« January 2006 | Main | March 2006 »

February 27, 2006

Hearing waves and bows

Gesturing in MMORPGs is an unusual experience. On the one hand, avatar gestures can appear impressively realistic (e.g., EQII or SWG), and in some cases may even be created through motion-capture of real human bodies. So when I type "/wave" my avatar performs a pretty realistic-looking hand wave. On the other hand, whenever my avatar performs such a gesture, the system also generates a public text emote describing the gesture (exceptions are SL and There). So typing "/wave" also generates a message like, "Bob waves to you." In other words, gesturing in most MMORPGs is multi-modal with both visual and textual components.

So essentially these systems simulate a world in which when people gesture, they also simultaneously announce what they are doing. When I shrug at you, I also say, "Bob shrugs at you." Imagine if we did this in real life. You would not only see people gesture, you would hear them gesture (since reading text is the correlate of hearing voice in most games). This totally changes the organization of gesture in interaction.

You would not necessarily need to coordinate a gesture so that the recipient could see it. As long as you knew he or she could hear it, that could be enough. In fact, in such cases, you wouldn't even need to perform the visual part of the gesture at all. I approach Nic from behind and say, "Bob waves to Nic" without bothering actually to wave. Nic doesn't even turn around but simply returns, "Nic waves to Bob." (There's a cheesy commercial or SNL skit in here somewhere, I'm sure.)

This is basically my experience of gesturing in virtual game worlds. Despite the impressive sophistication of the gesture animations, players tend to rely more on the text emotes instead. The result is a player experience that is more like a text-based MUD, than a three-dimensional, avatar-based world.

For example, take the following interaction from Star Wars Galaxies. In this encounter, Atac is in the player city in which she lives on Naboo, and she heads toward the player association hall to see if anyone is around.

[Star Wars Galaxies: Unseen Waves: Atac's perspective]

01 00:00 ((Atac rounds the corner of the PA hall))
02 00:00* Sin'thea waves to Teli Tubbi.
03 01:04* Sin'thea nods at Kimon Calari.

As Atac rounds the corner of the PA hall (line 01), she sees two text emotes, "Sin'thea waves to Teli Tubbi" and "Sin'thea nods at Kimon Calari." The text emotes alert her to the fact that her friend, Sin'thea, is nearby even though she cannot see the actual avatar animations with which they correspond. In fact, she cannot even see any other avatars from her current vantage point.

04 05:21 ((Atac mouses over Sin'thea's avatar))
05 07:02 ((Atac selects sin'thea's avatar))
06 07:14 Sin'thea: is that Teli I see?
07 10:12* You wave to her.

As she approaches the entrance to the PA hall, Atac catches sight of Sin'thea and mouses over her avatar to double check the name (line 04). Although she is still some distance away and Sin'thea is facing away, Atac nonetheless waves to her friend (line 07). Although Sin'thea most likely cannot see Atac's avatar and although it does not even actually wave (because the waving animation is overridden by the walking animation), she can still see the text emote, "Atac waves to you" (line 07).

08 16:20 Kimon Calari nods at Sin'thea.
09 18:00 ((Atac approaches Sin'thea from behind))
10 19:30* Sin'thea waves to you.
11 ((Sin'thea's avatar automatically spins around
12 and waves))
13 21:06 Goldtooth greets you.
14 22:24 Sin'thea: ATAC!!!
15 22:34 Atac: hiya
16 ((Atac's avatar waves automatically))
17 23:22 Sin'thea gives you a hug.
18 ((Sin'thea's avatar automatically gives half a hug))

Atac stops on the steps behind Sin'thea just in time to receive a return wave (line 10) and the two friends continue their greetings (lines 14-18). So in this case, the two players rely almost entirely on the text emotes that accompany gestures in achieving mutual orientation rather than on their avatars. In such cases, which are not uncommon, the text emotes render the avatars more or less irrelevant.

