One way that I’ve been thinking about the role of developers in Autonomous Worlds is a notion I’m calling embodied admin rights. It basically means something like:
“when developers intervene in their virtual world, they should do so in the most diegetic way possible”.
It comes as a solution for the following problem: The virtual worlds of the future will probably be very complex pieces of technology; meshing together infrastructure, smart contracts, and game logic to ensure the liveness and validity of the game world.
Accordingly, the first generation of worlds will likely be on upgradable “training wheels”, allowing developers to patch any bugs found post-launch.
At least, this is the situation with rollups, with billions of dollars of value locked up in essentially open betas, where developers can cease operation, censor users, and even submit invalid transactions. This is arguably a necessary evil (with various solutions proposed), but in any case, it’s clear that upgrade keys will be here for the foreseeable future.
I posit that the situation will be the same for Autonomous Worlds but 10x worse. Once all the physics-breaking, “invalid state” bugs have been patched (eg. infinite money, failed collision detection, corrupted data), there exists a long-tail of phenomena that are technically valid within the world's rules, but not congruent with some overall goal (be that fun, player advancement, variety of experience etc.).
For example:
A group of high-level players griefing everyone 24/7
In-world hyperinflation or some other economic collapse
A player losing their private keys
An accidental explosion that destroys a historic player-built monument
Maybe you believe that, unlike a blockchain, Autonomous Worlds should be opinionated about the transactions that take place on their ledger. Hence, “admin rights” that would allow developers to intervene in harmful situations. Of course, this intervention by definition conflicts with player agency and the world's physics, and therein lies the debate.
While some would prefer a unforgiving digital world - and these worlds will be much, much less forgiving than traditional games - the reality is that “Player #13562” who reads about the game on Twitter, starts playing, and is immediately killed by some high-level player, will just rage quit and never play again. Indeed, they might turn to a centralized competitor, or (gasp!), play a cozy traditional game.
I’m not convinced that this is a case of “just progressively decentralize”, either. These worlds will be a mind-boggling mix of human players, AI, and smart contracts, interacting in a highly systemic sandbox environment, over the course of years. Can you ever be sure that your world won’t devolve into a dystopian hellscape?
Whether any of this really justifies centralized control is up for debate. Regardless, I’ll suggest one approach that will at least make things less confusing for players.
Consider the following fictional scenario: Blocks is an on-chain sandbox world, where players can mine raw materials and use them to build structures. Movement is embodied; players walk at a finite speed, can collide with their environment, and may only interact with entities in their local area.
Blocks has recently seen a spike in popularity, but player retention is low, with newcomers being overwhelmed by the game's mechanics. The game is client-agnostic, so the developers can’t force players to read a guide, or play through a tutorial.
Instead, the developers have chosen an in-world solution, opting to build an information center near the spawn. It will act as an “embodied learning experience”, allowing players of any background and language to physically see and interact with the game's mechanics.
The developers need to build the information center quickly. Consider the following two ways that they could employ admin privileges:
Developer 1 manually edits the smart contract storage corresponding to the blocks at the build site. The information center materializes out of thin air. Nearby players are confused as to where the building came from and the low-level transaction doesn’t render on the game's block explorer.
Developer 2 grants themselves the ability to fly, and also mine blocks faster than a regular player. They soar over to a nearby quarry and quickly mine all the necessary materials. Then, they fly back to the construction site, and start building the information center piece by piece. Players excitedly watch the build from nearby, and some even help build and contribute to the design.
Developer 1 is a god. They edit the world at will and break the laws of physics, existing in a plane beyond the players. To an in-game observer, their cosmic actions are confusing and unpredictable. Can you be sure your inventory won’t be accidentally deleted or reality itself might corrupt before your very eyes?
Developer 2 is like a superhero or wizard; grounded by the laws of physics, but able to bend them when needed. Like the angelic Istari, sent to Middle Earth embodied as old Men to aid the Free Peoples against the threat of Sauron, these admins are players like any other, but sparingly use their powers to fix tough problems. They might even play the game alongside regular players - vowing never to use their powers for ill gains (although if they do, everyone can see what they are doing).
One way to think about this type of intervention is to ask yourself: “what is the least amount I have to bend the world's laws of physics to achieve the desired result?”.
To illustrate the distinction, here are four outcomes that developers might want, and some ways of achieving them, in order of less to more embodied:
Getting from A to B:
Teleporting < noclip < Flying
Gathering resources:
Creation ex nihilo < Having insta-mine < Having increased mining speed
Don’t want to be killed during important work:
Being non-corporeal < Being invincible < Having increased health
Building things:
Editing smart contract storage < Having insta-build < Building telekinetically
Note that, in practice, embodied admin rights can give admins the same power as just “edit arbitrary contract storage”. After all, a world mined to an empty void is just as unplayable as a corrupted smart contract.
The difference is, if you live in the world, you can see and feel the cause-and-effect of embodied actions. Admins don’t do anything fundamentally different to players; they just do what players do but dialed up to eleven.
In general, this is less about restricting power or even immersion, but forcing developers to use their power in a way that is legible to players; communicating (potentially legitimacy-damaging) information to a diverse, global playerbase in a language that everyone can understand.
You might say embodied admin rights are more credibly neutral, which means that:
“a mechanism is credibly neutral if just by looking at the mechanism’s design, it is easy to see that the mechanism does not discriminate for or against any specific people.”
For players, embodied developer intervention is more credible, because it interfaces with the level of abstraction that they are familiar with (the game rules) as opposed to the low-level machinery that the game runs on (the EVM). Non-technical players can decide for themselves if developer actions are fair.
Before wrapping up, here are some examples of embodied admin rights in media.
In the episode Make Love, Not Warcraft of South Park, the players of World of Warcraft are terrorized by a high-level griefer. The developers, Blizzard, declare that they cannot delete the griefer or intervene in any other god-like way. Instead, they give a group of players the Sword of a Thousand Truths, a weapon so powerful that it was removed from the game. Wielding it, the protagonists finally slay the griefer in an epic final battle.
Of course, this situation and its resolution are contrived (why not just delete the account?), but perhaps… Blizzard were practicing embodied admin rights! Deleting the griefer would be confusing and break immersion. Instead, they used in-game actions to stop the harmful behavior - a much more interesting, diegetic outcome.
Another example: I’ve always found it silly that Agent Smith from The Matrix chases and fights the protagonists like a regular human (same for the bad guys in Tron). Like, don’t you guys own the simulation? Can’t you just delete the good guys?
And so I conclude: these programmes were designed as “embodied moderators”; subprocesses that maintain order in a diegetic way, but not so powerful that they could destroy everything if they ever went rogue.
To recap, Autonomous Worlds will be huge and complex and full of bugs for a long time. Once all of the physics-breaking technical bugs have been patched, there is a near-infinite number of valid ways that these worlds could evolve, and not all of the resulting places would be very nice to live in. Embodied admin rights give developers the ability to bend - not break - the physics of the world if needed, making the world resilient to unforeseen future complications (which are legion).
Despite everything you just read, I would be actually extremely hesitant to add any type of centralized control to a digital world. But some people will add it anyway, and I hope this idea can make that less bad. Let's hope for Gandalf's and not Saruman’s!