Consent systems

I was asked on Twitter recently for references for my mention of “consent systems” in my talk on “What AR and VR can Learn from MMOs.” I didn’t have any handy at the time but today I had some free time so I went looking.

Textual antecedents

The basic concept can be found throughout roleplay social virtual worlds such as MUSHes. (For example: Black Ops MUSH, The Lady’s Cage MUSH, Star Wars Omens). These sorts of worlds typically do not have combat systems, and rely heavily on free-form emotes (though the commands there more often have the syntax “pose” or “emit” or the like). Like any other full roleplay environment, of course, fights and conflicts happen all the time.

This description of how it happens in practice is pretty good:

You can pose an ATTEMPT:

Reginald smirks, then brings an arm up to lean on Nidonocu.

And wait to see if they pose a reaction:

Nidonocu glances down at the arm as it rests on his shoulder.

-or-

Phineus side-steps, avoiding contact with the arm.

— Pokemorph.com wiki

Note that the above are all done with the free-form command. On most MMOs today, as well as in most MUDs, this might be called the “emote” command:

/emote does whatever they want.

Raph does whatever they want.

But of course most worlds also offer a huge array of pre-baked emotes, most of which permit targeting at other people, and which the recipient cannot control. Here’s a list from World of Warcraft, which includes stuff like

You bite <target>. Ouch!

You scratch <target>. How catty!

You cuddle up against <target>.

You bonk <target> on the noggin. Doh!

You brush up against <target> and fart loudly.

You give <target> a high five!

You lick <target>.

You massage <target>’s shoulders.

You pounce on top of <target>

You slap <target> across the face. Ouch!

You tickle <target>. Hee hee!

The basic contrast here is centered on who has the authority in the interaction. A standard MUD approach would grant the authority to the actor, not the target. A standard MUSH approach would grant authority to the target, not the actor.

The reasons for this divergence are complex and cultural; many MUSHes and MOOs would lean heavily on free-form emotes and poses, and would often have deep and intricate social behaviors heading well into extremely intimate territory. Players would form extreme attachments to personas and avatars, seeing them as proxies for themselves. The whole point of the game was social: roleplay, collaborative storytelling, or in OOC scenarios, of course, getting to know one another for real. Having a stranger come up and lick you (ewww!) in the middle of a Regency ballroom dance, let’s say, would be profoundly jarring. Hence, you see lengthy documents defining proper codes of conduct for participation in the MUSH space, such as this example from Arx.

1. Every player has agency over their own characters’ attempted actions, but not the consequences of the attempt.

This is something fairly universal to most role-playing games, but still often gives newer players some trouble. With the pose and @emit commands, players have the ability to write or describe anything, but really need to restrain themselves from describing the results of non-trivial actions. In other words, one character trying to hit another or attempt some difficult feat cannot just assume success. This would be decided by code built into the game (such as combat), the decision of staff (a GM arbitrating a decision), or the consent of all effected players in a scene (friends RPing a social scene and just making rolls as desired for randomizing results, for example). It is not appropriate to describe the actions of another character, or the results of actions upon them, since it removes agency from the other player. Players not having control over the results of their actions means their characters will have to deal with undesired consequences of actions, such as attempting something very dangerous and dying in the attempt. Staff fully knows how difficult players often take character death, and we don’t treat it lightly, but risk is necessary to add meaning to some storylines.

In MUDs, the point of the game was bashing orcs and killing giant spiders. Player characters die all the time, and it often means very little. The pace of the game was often far faster, and the array of “socials” (as the canned emotes were often called) were used in preference to the free-form emote command. An action like licking someone was taken extremely lightly, generally, and seen as an expression of utter goofiness, because people’s personas are held at a farther remove.

Though I never saw a set of socials that did this, it’s pretty easy to envision canned emotes that respect the privacy and personal space boundaries inherent in the spirit of the typical MUSH code of conduct. It would be pretty easy to simply edit the WoW canned emotes to be attempts instead of successes:

You try to bite <target>.

You offer to scratch <target>. How catty!

You attempt to bonk <target> on the noggin. Doh!

(I tried rewriting the “lick” one here and pretty much failed at coming up with something that wasn’t gross).

This would leave the decision on how to respond to the target, and you can even envision an “/accept” command that could play the second half of the sequence. You could even make the first half simply not happen until /accept is performed, and then play the original WoW text.

Adding graphics

When we moved to graphical worlds, some of this became obligatory if we wanted to display it using character animations. A canned “hug” emote will result in someone just sticking their arms out and waving them about; unless the target is within range and positioned correctly, the animation will be to empty air. And if someone isn’t also reading the chat box, it will be nonsensical. You can either choose to simply not animate all of these powerful social tools… or you can solve the problem.

It’s worth a brief digression here on why it can be worth it to do the expensive thing, and try to solve the problem. In a graphical world, the textual form of all of these is dramatically weaker in terms of emotive impact. (In a textual world, this is very much not the case). Graphics pull the eye, and a huge amount of our brain processing power is designed to cope with all sorts of subtle social cues. I’ve written about this before at length, and we can see how little work has been done in the field since. But social bandwidth is a powerful tool to improve player retention. It increases connections and commitment between players, and reduces the degree to which they objectify others (thereby encouraging better behavior and less griefing).

In Star Wars Galaxies we went to great lengths to create as socially rich a communications system as we could; we still used the MUD-style impositional emotes, but tied as much as graphical support in as we could. We detected words embedded in your chat in order to affect body and facial animations. We attempted to set avatar eye line based on who was speaking most recently. We included a rich moods and adverbs system. The screenshot to the left shows off several of these things, including selection of chat bubble style as an additional channel to convey emotional content.

