It's an age old question. It's one of those philosophical questions that regular people tend to stumble upon and that brought to life so many theories that are fun to think about that it is somehow engrained in pop culture at this point. Are we all inside the matrix? Are aliens controlling our minds? Is it an elaborate scheme of our government? Or are we all just a dream of the great Azathoth? But leaving all these arbitrary scenarios aside, the fact that none of them can be really disproven is why actual philosophers are also into this question. How can we know what is real? What does "real" even mean? Are we all chained up in Plato's cave?
In this post I want to give you my take on this whole thing. But for this, we'll need to agree on one central mechanism called modeling. It is the procedure by which a theoretical system called a "model" is found which fits a set of observations. This theoretical system can then be used to make "predictions" about things that have not been observed. It's the central mechanism behind machine learning, where the model is a computable function and the set of observations is the dataset. Note that all of this is theoretical, just like mathematics. The question if mathematics is real boils down to pure semantics and is thus not very interesting. Saying that it exists does not let us infer the existence of anything else, neither does saying it doesn't exist let us infer the nonexistence of anything.
Now comes the difficult part that I want to argue for, which you might or might not accept straight away: This modeling process is what we do. You, as you are reading this, are making new observations that you incorporate into your model of reality to infer things that are not immediately part of the observations. You are seeing letters that are part of words that you recognize as English language which leads you to infer that it was written with the intention to be interpreted as a text about whether the world is real. You only see the letters, yet you can infer that I wrote them and that I am kind of an idiot. You don't observe me to be an idiot, that's one of those predictions that falls out of your model. This model is central to you. It might as well be the thing you call "reality".
Let's assume for a moment you were not reading a text that questions the meaning of the word "reality". You'd have an intuitive grasp of the concept. Your body? Real. The room you're in? Real. Santa Claus? Not real (sorry). Planet Earth? Real. Me? Probably real. But what does this notion of "real" refer to? Here's how I would explain it: All your observations are signals that come from "somewhere". Luckily there are patterns in these observations that are coherent with a model that assumes a single "world" that produces these signals through an unfathomable multitude of processes. Hence, your model is built around the concept of a "world" in which you exist. This world is what many people thus refer to as "reality".
Obviously, your model of reality was not entirely made up by you. We humans have evolved language to share parts of our model with others, so your model is patched together from lots of different models of other people that you encountered during your lifetime, mixed with parts that you have inferred yourself. A fundamental component of language is words, such as "reality", so I think it's safe to say that the meaning of the word "reality" is this place that is governed by the rules of physics that is responsible for generating your observations, and which seems to be part of the model of nearly all people.
So, what happens now if you ask if we are all inside a simulation that perfectly mimics the world we are perceiving? You insert a useless layer of complexity into the model. You infer that there is some system that is more complex than the world, which is undetectable to as as only the parts that are like the world are actually perceivable. That is just not useful in terms of modeling. It does not let us make any reasonable predictions. Unless the simulation is imperfect and there are parts of it that are perceivable to a degree that you can show the imperfections experimentally. But unless the theory that the world is simulated is a more useful explanation of the experimental observations than a new law of nature, it's still not a very good model.
To be clear, I'm not saying that the world is not simulated. I'm just saying that we don't know. But the best thing we can do is form good models of it. And a good model is a good trade-off between high accuracy of prediction and low model complexity. Adding a nonsensical layer of "what if we're all dead and this is the afterlife?" will never improve this trade-off, as fun as it might be. So I will never take these questions serious.
So, how can we know what is real? We don't. We can define the word "real" however we like. But whatever we assign as its meaning is a part of the model we are constructing. Unless we assign *absolutely everything* as its meaning (which would not make it a very useful word), we use it to distinguish other things that are not real. Those other things would need to be part of our model as well, but defined to be "not real". Usually we define the things that are consistent with this world we've been talking about as real and those that contradict it as not real. It's a very useful concept, because it allows us to define which things to consider when making predictions about the world. But it's only useful within this modeling and prediction process.
To summarize, I think the question if the world is real is malformed. The scope that is implied is beyond our common understanding of the word "real". Thinking about it is not useful but very fun. Let's close this blog post with a little analogy: You're standing in front of a diorama, showing a piece of planet earth in precise scientific detail. Santa Claus is flying his sleigh above it. Someone now asks you to point out the pieces that are real and those that are not. Naturally, you point to earth and say "this is real" and point to Santa Claus and say "this is not". I wouldn't say you're wrong, but it's still just a diorama. A diorama inside Plato's cave.
