it matters. Various interactions have ranges at which they can be performed, including hearing normal chat (we do have "tell" as well, and global chat channels, and in-game email).

:-) It may be obvious to you, but this creates a serious sampling problem for a quantitative analysis. In my experience it even creates problems for qualitative analysis...

1. Two players chatting: you hear one, but not the other.

2. Human beings tends to have fixed favourite spots, thus you risk having some players overrepresented.

Raph> Chat is purely radial, and does not do line of hearing checks other than "inside/outside."

Does the radius vary with the amount of chat going on, or is it fixed?

Raph> That said, I agree that the macro system is too powerful--in particular, recursive macros were not supposed to be possible, and due to a quirk in the UI code, are what the folks in the article are using.

Mmm... But, if you can access functionality from the keyboard or trivial mouse movements, then how can you stop them from installing software to achieve the same thing?

17.

greglas> with larger scale you can get major fanboards, detailed walkthroughs, geographic (real-space geographic) communites, fanfests, etc. Interesting things start to happen with scale not just ingame, but out of game.

This sounds reasonable, but it isn't necessarily the case. A smaller game can have a tighter community with most users coming from the same area. (You don't need much more than 10K+ to get major fanboard, detailed walkthroughs etc). Besides, does the number of such phenomena matter if you don't have the capacity to study more than 1 or 2? It might actually be benefitial to have fewer as it is then MUCH MUCH easier to find respondents with similar characteristics (i.e. easier to control variables).

greglas> It also provides a shared vocabulary for conversation.

Yes, that is true. Unfortunately the big games are going to be copy cats and therefore you end up a body of research that perhaps has no validity... beyond a very narrow design.

I'm also not sure if systems like SWG are good subjects to study due to the heavy bias induced by the Star Wars fictionaly universe. The users may be motiviated by things that go beyond what exists in the world. This is only a good thing if you are interested in studying the fictional universe of SW, otherwise it is just some bias that is very difficult to assess the impact of.

18.

Ola> It might actually be benefitial to have fewer as it is then MUCH MUCH easier to find respondents with similar characteristics (i.e. easier to control variables).

Ola> This is only a good thing if you are interested in studying the fictional universe of SW, otherwise it is just some bias that is very difficult to assess the impact of.

Hmm... you're expressing your empiricist/informatics leanings here. I guess where you might be driving at is that we have these cantinas/starports in SWG because the SW fiction mandates that they should be present as social spaces? And, were we to start from "scratch" (or, say, reality), we would get a design more attuned to a utilitarian design calculus?

It seems we're coming at this from much different perspectives -- and maybe your perspective is more attuned to the goals of the authors of the paper here. I'm personally not so much interested in optimizing design (I like reading about it, but isn't my thing.) I'm more interested in thinking about how the data relates to policy issues raised by VWs and how social meanings and behaviors are negotiated and controlled. So the data on the "big" worlds has more value to me. I'm also less inclined to seek the kind of research purity that you seem to want, because I'm skeptical of the value (again, just for my purposes) of segregating out the "fictional universe of SW" in order to discover some other quanta. Imho, if a MUD or VW doesn't partake of the fictional universe of SW, it will need to partake of some other socially mediated "fictional universe" instead, right?

But I guess all I'm really saying is that we have different research agendas.

19.

Greg, yes, I am probably expressing my leanings towards design, but I think it hold in other fields too. I think it is in general preferable to study worlds where you can say that what happens in the system to some extent is a result of the design.

You don't know whether the players respond to the design of SWG or whether they respond to elements in SWG interpreted through their understanding of the SW universe and what it should be like. I.e. preconceptions about SW may have an impact on behaviour.

E.g.:

- you cannot assume that role-players in SWG are building their characters on what is available in SWG, but you have to include the whole SW universe of which different players have different knowledge.

- you cannot assume that players want to be Jedi just because of what Jedis are in SWG, but need to include what their position in the SW universe.

etc

greglas> I'm more interested in thinking about how the data relates to policy issues raised by VWs and how social meanings and behaviors are negotiated and controlled.