In a similar vein, gesture text emotes can also destroy some of the subtlety and indirection in interaction by making actions overly explicit. For example, in EQII, your avatar will do a series of sexy modeling-like poses if you type "/flirt." But it also announces in text that "You flirt shamelessly with X" thus categorizing your action explicitly as "flirting." So much for the subtle dance of seduction! Similarly in SWG, if you type "/wink" (more a facial expression than a gesture really), your avatar winks, but the system also produces the message: "You wink suggestively at X." I found this very problematic because often when I tried to wink at someone to indicate that I was joking, it came off instead as a come on due to the verbiage of the text emote. In real life, embodied gestures afford much more strategic ambiguity.

So should developers dispense with the text emotes that automatically accompany gestures animations?

YES... but perhaps not until they fix other problems with gesturing in virtual game worlds. Text emotes currently serve as a kind of Band-Aid for gesture systems that are broken. Without text emotes, players are much more likely to miss gestures directed to them. In real face-to-face, performing a gesture successfully requires making sure the intended recipient can see it. This involves seeing where the recipient is looking and holding (or prolonging the duration of) the gesture until the recipient has displayed some sign of recognition and understanding. This is not possible in most current avatar systems. They don't indicate to you when the other player has detached his view from his avatar's orientation (i.e., by panning and zooming); they don't indicate when the other player's view is obscured by menus (e.g., maps; except in There); and they don't allow you to control the duration of your gestures (see #1, #6 & #8 of my 10 thing about avatar interaction). So until such features are implemented, we may have to make due with systems telling us about gestures rather than better enabling us simply to see them.

Posted by Bob

Posted by at 03:21 PM | Comments (4) | TrackBack

February 17, 2006

The Level 60 Game

Anecdotally and from our own experiences, the game at level 60 is entirely different from the game pre-60. For one thing, level advancement is no longer the goal and most guilds become raid and instance oriented. We wanted to get a sense of this shift with numbers. And also, we wanted to see whether this is a gradual shift starting at level 40 or level 50, or whether this is indeed a drastic shift that occurs at level 60.

We decided to look at this via social network metrics. How different are characters in guilds at different levels? For this, we calculated the social network metrics (density, centrality, and combined connection times) for each character and found their means according to their level range.

The data suggest a sudden shift at 60 rather than a gradual change. Here are the 3 graphs showing the difference for the 3 metrics mentioned.

Server Sample: RP (High), PvE (High), PvE (High), PvP (High), PvP (High)
Sampling Period: Month of January
Sampling Resolution: ~12 minutes
Parsing Method: The sample unit is each unique character. Each character was tracked across the server logs. Total playing time, lowest observed level, highest observed level, guild affiliation, and zones seen in were parsed.
Data Filter: Only those characters who are in a guild.
Sample Size: 179,003 characters

Posted by nickyee at 12:36 PM | Comments (8) | TrackBack

February 16, 2006

Devs'll find work for idle avatars

Ever think character avatars in MMORPGs look a bit spaced out? Just standing there, kind of alive but completely entranced? Whether avatars appear stiff or lively certainly impacts our ability to mistake them willingly for living beings and lose ourselves in the fiction of the game world.

Now one strategy that game developers use to try to breathe some life into avatars, is to use "idles," or animations of autonomic behaviors, such as breathing, blinking and swaying that are activated when the avatar is idle. For example, the idles in EverQuest II are especially subtle and realistic, while those in World of Warcraft are highly stylized and cartoonish. In both cases, they make the avatars seem more like animate objects.

While idles indeed do a nice job of livening up avatars to a certain degree, there's still something missing. In public, people don't just breathe and blink, they do things. Things that are responsive to the particular social setting and to the particular people on the scene. Take people at a bus stop, for example. They glance at their watches and stare off in the direction of the awaited bus, displaying to everyone around that they are doing "waiting for the bus." They may step inside the bus but not feed the coin machine immediately, instead projecting this action by visibly rummaging through a purse. They crack a book or jump on their mobile phone, displaying a reduced availability for small talk to fellow passengers. And so on. This tacit level of communication is not yet implemented in virtual game worlds.

