Semi-Autonomous Avatars and the Metaverse

There's a moment when you log back in to single player mode in Grand Theft Auto.

The camera pulls back. Your character (Franklin, say) is walking out of a store and is waving goodbye to someone off camera. Then the camera slowly moves into a new position, hovering just above and behind Franklin, locking itself into third-person "game position".

It's a powerful illusion: first, that the game world was persistent: you might have logged off, but life in Los Santos went on without you. And second, that your avatar also lives a life of its own when you're not around. Sometimes you log back in and he's coming out of a movie theatre or cruising women on the street or exiting a convenience store.

The camera snaps back into place and you now inhabit the game character. You've taken over the controls.

Persistence in Games and Virtual Worlds

The GTA moment was seminal because it helped to reinforce the idea of persistent worlds. It provided a hint that when we log out, the worlds will continue without us.

And persistence is one of the key definitions of the Metaverse (a term which has otherwise become a sort of collective emblem of a shift in technologies rather than a specific destination).

More recently, a game like Rust has carried persistence into a deep game mechanic: the assets you create will be destroyed or stolen by other players when you log off. And so players band together, share calendars and set up schedules to guard their forts around the clock. The fact that the whole world gets reset once a month just adds...I don't know, a sense of existential futility or something.

An upcoming game like Seed (which I've come to believe will help us imagine new paradigms for the Metaverse) will drive that persistence deep into layers related to economies, wellness, politics and culture.

And so world persistence is profound on its own. But if your concept of persistence is driven mostly by multi-player game platforms, then you're probably missing the deeper point: that persistence speaks to it being a world, which indicates something which isn't static, which changes and does more than deliver a series of grinds and quests according to a pre-determined schedule.

Sure, when you log into GTA or RDR there are already people online doing stuff, but the world itself hasn't particularly changed since the last time you logged in.

Minecraft is more world-like because its persistence is coded right down to the atom. Everything is subject to change. Everything can be re-shaped by other players. By the time you log in again, someone will have opened a portal or built a castle.

But what kind of 'world' is it when the people who inhabit it can stop time? Why is it that if the world is persistent, its citizens can simply log in or log off?

Should our avatars be static in worlds that will increasingly have the physics, economies and environments of the physical one (or imagined versions of the same)?

Avatars and Characters: What's The Difference

For me, the GTA moment was more profound for its hint that our avatars might have lives of their own.

But before we go there, we should take a brief moment to note the difference between an avatar and a game character. And maybe it's enough to say this:

  • When a 'gamer' enters a virtual world, they often think of their avatar as their 'character'. They will talk about the avatar in the third person - "my character". They will talk about playing. There is a remove between the player and their representation.
  • But at some point, you see a shift (at least in a fully realized virtual environment). Their 'character' becomes YOU. It is not some third party.  You might be controlling that emblem of yourself. but you're not 'playing' it.
  • There are neurological reasons for this. The brain has difficulty distinguishing between the physical and virtual manifestations of our 'selves'
  • And so, when I talk about semi-autonomous avatars, I am not talking about game characters who are part of some story in which our sense of agency is limited.

Franklin in GTA is a character. You might identify with him, you might immerse in his life story, but he isn't YOU.

Set up a new 'toon for GTA Online, however, and you're getting a lot closer to being an avatar. The 'character' you use to play Fortnite is an avatar (especially when you spring for the skins) even if its life is mostly a series of grinds and the occasional Kaskade concert. And certainly when you log-in to Spatial.io your representation is clearly a version of you.

As we spend more and more of our time in synthetic worlds, these avatars, these extensions of ourselves, are US. Your avatar will have closets filled with clothes and NFT-backed sneakers, you will live in a $500,000 virtual house, and you'll head to a concert with 1.2 million other avatars (sharded, but still). Or...maybe not YOU (or ME), but someone will!

Will you own these virtual sneakers? Or your avatar?

Semi-Autonomous Avatars and Why It Matters

There are ideas we have about the Metaverse. Some of them have become so firmly ingrained that we stop questioning the assumptions.

The idea of a semi-autonomous avatar challenges a few of those assumptions:

That the Metaverse is a VR-only experience

This comes, of course, from Snow Crash and Ready Player One. The idea that we slip into our avatar like a skin. That an avatar only exists when we don a pair of goggles and log-in. That the correlation between how our body moves and the movement of our avatar is one-to-one.

But the Metaverse will defy the gadgets that we use to access it. As I've written previously, many people will have their first experience of the Metaverse while sitting in their car.

A semi-autonomous avatar reminds us that there will be 'instances' and sections of the Metaverse where our digital selves can act at least semi-independently of the devices we wear. We might be able to observe or control them through things other than glasses.