But isn't it then desirable to study smaller worlds where you:

1. can get a good overview over the situation

2. know the context of players' expressions (single instance worlds)

greglas> Imho, if a MUD or VW doesn't partake of the fictional universe of SW, it will need to partake of some other socially mediated "fictional universe" instead, right?

Yes, but that is part of the design and accessible through it. Isn't it?

20.

Abalieno: Why noone has ever tried to link the shards and make the actual link a gameplay element?

Many have in the past. It doesn't tend to happen with the big games because it would incur flash crowd issues. Most servers already run near capacity.

In addition, as more games include persistent world elements such as houses or shops, mobility for characters only becomes problematic.

Several games offer character transfers from server to server for a fee.

In this way you can use the "portals" as a way to regulate the server load

It's more likely to happen the other way; segmenting allows you to load balance more easily than free-flowing populations.

from the other side you link these portals into the rules of the game, perhaps letting the players open a portal when specific PvP results are achieved

You may want to read about the inter-MOO wars covered in Julian's book. :) Or about the invasion of one of the M59 servers by the clones from another server...

You'll find at least one game in the sub-100K club that claims peaks at 15K simultanous.

Yes, of course. That's why i said "almost never" and "typically." By my definition, that's massive already (my personal cutoff point is 256, which is a legacy figure from the days when that's how many telnet connections an out of the box Unix install supported).

Also note that a game like EQ segregates it's population: General, PvP, Roleplay, Europeans, Elite... So Greg's argument about diversity of playstyles doesn't hold at all.

Hmm, my experience has been that there's still quite a diversity of playstyles within those groupings. That said, more empirical data to compare and contrast those populations would be nice.

1. Two players chatting: you hear one, but not the other.

This is almost identical to real life...?

2. Human beings tends to have fixed favourite spots, thus you risk having some players overrepresented.

Also like real life...?

Does the radius vary with the amount of chat going on, or is it fixed?

Adjustable by users.

if you can access functionality from the keyboard or trivial mouse movements, then how can you stop them from installing software to achieve the same thing?

It's a question of ease more than of prevention. The built-in macro system does do things that with an external macro program would be a huge pain; it also cuts down on the use of external macro programs because there's a basic tool available to everyone. I'd rather have a level playing field...

Besides, does the number of such phenomena matter if you don't have the capacity to study more than 1 or 2? It might actually be benefitial to have fewer as it is then MUCH MUCH easier to find respondents with similar characteristics (i.e. easier to control variables).

Not if what you are interested in is the interaction between groups, not individual groups themselves. From a game design point of view, that's certainly what interests me.

Unfortunately the big games are going to be copy cats and therefore you end up a body of research that perhaps has no validity... beyond a very narrow design.

An alternate way of looking at this is to say that you'll end up with a body of research that is directly applicable to the big games, but may not have applicability beyond that. Given that the paper mostly confines its recommendations and evaluations to the big games in the first place, how is this a problem? I grant you that trying to extrapolate to real life or other types of games would be problematic.

21.

Raph> This is almost identical to real life...?

Not really, but that wasn't the point. The point was that the collected data may have a significant sampling problem. Whether you would get the same problem in another setting isn't really interesting. If you can increase the range to include most of the cantina and set the same range for different types of events then I guess it is less of an issue.

Raph> it also cuts down on the use of external macro programs because there's a basic tool available to everyone. I'd rather have a level playing field...

Yes, but my point was more: not having a macro tool isn't an option if there are some subtantial benefits for using one. In many games you are expected by the players to use external tools. Tracks back to text MUDs even. It's not like macroing is a new unknown issue...

Raph> Not if what you are interested in is the interaction between groups, not individual groups themselves.

You may study that they interact, but you cannot study why. You need to study the why's if you want to say something about causality. Causality ought to be interesting for design...

Raph> Given that the paper mostly confines its recommendations and evaluations to the big games in the first place, how is this a problem?

