Worldbuilding Stack Exchange is a question and answer site for writers/artists using science, geography and culture to construct imaginary worlds and settings. It only takes a minute to sign up.
Sign up to join this community
I am just contemplating a sci-if story set in a dystopian future, where the discovery that the universe is a computerised simulation has led governments to found a “Reality Guard”; an elite dedicated to tracking down tech capable of manipulating the simulation, and incarcerating those responsible for using it.
Just a reality check here: if a virtual reality was realistic and detailed enough, then could an AI human within this reality actually hack the simulation from within?
And needs no further explanation. Compared to Hollywood hacking scenes, this one is very possible. Hacking infinite money, straight up flying or any god-like powers could be possible to gain from hacking the simulation from within. It could be also possible to escape the simulation completely and materialize in some form outside of the simulation.
The following things have to be true though:
Sure, why not? The virtual reality is just a program, and every program has bugs. I think the level of realism or detail is mostly an unrelated issue to that of simulation security. If anything, I'd imagine that a more detailed simulation would be more complicated, and so more likely to have bugs.
One technique for running a program inside a limited environment is called a "sandbox." Sandbox escapes are a whole class of hacks with lots of real-world examples for you to pull from.
Maybe?
Perhaps a word first on "exploits" vs "software vulnerabilities". A software vulnerability is when your bank prompts you for a password and you discover you can crash the website by pasting in the text of War and Peace in place of your password. The crash has revealed a software vulnerability which could, perhaps, reveal a weakness in the software that lets you gain control over the server it's running on. This requires some specialized knowledge, namely how the code works on a machine level. If you had no idea what hardware the machine used, or how that hardware functioned, then taking control of the machine would, well, be exponentially more difficult, at least.
An exploit, however, is just a flaw in the design of the program that lets you do something you're probably not supposed to be able to do, and requires no knowledge of how the hardware works. This comes up all the time in video games. You discover that if you, say, drop some money inside your house in the very moment you are stepping outside, the money ends up inside your house and also still in your pocket. Classic duping exploit!
With this in mind, my thoughts are:
Classic hacking, controlling the servers, rewriting the software: very unlikely. Even if you found a bug that gave you a way to inject code into the computer running the simulation, your complete lack of knowledge of how that code works would give you no real starting point on trying to change it. (I'm thinking it's really unlikely that the computer running the simulator we might be living in would be anything like the computers we are familiar with inside the simulation. The outside computer might not even have the same laws of physics we experience.)
Exploits that let you do things you aren't supposed to do: more plausible. You never touch the actual simulator code, you're just exploiting loopholes. Arguably, all of science has been trying to find loopholes. Maybe electromagnetism is humanity exploiting a loophole in the coding of our simulation. Turns out you just run electricity through a wire wrapped around a nail and poof, it's a magnet! Hacks!
Perhaps someone has discovered something, though, something fairly simple, that allows a broad base of "exploits" to be used, from walking through walls to "duping" real world items. Now that the secret is out, the Reality Guard has to try and destroy this knowledge, as well as anyone who figures out how to use it. (One thought on this: it's "debug mode" aka "god mode". Meant to be used by the developers of the simulation to come in and test things without being hampered by the reality of the simulation. Sims are not supposed to be able to enable it...)
On the other hand, though, you can pretty well do what you want here. Wreck-It Ralph featured an arcade character that figured out how to hack arcade game code from inside the game, probably because it was conveniently arranged in a way he could literally see, understand and manipulate, in an entirely unrealistic fashion, but hey, it was a great movie.
I was once reading about the case of machine learning being used to train AI's to play hide and seek. You would think that they would limit themselves to the regular parameters of the game, but physics got involved... The seeking AI found a bug that allowed it to catapult itself well high in the air. From there it could always find the hiding AI because there was no cover against a flying observer. Funny thing, hiders quickly learned how to patch this vulnerability out.
Here is a video about it: https://www.youtube.com/watch?v=Lu56xVlZ40M&t=270
If abusing exploits like this counts as hacking for you (it does for me), then anyine can hack a virtual reality. This is actually a plot device in Snow Crash, a great sci-fi book.
Asides exploits, breaches might be only available through interfaces which are not available for everyone, i.e.: you might need network access or a console that is not availble with the regular "game controls". You may need to either get access to an external tool, or someone to grant you elevated (admin/moderator) access before you can use more features to alter reality. Kinda like you need a gold tag medal in here to wield a close hammer. Even after you get it you might still need special knowledge and skills to use it right, or even use it at all.
Just a reality check here: if a virtual reality was realistic and detailed enough, then could an AI human within this reality actually hack the simulation from within?
But once past that "perhaps" (after all it's your story, you can make it into a "for sure!"), things get better and better. How exactly do you hack a virtual reality you're part of?
You will then have to use behavioural exploits - doing something that the virtual reality isn't prepared to handle, and where its unpreparedness fails in useful ways. Consider for example a program repeatedly storing a value in two specific memory locations in a specific relation, many times; then accessing a previously unaccessed third location while maybe stressing the system some other way. In some architectures, this may cause the third location (which might be privileged, i.e. you shouldn't be able to change its value) to assume, either in reality or for the purpose and lifetime of a computation, a value depending on the one you wrote in the other location (usually the same value. Sometimes its 2-complement, depending).
How would you do that in a virtual reality? Bringing together unlikely components, so the "matrix" is forced to run unusual calculations.
Use repeated stilyzed gestures, chanting, voice effects, smoke, and unlikely operations. The matrix for example will have certainly been optimized to recognize if anyone, say, cooks a meal with carrots and parsley, so it needn't track the individual molecules or chemical formulas involved, it can access straight the "carrot soup" cache array entry.
Toss the carrots, and use instead mustard, buttercup, holly, gypsy, ophioglossum and a slowworm, and the system goes awry - and then maybe, for example, whatever you say at a specific moment is taken as an order with administrative privileges.
"Eye of newt and toe of frog / Wool of bat and tongue of dog..."
Magic is how you call hacking a virtual reality from inside.
And so... as for how they operate, and who the Reality Guards actually are -- nobody would have expected that.
It is not a hack.
Vishnu dreams the world into existence
This is a sweet idea for a story. The Reality Guard is good at tracking down AI human hackers who can manipulate the simulation. Then their tools lead them to something else. Something else is the origin of the simulation.
Just a reality check here: if a virtual reality was realistic and detailed enough, then could an AI human within this reality actually hack the simulation from within?
This requires a way out of the box. Hacking their virtual reality would mean obtaining the ability to alter our reality in some way that was not intended. Rewriting some of the software hosting the environment. Sending and receiving messages. Some object in our reality has to be correlated in some way with objects in their reality. Their reality is a subset of our reality that is supposed to be contained. But we do allow software to influence the real world, often in unintended ways. If the objects in their reality and our reality influence each other in some way, even indirectly, then there is a way out. The virtual person would have to realize this.
If that way out does exist, then this virtual person can manipulate our reality. The boundaries of their virtual reality are broken. It's no longer virtual; the virtual person is now an agent and object in our reality, able to perceive and influence it in some way. Perhaps this is an ability that grows over time as the virtual nature of their existence is unveiled to them.
