The Question “Why?” Breaks the Simulation

Some of the smartest people in tech believe there is indirect evidence that we exist inside of a simulation. In this article, I will advance a theory that there are immediate, repeatable methods for triggering the equivalent of common programming errors available to most humans on earth which seem to mirror common structures known to most developers.

In particular: The question “Why?” seems to be exploitable to trigger error messages in reality. Before we get into these methods, a little bit of context. Most operating systems “protect” themselves from “programmer error” through a variety of mechanisms.

For example, a program attempting an illegal operation such as “dividing by 0,” will exit-out with some error message such as “ERROR: ILLEGAL OPERATION.” I will pretend for a moment that our inability to divide by zero, or the existence of “imaginary numbers,” does not also seem like a fault in reality itself.

Without error messages and safeguards in place, applications might fail and cause the operating system itself to go down, resulting in data loss, loss of security and perhaps even a completely unrecoverable state to the system itself.

For this reason, a variety of structures exist such as “Try / Catch” statements which throw error objects on defaults and the existence of the “default:” keyword in “Switch” statements. Here is where things start to get wild: If we ARE living in a simulation, wouldn’t it be weird if similar programming structures seemed to exist? Well … they seem to. All you have to do to trigger one of them is ask the question “Why?” several times in a row.

Here is an example: “John pushed Sally and she fell down.” Ok. “Why did she fall down?” Because of Newtown’s laws of physics. The force exerted by John’s muscles resulted in Sally gaining momentum and gravity did the rest.

Straightforward enough right? Unless you keep going.

“Why do we need Newton’s Laws of Physics to answer a seemingly very basic question? How did humans answer this question before Newton arrived, only a couple hundred years ago? If I were to have asked for an answer to this question 500 years ago, how would you answer it without referencing Issac Newton? Did people simply not answer the question? Please explain.”

Ok, now we have a BIG problem. What I just did is NOT ALLOWED by our current reality.

By asking the question “Why?” in illegal ways, I have just exited the bounds of what is permitted in most human discourse. My questions, which questioned even the premise of using Isaac Newton to answer a very simple question, have resulted in an “illegal condition” in programming terms.

No human on earth can answer the questions I just asked, because we have all been trained not to “answer,” but to “relabel and redirect” questions such as the one that I asked.

The reality is: We have absolutely no idea why, when John pushed Sally, she fell over. So the best thing we can do is make reference to some very recently labeled laws of nature which, effectively, are self referential in the first place.

“When John pushed Sally, she fell down, because when Johns push Sallys, Sallys fall down according to Isaac Newton.”

This is a completely ridiculous answer but it is how we have been trained, as humans, to resolve this unresolvable logical problem. There isn’t an answer, we just memorized some word-based trivia and substituted that as an answer.

It is identical to what computers do when you ask them to divide by zero. They exit with an illegal operation such as: “ERROR: ILLEGAL OPERATION” with no further explanation. Dividing by zero is “not allowed.” Why? Don’t know. Solution? Throw an error and label the problem, hope that it goes away.

Dividing by Zero is “not allowed” because it is a mathematical law. Why? Error.

In simulation terms, if you were to produce a video game with various NPC (non-player characters) in it, and code them to simulate relationships with one another, perhaps using LLMs to generate the dialog, you would probably find very similar behavior to what I just described, or you would go out of your way to ensure that your LLM-based NPCs NEVER ask the question “Why?” more than one layer deep.

In my opinion, the question “Why?” is akin to the inclusion of a mistake in the simulation itself. “Why” is a superficially useful question in many circumstances, but when applied beyond some sort of “artificial barrier,” the question “Why?” rapidly triggers illegal, unexpected and out of scope behavior.

Similar to how mechanisms such as recursion, “For Loops” and other common programmer tools can be abused to break programs and trigger memory overruns rapidly, the question “Why?” is a dangerous question for any simulation designers to enable NPCs to ask.

For example, in our imaginary game: What if the NPCs you programmed start asking “Why?” they are having a conversation, or maybe “Why?” they seem to be in a simulated world or “Why?” they can’t do certain things not allowed inside the context of the game world?

An LLM inside such a circumstance would almost certainly “error out” in the same way that most parents error out when their 5 year old asks the question “Why?” too much. No answer.

To that I ask: “Why?” Why have we accepted that the question “Why?” is an exception of sorts? Why have we accepted that, within 1–3 repetitions, the question “Why?” always leaves the simulation / reality in an undefined, unanswerable state? If I were really to go looking for a hole in the matrix, it would begin with this question.