Actually, I didn't find any design guidelines derived from their research in the paper so I can't really answer your question. My points were more directed towards Greg's arguments for studying 100K+ systems.

22.

Just a quick note to point out that at least one of the 100k games, Runescape, allows users to login to any of the servers and switch (essentially) at will - other than a 30 second timer in the server, which was added to deal with a minority who used 3rd party programs to rapidly switch servers gathering resource spawns.

The users thus 'load balance' the servers, and have informally turned certain servers into the main bazaars; much as the starport in the paper referenced here is used by players to advertise vendors/provide services, Runescape members will log onto world 2 and go to certain locations there to buy/sell from other players. (Certainly it goes on in other places and on other servers; if you don't find what you are looking for on your current server, then perhaps it's time to go to the merchant world.)

23.

Ola> I think it is in general preferable to study worlds where you can say that what happens in the system to some extent is a result of the design.

Design always shapes play--I don't know what you mean by the comparative extent to which this happens in SWG as opposed to some other game. Check out Professor Whang's talk at SOP II (and the Culture of Play panel resonates with this):

http://web.stream57.com/nylaw/103004_virtualworldsinasia0003.htm

Whang looked at Lineage as played in Korea and Japan, and found significant differences in a many aspects of play. People come to Lineage, like they come to any text or object, with certain understandings. You can't speak about the influence of the Lineage design on play without taking account of how that design created certain affordances that were meant to be exploited by the players based on their prior understandings.

Ola> You don't know whether the players respond to the design of SWG or whether they respond to elements in SWG interpreted through their understanding of the SW universe and what it should be like. I.e. preconceptions about SW may have an impact on behaviour.

And when I respond to your comments here, am I responding to your comments or am I responding to my understanding of your comments? You'll never create a game design insulated from player preconceptions -- if you did, that design would be unintelligible.

Ola> But isn't it then desirable to study smaller worlds where you:
1. can get a good overview over the situation
2. know the context of players' expressions (single instance worlds)

It's desirable to study everything, I'm afraid. The question is priorities.

24.

Greg, bad idea. You are evading the issues under the pretense that because you cannot eliminate an issue you shouldn't try to limit it. That's flawed logic.

25.

Abalieno: Why noone has ever tried to link the shards and make the actual link a gameplay element?

Ola: Many have in the past. It doesn't tend to happen with the big games because it would incur flash crowd issues. Most servers already run near capacity.

Last time I was on ToonTown, it allowed you to jump shards at will. Since the game doesn't have social downtime and doesn't allow trading, the shards tend to self balance a somewhat - Too many people on a shard means no buildings to hunt. Too few people on the shard means no one to group with.

Randy

26.

Maybe I'm just avoiding your issues?

Logic usually proceeds on a certain set of assumptions. If I'm doing Lobachevskian geometry, I'm bound to confuse someone who thinks I'm doing Euclidean geometry.

I'm fine with you looking for your issues and with me evading your issues. If you want to examine environments that limit the influence of prior player understandings in order to study the effects of "pure" design, more power to you -- have fun. But why do you want me to share that interest and methodology?

27.

Greglas wrote:
Yes, that probably does need some explaining. The short answer is that we've always been kind of inordinately keen on scale here -- see the 100K lists and Ted's numbers, so it's kind of our thing to care more about "massivaely multiplayer."

But that begs your question of my question. For me, personally, the "big" MMOGs are a little more interesting, even when in shards, due to: 1) major financial backing, 2) large player cultures with low owner/player ratios, 3) more certainty of diverse play styles/player types, and 4) more likelihood of a critical "readership."


But you spend a lot of time talking about Second Life, which is far from big, does not support a lot of playstyles, does not have a low owner/player ratio, etc.

--matt

28.

greglas> But why do you want me to share that interest and methodology?

I don't. I assumed we were discussing the general problem of selecting a case and the merits of the few 100K+ systems.

29.