In games, much of the time when avatars are "idle," players actually are not. They are busy with a whole array of personal (in-game) activities, similar to those in real life, such as rummaging through backpacks, consulting maps, searching spell books or directories, reviewing skill trees, chatting privately, and more (see earlier post). So players are already engaging in embodiable activity that is appropriate to the situation. Simply animating this would go a long way in terms of making avatars appear more like living, competent beings than spaced-out zombies. Characters could actually look "busy" and do so in a way that is socially meaningful and appropriate to the situation, rather than canned.

Now animating such personal activities not only makes avatars appear more lively by enabling others to make sense of what they're doing, it also provides important cues for achieving coordination. Hiding player activity from public view, on the other hand, can cause slippage. For example, in the following excerpt, two players in Star Wars Galaxies are doing a set of missions together to gain access to Jabba the Hutt's palace. They have just completed one objective, killing a kaadu, and are about to travel by speeder to the next (note: double parentheses mark transcriber's descriptive notes).

[Star Wars Galaxies: Jabba's Missions; Atac's perspective]
01 01:38 ((Nike runs past Atac toward speeders))
02 02:44 Atac: is that it?
03 03:10 ((Atac turns around and looks in Nike's direction))
04 04:16 ((Atac begins to follow Nike))
05 09:34 ((Atac mouses over an empty slot in her toolbar))
06 12:06 ((Nike mounts his speeder and begins turning it))
07 13:46* Atac: wait
08 13:52 ((Nike stops turning speeder))
09 15:15* ((Atac opens inventory))
10 17:24 ((Atac opens backpack within inventory))
11 20:52 [GroupChat] nike: did you get it?
12 28:48 ((Nike turns speeder))
13 29:46 ((Nike begins to move forward))
14 30:24 [GroupChat] atac: get what?
15 31:56* ((Nike passes Atac on his speeder))
16 32:16 ((Atac mouses over and inspects some stimpacks))
17 37:28 ((Atac drags stimpacks))
18 38:54 ((Atac drops stimpacks onto empty slot in toolbar))
19 40:18 ((Atac closes backpack))
20 41:00 ((Atac turns around to look for Nike))
21 42:12 [GroupChat] nike: the kaadu egg...

Here Atac tries to coordinate two activities: replenishing the stimpacks in her toolbar (used for healing in combat) and traveling with Nike to their next destination. When Nike appears to be about to leave the scene on his speeder (line 06), Atac types "wait" (line 07) and then opens her inventory about 1.5 seconds later (line 09). However, there is no avatar animation or public signal to Nike that she has done so. As a result, he can't see what he's waiting for. So he takes a guess, "did you get it?" (line 11) referring to a quest item from the dead kaadu. Nike then starts to drive off (lines 12-13) as Atac requests clarification by saying "get what?" (line 14), which shows that, that was not in fact the reason for the waiting. Nike drives off (line 15) and Atac selects some stimpacks and drags them to her toolbar (lines 16-19). She thus fails to get Nike to wait for her while she rummages through her purse. This may seem like a minor slippage in coordination, but it can have more major consequences if, for example, you lag behind your group mates in a dungeon while you search for an item or a spell and a mob pops and kills you (this has happened to me on a few occasions). In general, such slippages make the avatar-interaction system feel awkward and clunky.

Now game developers are not entirely unaware of the social importance of animating personal activities. In There for example avatars don goggles when the player opens a web page or a headset when the player opens an IM window. In WoW, avatars bend over when opening a chest, showing everyone who is going for the loot. EQII does a nice job of animating the various types of harvesting: avatars bend over when "gathering," squat slightly when "trapping," chop with an axe when "foresting," and swing a pick when "mining." This not only displays publicly what a player is up to when lurking about in the bushes, but also displays to other would-be harvesters that he or she is indeed going for the insidious root, instead of the nearby lithic stone. And of course, most games have long had combat animations that show group mates that the tank is starting to pull or the healer is starting to heal. All of these avatar manifestations of player actions greatly enhance players' ability to interpret what each other is doing and tightly coordinate their actions, as well as make our virtual bodies seem more alive. Developers should expand the use of these kinds of personal animations. (Raph, please give us more!)