That we will want to move seamlessly between worlds

This is a key tenet/conventional wisdom of the Metaverse. It's this idea that we want to log in to some kind of waiting room and use it to pop in and out of a constellation of virtual worlds.  

I've never entirely understood this idea. I suppose it's driven by our experience of the Web - as if we move around Websites seamlessly (when in fact there are still a million friction points that prevent our identities, wallets, permissions and 'inventory' from travelling with us as we surf).

Regardless, it doesn't necessarily solve a clearly identified problem. WHY do I want my avatar to jump from Minecraft to Fortnite again? And if I could LEAVE a version of my avatar back in Minecraft to guard my farm, wouldn't I?

I get the idea of IDENTITY. But more often than not most users prefer to slip between identities rather than be burdened with a single one. It's no different than the personas we 'wear' as we move from home, to work, to community. We bring different selves.

This isn't to say that we shouldn't create standards or that we shouldn't be able to bring our avatar from one world to another. It IS to say that there are other use cases as well.

That the Metaverse is a "lean-forward" medium

It's a 'truth' given that there are only two types of media: lean back and lean forward.

Matthew Ball has reinforced the dichotomy between two types of companies as he looks at the future of entertainment:

"Just as gaming seeks Hollywood to adapt their stories in order to build love, Hollywood seeks out gaming to adapt theirs. But in this latter case, Hollywood faces existential threats".

Hollywood can create love. But in his calculation it hasn't mastered the art of the lean-forward experience.

In a macro sense, this division might be true. But it ignores the very messy middle: the worlds which aren't love. Which don't even require much attention.

Pop into GTA these days and listen to the chat. If you run into a group with any experience at the game, you get the sense that they're barely paying attention: they're running another supply quest for their motorcycle club but mostly trash talking other people in the channel.

Attend a virtual dance and you can't even be sure half the people have their eyes on the computer. They're probably watching Netflix or streaming onto Twitch instead where they're chatting up their superfans.

In fact, I'd propose that there is a significant majority of content that is successful because it allows for split attention. Your avatar is there, it's dancing, but there isn't really anyone home.

The semi-autonomous part? At least it has some good pre-recorded dance moves.

The Metaverse is Entertainment

Which brings us to a final myth (although I could go on): which is that the Metaverse will be entertainment.

If I can imbue my avatar with a set of automations, it can also perform tasks. An avatar that can perform tasks, in worlds which will have their own economies, is an avatar that can make money.

Large chunks of the Metaverse will have economies and auctions and shop keepers and fashion shows. It will have round table discussions on the state of bitcoin and mini stock markets where you can trade NFTs.

But even putting that aside, the speed with which automation is becoming a key underpinning of the Internet itself means that the Metaverse will adapt those same technologies.

I have a workflow which connects Tweets to Airtable and over to Notion and then back to an email reminder system and ToDoIst item. I use Automate.io to hook it all up. I take a single action and it creates a cascade of value through a series of systems.

All I need to do is hook it up to GPT-3 and it could maybe even just auto-generate these posts! I'd have a fully functioning enterprise that required almost zero human intervention.

DAUs will be set up in the Metaverse. They will mostly run themselves and exist entirely in a synthetic world. We will able to participate in them (or our avatars will) and we will be able to vote and take actions. And some of those actions we'll be able to automate.

In other words, because there will be economic value in the Metaverse, many of us will want to maximize the value that our avatars create.

The Ambient Metaverse  

Currently, the idea of a semi-autonomous avatar brings scripting hacks and automated farmers to mind. They're considered hacks because they're associated with game environments and are used to bypass the (written or unwritten) rules.

Or, they're not considered ways to make an avatar autonomous: the macros you use in Warcraft are just enhancements. Primarily because they're seen to aid the player....who is controlling a character.

But our avatars will end up with all sorts of macros and sub-routines. They will be able to act a bit on their own and give the appearance of presence.

I know people who leave their avatars logged in and resting in a virtual bed while their human controllers sleep. They feel a need to send a signal to others in the synthetic world that they are 'present' even if the human behind the avatar is asleep.

On the other end of the spectrum, our avatars may be mostly invisible. We'll move through virtual worlds seen through our glasses or while driving our cars. In those cases, the autonomy of our avatars will have some real meaning because the sub-routines that they perform will be a large point of our presence in those corners of the Metaverse.

We have lived with the myth of Ready Player One: a lean-forward, entertainment and game-focused 'Metaverse' (owned, mind you, by a benevolent dictator) that you log into when you throw on a haptic suit and some goggles.

The reality is that the Metaverse will often be ambient. We'll skim it. We'll dip in and out or barely even notice it. It will be always on and it (and our avatar) will live a life of its own whether we pay attention to it or not.



We're in this together. I'd love to hear your thoughts.

Let's start a conversation. As a subscriber, all you need to do is hit reply.