Matt> But you spend a lot of time talking about Second Life, which is far from big, does not support a lot of playstyles, does not have a low owner/player ratio, etc.

I take it you're using the plural "you" here? Second Life is cool and I like reading about it, but I don't think I've personally posted a lot about Second Life. I've only spent a few hours in it and I don't think of it as an MMORPG at all -- it's maybe a MMORP__.

My question -- way up there in the thread at this point -- was: "I'd be interested in hearing where/if people think that there are designer-pushed social event spaces in other big MMORPGs." And Jeff gave an answer, but mostly I've been trying to defend asking that question ever since.

Re my motives again, I was speaking about the singular "me" in the bit you quoted. I can't presume to speak for the other 14 authors on the blog. Obviously, the word "big" and my interest in scale struck a nerve here. I actually excluded smaller scale VWs from the question because I *presumed* that smaller scale VWs would have such spaces.

30.

> Abalieno: Why noone has ever tried to link the shards and make the actual link a gameplay element?

> Ola: Many have in the past. It doesn't tend to happen with the big games because it would incur flash crowd issues. Most servers already run near capacity.
> Ola: It's more likely to happen the other way; segmenting allows you to load balance more easily than free-flowing populations.

The reason why I imagined and suggested it it's exactly the *opposite*.

When SWG launched it was hell (aside the bugs) because everyone was trying to enter or Starsider or Bloodfin. Closed servers bring directly to capacity problem while my issue solves directly this problem.

Just think at a big game launching. You'll need a way to balance the population so that everyone doesn't try to go in a server. I suggesting this by "regulating" directly the rules to access a shard. What I mean is that following my idea you are able to open/close the servers only when they have the planned capacity and balance. When they are full you close them.

This brings obviously to the complaints of the players but for SWG the complaints were VERY LOUD. Because once you choose a server you are blocked in it forever if you don't want to forget that character and build a new one somewhere else.

Instead I believe that forcing servers in an environment where this choice isn't *definitive*, is way more acceptable. During the release the players could accept to finish in a different shard than where their guild is planning to go (for example), because they'll have the possibility to gather again in another moment. Without loosing their progress.

The portals mechanic I imagined works as a mix of rules. There are rules that set if a portal CAN be open based on the server load and PvP balance. If the shard meets the requirements, the portal is set as "available". Then it becomes a PvP mechanic where you need to conquer power nodes in the shard to make the portal active and use it.

This means that we still have a STRONG way to control both the load of the servers and the balance. At the other side the players won't be able to jump everywhere as they choose so (so we allow the game and each shard to develop smaller communities and economies). But at the same time noone will be forced *forever* in a situation with no escape. If you miss a portal a day, you'll be able to try the next one. At a point you are still able to join your friends and I think the result is way more acceptable.

31.

Greglas wrote:
I take it you're using the plural "you" here? Second Life is cool and I like reading about it, but I don't think I've personally posted a lot about Second Life. I've only spent a few hours in it and I don't think of it as an MMORPG at all -- it's maybe a MMORP__.

Sorry! I thought you were speaking as an official TN representative (if such a thing exists!).

--matt

32.

> Ola: Many have in the past. It doesn't tend to happen with the big games because it would incur flash crowd issues. Most servers already run near capacity.
> Ola: It's more likely to happen the other way; segmenting allows you to load balance more easily than free-flowing populations.

These were Raph's comments, not mine! :-)

I personally think there can be advantages in keeping the servers separate as releasing a new server is kind of like doing a reset. Creating a level playfield. Say, if you have just corrected the economy you don't want spill over from existing servers to the new one. Assuming you can make worlds that start small and grow...

33.

Oops, sorry.

I'm still convinced of my points, though.

34.

Well, since Ted asked, here's my initial opinions on the MMO social paper recently offered by Nicolas Ducheneaut and Robert J. Moore of PARC.

