Today I learned from an academic talk given by Prof. Gilles Brassard (Montreal): In the context of artificial intelligence one discusses not only civilizations (level 1) which are simulated by civilizations of real persons (level 0), but also civilizations (level 2) simulated by simulated civilizations from level 1. And iterated further on level n, n+1, …

How to define the characteristics of the species of „real“ persons on level 0 to distinguish them from the characteristics of the species of simulated persons on level n > 0?

  • Please add some details here. I don't know what you are talking about. I will say, though, that in normal English, the difference between a real person and a simulated person is that there is something it is like to be a real person, but there isn't something it is like to be any sort of simulated person. Jun 27, 2022 at 18:49
  • @David Gudeman I added some reference. - You define "real" person by "it is like to be a real person". I consider this a circular statement; analogously your statement concerning about being a simulated person.
    – Jo Wehler
    Jun 27, 2022 at 18:57
  • Is the Turing Test relevant here?
    – BillOnne
    Jun 27, 2022 at 19:04
  • Even if different people can think they are the same thing (say level 0 you thinks they’re natural and are but don’t yet know and level 2 you has the exact same experience but is wrong without knowing they’re simulated), they still have their own unique internal and/or external relations (being right/wrong, being natural/not). Idk if that helps.
    – J Kusin
    Jun 27, 2022 at 20:27
  • Dunno. Bring me a simulated person so we can compare.
    – armand
    Jun 27, 2022 at 23:05

3 Answers 3


How to define the characteristics of the species of „real“ persons on level 0 to distinguish them from the characteristics of the species of simulated persons on level n > 0?

There are, broadly, two answers:

There is also a third, frame-challenge answer: "You can't simulate people in computers, so there's nothing to distinguish." I think this is much more sensible than the philosophical zombie argument.


I mean probably the only straight forward answer you could give, without further knowledge of the nature of the civilizations or the simulated realities, would be that a level 0 civilization is a level n civilization for which no level n-1 civilization exists. So to say an "unsimulated civilization".

Where "level n civilization" makes no implications on the similarity or differences between levels beyond the fact that a level n+1 civilization is hosted within the shared reality of a level n civilization.

So before you could give any answer beyond that you'd need to define what "reality" is both for level 0 and level n>=1 civilizations, what their civilization would be like (at least in terms of broad constraints) and how you would construct the demarcation between "simulation" and "reality".

Because depending on how you do it you could argue that a level 2 civilization is possible or impossible. or what it would look like in terms of size and complexity.

Like say you want to simulate battle. Then the level 0 civilization would actually go to war, a level 1 civilization might perform a life action role play, still largely following the concepts but cutting out the part where anybody gets actually seriously harmed. A level 2 simulation would be a general moving figures on a map and a level 3 simulation might be someone playing chess.

Now you could argue that all of them are only level 1 because they all try to mimic a level 0 event and they all arguably happen in the level 0 reality. Or you could argue that while they are processed in the level 0 reality, the scope in which they are happening is actually a level 1+ reality because the rules and the internal logic does not allow for interaction with the level 0 world, so it's somewhat self-contained. And you can argue that this doesn't actually work, cause of examples like where an upset stomach results in a draw in chess despite the body of the player not being technically part of the realm of the chess game, but it's also not entirely possible to remove them from it either (same for "the mind" which inevitably is part of the chess game and isn't, sorta).

But you could also argue that they are simulations of simulations of simulations. Like the role players mimic the battle, but the general mimics the role players (maybe novice general without battle experience) and so on. So in that case you'd actually have legit levels of abstraction where you can go beyond 1. However with the same catch as before in terms of technically being in the spacial domain of level 0.

Also you've got a decreasing level of complexity. Each level removes features to focus on the ones actually relevant to them. Meaning in that kind of simulation hierarchy the most complex one or the most "irrational" one would be the level 0.

But that makes assumption on the nature of the simulation and you could also imagine an improvised play about people improvising a play. Where you don't lose complexity ascending the levels of simulations. Like you could argue that the actors would need to ignore the fact that they are in a play and that there is an audience. But would they? They could very well interact with the audience without breaking the 4th wall. "Hey we're making a play about making a play what do you think about it?" So they utilize all of a level 0 world but still pretend to be in a level 1 world.

And even worse what if at some point they decide, no actually it's not a play our roles are who we are and we're the real world now, so it's actually you the audience) who's putting on a play for us. Would that change who's the level 0 and whose the level 1 civilization?

Seriously just because we think of computers and AI if we think of simulations does not mean that all simulations would look like that. But if you need to you can apply the same logic on computer simulations as well.

A human sitting in front of their computer would be level 0, a program running on that computer would be level 1 and if that program started another program it gets complicated. Like is that program running in parallel in the shared ecosystem of the computer or is it running within the program so within the allocated space of resources and is it then actually running within or encapsulated from it and therefor in parallel and aren't they still both running in level 0 or at least level 1?

So TL;DR one would need more information to answer that question beyond the trivial answer and even then you'd probably have a hard time doing so, without making some axiomatical decisions.


How to define the characteristics of the species of „real“ persons on level 0 to distinguish them from the characteristics of the species of simulated persons on level n > 0?

This is a pseudocode example of a "simulated person" that can say "Hi! My name is John.":

Class Person{
  String name;
  public Person(String name){this.name=name;}
  public say_hi(){print("Hi! My name is"+this.name);}
Person john=Person("John");

Which will output:

Hi! my name is John.

Now, I will do the same:

Hi! my name is Rodolfo.

Now, try making the difference between that person, which I've simulated (S+1) and me (S+0). Evidently, differences and similarities are quite abstract and can be approached from an infinite set of perspectives. Better not to address them in detail, that would be quite subjective.

I can write an S+n:

Class Simulator{
  Simulator simname;
  Person person;
  public Simulator(String simname){this.simname=simname;}
  public Simulator(Person person){this.person=person;}
  public simulate(int instances_number){...;}
Simulator jill=Simulator(Person("Joe"));
Simulator jack=Simulator("jill");

What would be the differences, other than the implementation?

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .