Referencing the Unreferencable

post by Chris_Leong · 2020-04-04T10:42:08.164Z · score: 17 (3 votes) · LW · GW · 5 comments

Jessicata wrote a post on The Absurdity of Un-referencable Entities [LW · GW]. The comments there give many good reasons against her claim that unreferencable entities don't make sense. Nonetheless, I think it is value to understand unreferencable entities in more detail as there are many circumstances in which they come up. Before we start, there are two key points we need to understand about references.

Firstly, "referencable" is relative to the agent. If the universe consists of two non-interacting boxes, then agents in the first box can reference object in the first box and agents in the second box can reference objects in the second, but neither can reference objects in the other. Similarly, if an agent is in a simulation with no access to the outside, that agent can only reference objects in the simulation, while someone outside can reference objects in both the external world and simulation.

Secondly, even when given an agent as a reference frame, we still need to clarify what counts as a reference. Here's a list of entities. Some people would say that we can't reference some of these entities as merely be able to saying the words doesn't mean we can pin down the object enough for us to be able to truly say we are referencing it:

Each of these kinds of entities is referred to in a slightly different manner. Precisely defining how each of these works would be a very involved task, so I'll just sketch it out quickly. The purpose of the following paragraph is only to show how we could go about doing this:

For phenomenal content, we have the ability to recall what we experienced and interface it with the rest of the brain. We typically reference objects by experiencing visually or through our other sense or indirectly via scientific instruments and saying "Whatever is creating that experience". For primitive properties, a person will normally understand the meaning if they are told a bunch of objects are in the class and a bunch of objects aren't. For non-primitive properties, we can use the same method or we can define these in terms of other properties.

Again, the specifics aren't important. The point is only to illustrate that even though these references look alike on the surface, the manner in which these entities are triangulated greatly varies.

The Unreferentiable:

We've explained above that some agents can reference objects that are unreferenceable to other agents. However, even if an agent can't reference entity X, an agent can construct a model of a world containing itself and an entity X' that stands in for X. It can even imagine an agent Y that can reference X'. How is this possible? Doesn't providing a stand-in for X involve referencing the object we said was unreferencable?

The resolution is to recall that there are difference senses in which we can reference an object. Let's make this concrete: Imagine a programmer and a manager observing a perfectly closed system containing an AI human and an AI dog. The AI human can reference the AI dog by "seeing" it and presuming there is an entity behind what it is seeing. It can't reference the external agents in the same way. However, it could construct an internal model containing a programmer and an AI system which the manager could interpret as a reference to the programmer. The AI human can't include a full copy of itself inside the model, but it can include general description of itself which will be recognisable to the manager.

Referencing an entity through perceptions and via positing a reflective model are entirely different approaches to reference. Many people take the first to be what a reference really is, which is fine, so long as they offer a word for the second non-reference kind of reference. I personally like the term Reference by Analogy.

One potential objection is that such references could fail to be resolve such as if there are two programmers, making the reference underspecified. But that's hardly unique to this situation: what if there isn't actually a chair object behind the chair image, but it is all in the imagination?

I want to finish by emphasising that the point is simple. An entity can be unreferencable in one sense, but referencable in another. In fact, we should probably just taboo this word in the majority of circumstances.

5 comments

Comments sorted by top scores.

comment by jessicata (jessica.liu.taylor) · 2020-04-04T18:54:37.634Z · score: 4 (2 votes) · LW(p) · GW(p)

If you fix a notion of referenceability rather that equivocating, then the point that talking of unreferenceable entities is absurd will stand.

If you equivocate, then very little can be said in general about referenceability.

(I would say that "our universe's simulators" is referenceable, since it's positing something that causes sensory inputs)

comment by TAG · 2020-04-05T12:51:04.573Z · score: 3 (2 votes) · LW(p) · GW(p)

Equivicaction is using a term in different senses *during the course of an argument"..that is under conditions where it should normatively have a stable meaning. It is still the case that some words are ambiguous, and that recognising ambiguity can solve problems.

comment by Chris_Leong · 2020-04-05T01:07:03.322Z · score: 2 (1 votes) · LW(p) · GW(p)

We will still be able to talk about unreferencable elements, but only by "referencing" them in a different sense than what we mean by "unreferencable". The key is that it might seem like we are only using the word in one sense until we really break down the definition, at which point it becomes clear we are using it in different senses. And it's not equivocating because we are only using the word referencable in a particular way. When we "reference" unreferencable elements we don't call it "referencing" even though we are in the casual sense.

comment by Donald Hobson (donald-hobson) · 2020-04-05T23:34:34.180Z · score: 3 (2 votes) · LW(p) · GW(p)

Suppose that I am in a simulation, and the simulator drops in a hard disc containing a detailed description of the world outside the simulation. This description says how to reply. The simulator is clearly referencable. Gradually reduce the amount of evidence. You are looking at a pattern in coinflips, it might be a message from simulators, or maybe noise. You are looking at physical constants and wondering why the simulators chose . There is no sharp line from referencable to unreferencable. Just a gradual increase in uncertainty.

When I say "there is a chair over there" I am not refer to a single hypothesis, a particular arrangement of atoms. Instead I am refering to an implicitly represented ensemble of hypothesis. This ensemble contains universes made of atoms, strings, platonic elements and much else besides. Within the set of atomic universes, the set contains all arrangements of atoms that contain a chair over there. So within this set is a universe of atoms, defined in terms of a long list of coordinates, in which [the moon is made of green cheese, and a solid diamond rocking chair is in the indicated direction ][Translated from a big list of numbers]. So "the simulator has green hair" is only a valid proposition over the subset of possible universes that contain exactly one simulator. The probability you assign to this subset can vary. When it is almost 1, "the simulator has green hair" feels either true or false. You feel like you can reference "the simulator"

comment by Chris_Leong · 2020-04-06T00:20:45.693Z · score: 2 (1 votes) · LW(p) · GW(p)

"Suppose that I am in a simulation, and the simulator drops in a hard disc containing a detailed description of the world outside the simulation. This description says how to reply. The simulator is clearly referencable" - yeah, the situation can be dynamic. The simulator can be unreferencable at the start and then become referencable later like if a cat comes into my vision I can then say "that cat".