Overall, this is a nicely structured paper with a clearly described methodology, background info, assumptions, observations and conclusions. They make some pretty good recommendations for future design vectors vis-a-vis how the usage and layout of virtual spaces and the emergence of social structures essentially go hand in hand. We have only to witness the unfolding (and subsequent unravelling) of Earth & Beyond to see this at work. In fact, many of the problems they highlight are already well known within the MMO development community. Sad to say, politics, resources and time are more frequently the unbeatable level-bosses which rational game designs must conquer. More so, I suspect, than simple lack of forethought by experienced MMO design teams.

Painfully missing from this paper (as the authors correctly disclaim) is first source information about private conversations and non-visible social contexts such as friendships, guild membership, grouping, etc. This severely limits the authority of their conclusions, since hidden behind the veil of public chatter are all kinds of meaningful social interactions and discourse. Review of the data from several large MMO operations tells us that ~70% of all interactions (player to player chats, structure changes and business transactions) take place privately, even when conducted at/near public locations. This then is the handicap that all 'outsider' studies face: lack of access means they are forced to work with a statistically noisy sample. Despite the volume of information, painstakingly collected, there remain vast, critical gaps in their data-corpus.

Still, these guys are on the right track; their results are consistent with regular reports on social metrics in MMOs populations. One of these can be clearly seen in their graph: the well-known tendency for nodes to follow "preferrential attachment" rules, eg. most of the communications are from/to a disproportionately small fraction of the population. Sociologists and Communications Theorists traditionally plot this as a log curve, with a analysis of the exponent. Additional study of the generative mechanisms behind such preferential connections depends highly on the social context, but are commonly linked to factors such as demographics, topology of the space, game theory (eg. players maximizing individual or group utility), etc., etc. Most MMO of the top designers are *very* aware of these phenomena are are designing accordingly.

As a professional MMO developer, I can honestly state that similar papers and studies frequently circulate internally within the design teams of large MMO publishers. These internal papers are far more accurate than all-but-the-best public or academic papers on MMO social phenomena, due largely to their direct access to primary sources such as server logfiles databases. Sadly, most of these papers will never see the light of day, since their information and conclusions are based on highly confidential information. This is regrettable, since the state of our craft will likely not advance very rapidly until one of the big publishers breaks the ice and 'goes public' with actual hard data for academics to ponder and dissect. This is one way in which the interactive industry is NOT like the old film industry. The USC film school doesn't need access to proprietary datamining in order to understand why people like certain films... but the USC Interactive school will need such data in order to really understand large virtual communities consisting of several hundred thousand participants. Like large eco-systems, such complex communities are generally impossible to fully comprehend without a little help from statistics, math and some solid data.

--Mike

35.

Michael -- that's really interesting. I'm a bit confused by one bit, though.

You say the "top" and "experienced" MMO design teams (I'm curious how many are in this group and how it relates to the scale issues above) are "*very* aware of these phenomena are are designing accordingly" and that "the problems they highlight are already well known within the MMO development community." That makes sense. You also add that "politics, resources and time are more frequently the unbeatable level-bosses which rational game designs must conquer" -- implying that the vision of what should and can be done is far in advance of what is done in practice. That makes sense too.

You say that the shared understanding of the top and experienced teams is recorded in circulated in-house papers "based on highly confidential information." Again, that's fine. It is easy to see how you can make claims about general industry knowledge and practices -- that has to be porous information. The project staffing of MMORPG design teams, like the staffing in the rest of the game industry, seems highly fluid -- devs seem to float from company to company a lot (with NDAs, of course).

The part that I don't get is this: "our craft will likely not advance very rapidly until one of the big publishers breaks the ice and 'goes public' with actual hard data for academics to ponder and dissect." This last bit strikes me as curious. What you've described so far is essentially a standard privatized research model. I'm curious to know why the papers need to go public and be dissected and pondered by academics for the craft to advance rapidly? If the top, experienced teams are already doing the data-mining, sharing insights (to some extent) across companies, know their stuff, and are tryinging to design accordingly, then things are well in hand, right?