In Metaplace, we had a far far poorer animation system because we were working with 2d images, sprite frames, streamed over the network. This meant adding new animations was extremely challenging and expensive. I asked for a very limited set of social actions: waving, pointing with arm outstretched, and sitting. And I built a hug system out of the point animation, after there was a bit of an outcry from players for it.

But Metaplace was a social world, not a MUD. The culture of the space was far more like a MUSH or MOO than it was like a hack n slash dragon slaying space. To make the hug work, we needed an automated consent system. The idea isn’t new; getting added as a friend in any game works the same way, and of course even Facebook offers consent systems for certain social activities, such as tagging people in photos or posts.

The flow of it went like this:

• Buffy initiates a /hug command on a target Bob.
• Bob receives a pop-up informing them that Buffy desires to hug him.
• Bob accepts or rejects. If reject, nothing happens. If timed out, nothing happens.
• If Bob accepts, Buffy is notified.
• Buffy pathfinds to Bob.
• Upon arrival at a precisely calculated position relative to Bob, Buffy plays the arm outstretched emote, but also notifies Bob of arrival.
• Upon notification, Bob pivots in place to mirror Buffy’s position. The two overlapping avatars and outstretched arms overlap in such a way that it looks like a hug.
• Any movement by either player breaks the tandem animation.

A /kiss system was added as extra support on top; it did exactly the same thing, except that it played little floating hearts over the couple.

This sort of thing is what is done in pretty much all forms of fighting game anyway. When a player in a fighting game executes some cool combo that results in grabbing the opponent in some way and slamming them to the ground, it’s termed a tandem animation; you have to get the two characters precisely positioned (often by “cheating” the real physics) in order to play the awesome cool effect with both characters moving in synchrony. This is why you typically lose control of the avatar while something like this is going on; we can’t let you have control and also have it look good. (It can also be used in pro-social ways, as in the synchronization of dance moves in Star Wars Galaxies).

LARPing

What I was advocating for in that talk which prompted this little essay, was for systems like the one in Metaplace or the one in MUSHes to be considered when designing social VR. In VR we rely heavily on the graphical equivalent of /pose and @emit, with actual physical movements tracked with the headset and controllers providing the “freeform” aspect of the input. Social VR doesn’t currently use canned commands all that much outside of facial expressions, because hands and heads can be so expressive. On the other hand, head and hands alone is of course going to hugely limit the canvas of possible social interaction, so I wouldn’t be at all surprised if more advanced avatar puppeteering would someday require either freeform tools for the body (VR suits or body tracking) or avatar control tools that work more like the systems in Galaxies and Metaplace.

But if social VR is going to lean on freeform input, then it needs a consent culture akin to that of the MUSH world — and actually, probably one even more robust than that, and that’s where the LARP enters the picture, particularly the Nordic LARP.

Nordic LARP is not just people dressing funny and running around in the woods shouting “Fireball!” at one another. It is used in school curricula in Scandinavia; it is used for therapy, for historical education, and for anti-bullying lessons.

The concept of consent can extend well beyond the simple examples I have discussed here. It essentially extends all the way through your Code of Conduct. In the Nordic LARP community there has been a lot of discussion and attention towards this topic. Here’s the briefest possible dip into the subject, from “The Consent and Community Safety Manifesto” on nordiclarp.org:

The 10 Principles of Consent and Community Safety

1. People are more important than larps. Larp organizers and communities must value the safety of their participants as their number one priority.
2. Each person’s body is their own. They alone may set their boundaries and say what makes them comfortable…

The 20 Statements of Support for Community Safety

1. I will give a clear and honest “yes, please” or “no, thank you” when I am asked for my consent, and negotiate more specifics if I feel they are needed.
2. I will respect the boundaries another person sets and accept that my boundaries may be different from someone else’s.
3. I will not touch another participant in-game or off-game, without their consent…

The 10 Larp Designer Commitments to Community Safety

1. With my community, I will create and maintain a Code of Conduct that outlines expected, encouraged, and prohibited behavior.
2. With assistance, as needed, I will create a harassment policy and reporting procedure for my larps which condemns harassment and establishes a clear and confidential way for participants to come forward if it is violated.
3. I will designate one or more people on my organizer team as Community Safety Coordinators and give these people the resources and respect they need to conduct Community Safety business…

— heavily excerpted from nordiclarp.org

This has, of course, a lot in common with codes of conduct and EULAs and TOSes found elsewhere, with the notable difference that it offers that set of designer commitments; that of course calls to mind older rights documents for players.

Once you start thinking in terms of consent, it can force you to think in terms of consent up and down the entirety of the social system for your world. LARPers simply bump into the issues much faster, and therefore have tended to discuss them relatively unprompted, precisely because real-world engagement greatly increases vulnerability.

The bottom line

As social VR develops, it would be wise for its developers to think on the ways in which they are effectively treading old ground. I’ve singled out MUSHes, MUDs, MMOs, and LARPs here as models, but in point of fact, any public space designed for play is going to have developed best practices that are likely relevant. Theme parks, wilderness areas, and concert venues likely have applicable processes, cultural signposts, and policies too. To host a multiplayer virtual environment is to host and govern a space, and there are no shortage of models precisely because spaces are where we normally live.

And heck, it’d be nice if Facebook had a few more consent systems too. Adding people to groups without their consent? That’s a dark design pattern…!