Imagine that whenever you type /time or click on the in-game clock, your avatar looks at his or her watch (hourglass, sundial, whatever). Now when it's 3:00am and I finally need to ditch my group, I can forecast my departure a bit more subtly than simply typing, "I gotta go to bed." Instead my group mates can draw the inference, "you gettin tired or something?" Or imagine that whenever you open your map, your character also opens and studies a map. Not only does this display to your group mates that you're not ready to start the next attack and the like, but it also enables passersby to notice what you're doing and offer assistance ("Are you lost?"). Such seemingly small mechanisms will create a more intricate texture of the social fabric in future virtual worlds.

Now players might argue that they don't want others to see that they're consulting a map ("I don't want to look like a noob!"). That may be true, but is the reduced exposure really worth sacrificing a whole layer of social life? Due to the limitations of MMOGs to date, we have become accustomed to an odd sense of public privacy in which we can hide much of what we're doing while in the virtual presence of others (e.g., we can even "talk behind someone's back" right in front of their face). But a fundamental part of human sociality is seeing others and being seen by them, incompetences and all.

Posted by Bob

Posted by at 04:08 PM | Comments (9) | TrackBack

February 08, 2006

Why players adopt 3rd party VoIP apps

One reason, although certainly not the only reason, players of MMORPGs use 3rd party VoIP applications, such as TeamSpeak or Ventrilo, is that standard text chat is often too slow and cumbersome for many activities that require tight coordination between players, such as raids or PvP. While text chat is good for certain kinds of activities in games (e.g., chatting across zones, advertising wares), it is not so good for real-time joint activities (e.g., traveling together, fighting together). There are of course a few different reasons for this, most of which Pavel Curtis pointed out years ago. One is that the typing itself is somewhat slow (especially compared to speaking). Another is that text chat increases the burden on the users' hands, with which they may also be wanting to do other things at the same time, such as firing off a spell or navigating their avatar. Yet another reason is that most games (with the exception of There) implement standard text chat in which the composition of turns is kept private. In other words, turns are posted on a message-by-message basis (rather than a word-by-word or character-by-character basis). This creates interactional lag...

Now although players can be quite good at managing this interactional lag, it can at times cause slippages that impact their joint activities. If we look at avatar interactions under a "microscope," we can see how this slippage occurs. The following transcript (or annotated chat log) comes from a screen-capture video collected from EverQuest II. It begins as Ataca and Rattington are roaming around outside the gates of West Freeport looking for some animals to hunt together for xp. (Note: double parentheses mark transcriber's notes and italics mark system-generated messages.)

[Hunting Armadillos: EverQuest II; Ataca's perspective]
01 07:38:16 ((Ataca stops running))
02 07:38:54 ((Ataca targets an armadillo))
03 07:41:38 ((Rattington stops next to Ataca))
04 07:41:38 Ataca points at a banded armadillo.
05 ((Ataca's avatar points)
06 07:43:06 ((Ataca rearranges icons in her toolbar))
07 07:50:00 ((Ataca mouses over an attack spell))
08 07:51:00 ((Ataca mouses over armadillo))
09 07:54:16 ((Ataca initiates an attack by clicking a
10 spell icon in her toolbar & her avatar begins
11 a spell-casting animation))
12 07:57:42 You surround A young armadillo with arcane chains!
13 07:58:18* Rattington says to the group, "I see two that
14 are grouped but I think we could take them."
15 08:00:10 ((Ataca clicks another spell icon))
16 08:01:58 Rattington hits banded armadillo for 4 points of
17 crushing damage.
18 08:03:34 You strike A banded armadillo with a storm of lightning!
19 08:04:08 Rattington hits a banded armadillo for 3 points of
20 crushing damage.
21 08:05:02 Ataca says, "hehe"