There seems to be an implication, though, that you've got some kind of market failure occuring here -- you think the proprietary control of the data is actually a drag on innovation. Can you explain that a bit?

36.

The essays at Play On, especially Accountability http://www.parc.com/research/spl/projects/PlayOn/accountability.htm, gave me a lot of insight into one thing about online worlds and why the players are the type of people they are. And it sounds like a game created according to the advice of the PlayOn team would be a sort of purgatory for me, and many other players like me (certainly the minority of possible players, but a much larger proportion of actual players, for reasons that are now much clearer). That's because we are er... socially maladroit. Or introverted. Well, those are overgeneralizations. I will be more specific.

I don't really *want* to be waiting while someone talks on the phone in a game, or messes with their datapad. I certainly don't want it to be obvious what I'm looking at from looking at my avatar. I don't want people I'm chatting with to know how many other people I'm chatting with. And I like multitasking. It keeps game conversations from being as tedious as face to face ones are, with the whole waiting till the person is done to start and having to listen at the speed they talk, instead of using the fact that people type/talk slower than they read/listen to add in a few more conversations simultaneously. But I suppose it would be a boon for the many people who would hate the current interaction style of games, for the reasons listed. And it gives me a good idea of why I am so much more socially adept in IRC-style chats than in meetings. The whole intermixed threads of conversation is my natural flow.

I wrote "I don't want people I'm chatting with to know how many other people I'm chatting with." Why is this? I worried that I desired to be deceitful. My theory of why is this: Because they'd think I was being rude, and not giving them my full attention. What they would not realize unless they knew me well enough not to care, was that it is actually easier for me to give them my full attention when I'm also giving full attention to someone else, because that means my attention is *filled*. If my attention isn't full, it goes looking for something else. And if that fills it, I might actually forget about them in the middle of the conversation with them (I've done this. It's frustrating). Whereas if I'm having 3 conversations, each of which takes up 1/3 of my attention, I can give full attention to each of them (up to the capacity it allows).

37.

Dee, I don't think you have to worry. To make all hidden channels and acitivities visible is impractical and impossible (e.g. reading guild chat). It also affects roleplayers ability to hide OOC coordination and background activities that are neutral would become more intrusive. Worse, people will start to worry about what you are saying behind their backs...

What you could do though was to have a particular mode in special locations where "I am all yours". I.e. a symbolic act where you explicitly dedicate your full attention to the person. Great for cybersex... (Guild chat could be buffered.)

AFK detection and efficient detection of lost connections are useful, but that is already in use.

That said there are fields such as studies of cooperative work where these issues are more imperative...

38.

Abalieno> Why noone has ever tried to link the shards and make the actual link a gameplay element?

Abalieno> The purpose can be about forming a real community that isn't enforced in a specific server, letting the players to move if they so choose, but always between "cloned" shards (so not offering something different from shard to shard).

Abalieno> In this way you can use the "portals" as a way to regulate the server load and manage the population unbalances plaguing (PvP) games like World of Warcraft or DAoC.

Runescape allows free movement between shards. The link isn't in-game, but given that it takes only 15 seconds to log out from one server and log into another, there's no real difference. This leads to some interesting situations with load balancing.

For example, since each shard has a limit of 1250 simultaneous users, the first few servers on the list will fill up to that limit, while the rest of the players will distribute themselves fairly evenly between the other servers.

Most in-game trading takes place on those high-population servers, while trading that is arranged out-of-world takes place on a low-population server -- since the high-population servers are at or near the capacity limit, you can't be certain that both participants will be able to log on.

39.

> you think the proprietary control of the data
> is actually a drag on innovation. Can you
> explain that a bit?

Well, it's just an opinion, of course... but when I look at the amount of directly-relevent, advanced work being done in academia on economics, social networks and communications theory, I feel this level of sophisticated analysis is generally beyond the budget (or interests) of most publishers. So the really state-of-the-art work (on virtual economies, social structures and community) that needs to happen for our understanding of virtual worlds to advance is being hindered by the 'proprietary information' wall that prevents public research from occuring.

When I say that the top teams have been doing this type of analysis, I am referring to internal papers/studies that are largely equivalent to the papers we see publically today, albeit with much better data. What I am *not* seeing are dozens of Doctoral-level thesis. Most of the pure research is being done in far less structured, word-of-mouth, craftsman/apprentice like manner. This is great for handbuilding incrementally better virtual worlds, but it's not building on a body of empirical scientific research. We won't be seeing any 'Apollo projects' coming out of the interactive world at this rate.

This is where the analogy between the maturation of the film-making industry and the maturation of the interactive industry breaks down. Film is about telling stories, albeit artistically and technically demanding stories. In contrast, virtual worlds (above a certain size, let's say ~100,000 people) are living, dynamic, complex ecosystems. You simply cannot hope to *really* understand them (never mind significantly improve them) without a lot of hardcore science. Sure, there can be story and art and other thema/flavor, but it's really all about the people, and all they ways in which they interact.

Building an airplane wing was a craft until 1939. After that it was all science.

In general, I think it is quite possible for MMO companies to release information which has been 'sanitized' of any risky confidential information, especially on older or cancelled products. However, the question they will ask is "Why should we do this?". I probably don't need to lecture anyone in this forum on the challenges inherent in trying to convince product managers to see beyond short term worries and focus instead on less tangible strategic benefits. :-)

40.

Michael Steele wrote:
In contrast, virtual worlds (above a certain size, let's say ~100,000 people) are living, dynamic, complex ecosystems.

Are there any virtual worlds that have 100,000 people in them? None of the Western ones do that I'm aware of. All the products (that I'm aware of at least) with that many users split the product into many virtual worlds, not one big one. You could consider the entire community around the product a virtual world but at that point you may as well consider the user community around Linux or motorcycles a virtual world as well.

--matt

41.

Dear Terra Novans,

We are just back from presenting our paper at CSCW and, to our delight, we found mention of our work on your blog! We have been following Terra Nova for a while and the discussions here are always insightful, we are glad to be part of it.

A couple of comments on previous posts (we have not digested them all!):

- As Ola and Greg pointed out, we do not mean that SWG is the *first* game trying to encourage social interactions. What we do mean is exactly what Raph points out: large MMOs are qualitatively different from smaller, older virtual worlds like MUDs, if only in their scale. Designing public spaces becomes particularly important when distance matters and players simply cannot know everybody else – public spaces can offer a clear locus of information sharing and exchange. The question becomes, then, how do you design successful social spaces in large virtual worlds? This is what we are trying to answer with our paper.

- Michael rightly points out that our paper does not consider private messaging between players and other non-visible social contexts. It is particularly interesting that, as Michael mentions, 70% of interactions take place privately – we could not compute such a number from our analyses, but it rings true for SWG based on our ethnographic data. However (and in relation to the point above), we do not think it really matters for the question we are addressing. Again, the issue is how large MMOs can support successful, interactive, *public* social spaces because they play an important role in any community. Sure, players can talk entirely in private, but how do you overhear information then? We learned a lot about grouping and socialization simply by standing at coaches in EQOA, for instance – this would never have happened if the messages had been private. Also, private messaging presupposes in most cases that you know the person you are contacting – there are no chance encounters in IM, whereas public spaces have the potential to put strangers in contact with each other. Finally, even if you argue that group conversations can still happen in private (e.g. SWG’s group chat), you still need to form the group in the first place – which leads back to the necessity of public social spaces where you can hook up with others.
This being said, it would be fascinating to have access to private communications across the entire world and analyze them. One of the points we discuss in our forthcoming Other Players paper is that, while public social spaces in games like SWG might not be truly sociable, the *entire* game may very well be. To test this, you’d need to be able to consider the sum of each player’s interactions, both public and private.

Thanks again to you all for your comments. We would love to stay in touch.

42.

Dee, you make an interesting point: “…I like multitasking. It keeps game conversations from being as tedious as face to face ones are….”