In this episode, Ataca approaches some banded armadillos, selects them and "points" to them (lines 01-05) by clicking on a toolbar icon for "pointing." In this context, such an action can be seen as a proposal to attack the armadillos, although the more standard practice is to "hail" them by selecting them and pressing the H-key. Rattington can observe Ataca's pointing by virtue of an avatar animation (line 05) and a text emote (line 04). Ataca then privately rearranges icons on her toolbar (line 06) for about 8 seconds, while Rattington appears to do nothing. Ataca then proceeds to initiate an attack on the armadillos with a click of a spell icon on her toolbar (lines 09-11). Rattington can observe this by virtue of a spell-casting animation of Ataca's avatar, an alert sound, and eventually a text message after the spell is cast (line 15). But then an interactional slippage becomes apparent. Four seconds after Ataca has initiated the attack Rattington says, "I see two that are grouped but I think we could take them" (lines 13-14). In other words, he offers his assessment about whether the armadillos are suitable targets; however, it appears too late to be consequential for the attack due to the interactional lag caused by the chat system. He almost certainly started typing his turn before Ataca initiated the attack, but she could not see his turn unfold in real-time. Rattington then joins in the attack (line 16), and Ataca chuckles at their apparent lapse in coordination (line 21).

These two low-level characters do in fact eventually defeat the two perturbed armadillos without much difficulty. However had this been a 60-person raid against an epic dragon or a PvP encounter against an opposing team of players, this kind of lapse in coordination can have greater consequences. In such situations, having a real-time medium for talking with your teammates, like voice (VoIP or co-presence) or even word-by-word chat, is a distinct advantage. Tighter coordination between players can be achieved when avatars, which move in real-time (or nearly real-time), have voices to match.

Of course, VoIP does not magically solve everything. It creates interactional issues of its own, but that's a topic for a future post. Some of these issues will no doubt be discussed at the panel on "Community and Communications in Massive Multi-Player Online Games" at VON 2006, Tuesday, March 14, 2006, 1:00pm - 2:15pm in which Nick Yee and I will be participating.

Posted by Bob

Posted by at 11:41 AM | Comments (8) | TrackBack

February 02, 2006

Centrality, Class and Gender

After the previous analysis, we ran an additional one that included the character gender variable. Here, our results were puzzling. Across all of our metrics, male characters were better connected than female characters. And this was true for all classes, with the only exception of Priests. In other words, male characters of all classes are better connected than female characters of all classes, except for female Priests, who are better connected than male Priests. This gender difference was clear and consistent across our three measures of centrality.*

We then ran several analyses to filter out possible explanations and help clarify what may be happening:

1) Are male characters typically in larger guilds than female characters? We ran a quick t-test. While significant, the difference was between 51 and 55, so it can't really account for the difference we're seeing.
2) Do male characters play more than female characters? Again, the t-test was significant, but the difference was insubstantial (1437 vs. 1481 minutes).
3) Do male characters group more than female characters? The t-test here comparing grouping ratios was not significant.
4) Are male characters higher level than female characters? The t-test was significant. The difference was between 33 and 35 - an insubstantial difference.
5) Are there more level 60 male characters than level 60 female characters? 25% of female characters were level 60. 28% of male characters were level 60. Again, a insubstantial difference.

So we're at a loss as to why we're seeing the pattern we're seeing. In every respect we can measure, male and female characters seem to be largely equivalent. Thus, we have two findings we're not sure how to explain. First of all, why are male characters better connected? And secondly, why are female Priests the exception? Any suggestions?

*Server Sample: RP (High), PvE (Medium), PvE (High), PvP (High), PvP (High)
Sampling Period: Month of November
Sampling Resolution: ~12 minutes
Parsing Method: The sample unit is each unique character. Each character was tracked across the server logs. Total playing time, lowest observed level, highest observed level, guild affiliation, and zones seen in were parsed.
Data Filter: None
Sample Size: 179,003 characters

Posted by nickyee at 12:35 PM | Comments (12) | TrackBack