Certainly if you find conversations in real life to be tedious, you would not want to make online interactions more like them. Many people like current systems, with their quasi-synchronous chat and low mutual awareness, for the kind of interaction they afford. Using these systems no doubt has become second nature to many people. I often meet some resistance from people when I start to propose ways to make virtual face-to-face more like real life.

Our recommendations from the PlayOn project about how to “improve” the simulated face-to-face interaction in MMORPGs certainly presuppose that designers are interested in reproducing the experience of real face-to-face.

Of course some may not be interested in this at all (nor should they be). Cyberspace enables us to create new worlds and forms of interaction that are impossible under the constraints of real life. For example, there’s no constraint in cyberspace that avatars must be humanoid or have a fixed size. Nor as current systems demonstrate, is there a constraint that users must take turns when they chat. So why not exploit these new possibilities?

I’m all for new forms of social interaction, yet when I look around at current virtual worlds, I don’t see attempts to create new forms but rather attempts to emulate real-life. This appears to be the case in Star Wars Galaxies for example. (Why go to the trouble of using motion capture to animate avatars if you’re not interested trying to reproduce the feel of real life?) And this makes sense because the whole point with this game world seems to be to give players the experience of inhabiting the Star Wars universe. If they enabled you to change the size of your wookie at will so that it could be 7 feet tall at one moment and 50 feet tall the next, you could no doubt do some interesting things, but that’s no longer an experience of Star Wars.

So our recommendations (so far anyway) are certainly most relevant for designers who are aiming to create online experiences that feel like real face-to-face (in other words, The Black Sun nightclub from Stephenson’s Snow Crash). As a sociologist who studies the mechanics of real-life social interaction, I personally find this challenge especially exciting. However, as a player I also find that I want an online experience that feels like I’m really sharing a physical space with a live person. I have had many brief moments like this in current games, but they invariably dissolve when the other’s avatar remains silent and still for too long or our conversation develops too many parallel threads.

43.

Carnildo Greenacre:
Runescape allows free movement between shards. The link isn't in-game, but given that it takes only 15 seconds to log out from one server and log into another, there's no real difference. This leads to some interesting situations with load balancing.

For example, since each shard has a limit of 1250 simultaneous users, the first few servers on the list will fill up to that limit, while the rest of the players will distribute themselves fairly evenly between the other servers.

Most in-game trading takes place on those high-population servers, while trading that is arranged out-of-world takes place on a low-population server -- since the high-population servers are at or near the capacity limit, you can't be certain that both participants will be able to log on.

Okay, the bad consequences are there because it's an "OOC" system about the *players* and not about the characters. In my idea you can move from shard to shard as an "in game" mechanic. You need to conquer power nodes, open portals. Not just switch servers freely.

The point is to not transform it in a virtual system where you can be everywhere, all the time. Taking the best from every situation. Moving, in my idea, is time consuming. It asks you a dedication, to play the PvP, contribute to open a portal etc...

This is why I expect it to work.

44.

Abalieno wrote:
The point is to not transform it in a virtual system where you can be everywhere, all the time. Taking the best from every situation. Moving, in my idea, is time consuming. It asks you a dedication, to play the PvP, contribute to open a portal etc...

This is why I expect it to work.

I don't follow the logic between the first paragraph and the "That is why I expect it to work." What's the chain of reasoning there?

--matt

45.

I was commenting what happens in Runescape. That happens because you can switch between servers at will, taking the best you can from each one. In a PvP environment you could simply join the winning part all the time.

This doesn't happen in what I imagined simply because from a side the flow and the popluation distribution is directly controlled to be balanced, and from the other side it takes time. So you cannot switch at will. And it's not even a timesink because you don't need to "wait" but directly gain powernodes by joining the PvP game.

My goal is to give the players the possibility to move, but not as a daily activity. I want them to play always in the same server for the most part. All the high-level stuff tied to the guild mechanics is hardly tied to a single shard and it will need maintenance.

The comments to this entry are closed.