YouTube channel with IF analysis

If you follow my blog or follow me on Twitter you may have already seen this, but if not: I started a YouTube channel on which I’ve so far posted two videos analysing interactive fiction. In episode 1, I talk about Adam Cadre’s 9:05. And in episode 2, I talk about fictional truth and secondary worlds. Enjoy!


(Also, I should learn to make thumbnails, because these random frames of me speaking are horrid!)

What a lovely pair of though-provoking videos!

After watching the second video, I’d like to go off on a tangent of my own. While your line of reasoning was expressed with clarity, and easy to follow, I kept thinking about how all of this would relate to quantum mechanics.

Spoilers for 9:05 below.

You point out that one thing that distinguishes fictional worlds from the real world is that the latter is “maximally specific”: You have a certain number (currently unknown) of hairs on your head, but it is not true that Bilbo has a certain number of hairs on his feet. So far I agree with the example. If we zoom in on one of your hairs, we would find that it consists of a definite number of atoms, each with a definite number of electrons. Still maximally specific. But if we zoom in on one of the electrons, and start talking about its precise location, then we find that the electron can be (and, in a sense, is) located in infinitely many different locations. We can only know the probability that the electron is located in any particular spot, and that the probability is highest in a spherical shell at a particular distance from the nucleus of the atom. So at this extreme zoom level, the real world is not maximally specific.

When we shoot a single photon at a board with two slits, it would seem that the photon could pass through either the first or second slit, bounce off the edge of the slit and change its direction somewhat, and then hit the far wall at some specific spot. But in reality, the photon takes both paths, as well as infinitely many other paths that go all over the place and don’t even have to follow straight lines. All of these infinite possibilities then interfere according to some mathematical laws that people have figured out and confirmed experimentally over and over, and the interference affects the probability that the photon reaches the wall at any particular spot. If we run this double-slit experiment several times (each time with one single photon), the probability function can be seen, visually, as bands on the wall. Bands that do not appear if we hold a piece of paper in front of one of the slits.

Meanwhile, in 9:05, the narrative could take a path that puts a dead body under the bed, or—as you convincingly explained—it could take a path that does not. As a thought experiment, we could even argue that until the story is over, there both is and isn’t a dead body under the bed, just like Schrödinger’s cat is both dead and alive until we open the box and observe it. Because of how 9:05 is constructed, it is very likely that there ends up being a dead body under the bed, but you give an example of a reading where this did not happen. We could say that the dead body exists in the fictional world with a very high probability, but not quite 100%.

For the people living inside one particular instance of the fictional world of 9:05—if this notion makes any sense at all—there either is (100%) or isn’t (0%) a dead body under the bed. All they have to do is look under the bed and observe. But across all possible instances of this fictional world—all the playthroughs that ever happened, or will happen—the dead body exists with, say, 99% probability.

Another thing that may or may not happen in the fictional world of 9:05, is that the protagonist manages to leave Las Mesas. When the story is read for the first time, this is highly unlikely to happen. On the second reading, it is highly likely. If we assume that most people read 9:05 exactly twice, then across all playthroughs, the overall probability that the protagonist gets arrested is about 50%. This figure comes about because information gained from the first playthrough affects the outcome of the second; there is a kind of one-way interference. One could say that information travels between separate instances of the same universe. For people inside any particular instance of the fictional world, our protagonist either ends up in Las Mesas (100%) or outside of it (0%); they could observe this variable. But before they do that, the probability is 50%. If they could measure this probability, like we can measure the probability of the photon ending up on a certain spot on the wall in the double-slit experiment, then they would find this probability to be about 50%. And that figure depends on factors outside of their particular instance of the fictional world. As the story is written, however, they cannot measure this probability because they cannot repeat the experiment.

Perhaps a different interactive story could be written, in which the fictional inhabitants of the fictional world would be able to measure the probability of some out-of-world event. Their experiment couldn’t be described explicitly in the narrative, for in any particular playthrough all the variables would be fixed. But perhaps a fictional world could be crafted in such a way that it would be conceivable, without contradiction, for the inhabitants to carry out such an experiment. I’m not sure if this is possible, but I find the idea intriguing.

Note also that the two variables in our 9:05 example are entangled because of how the story is set up. If there is a dead body under the bed, the protagonist leaves Las Mesas with about 50% probability, as discussed above. If not, the protagonist leaves with 100% probability. This is because the game-over message in which the protagonist remains in Las Mesas (gets arrested) reveals that the dead body exists. This is somewhat confusing, because it seems that cause and effect are reversed: The end of the story seems to affect what happened at the beginning of the story.

And so, if the people in the fictional world had figured out how their world worked, and were somehow able to measure the probability of the protagonist leaving town, and that probability turned out to be 100%, then they could be reasonably sure that there would be no body under the bed. It would be evident from the experimental data that somebody was holding a piece of paper in front of one of the slits.

To be perfectly clear, I do not claim that “fictional worlds” have to behave or be defined in a certain way. I’m not trying to use natural science to make a point about how made-up stories “ought to” be written or read. What I do say, is that our understanding of the real world at the quantum level ought to inform our thinking about fiction, especially interactive fiction. In particular, I think it’s too simplistic to claim that fictional worlds have to be fundamentally different from the real world because of the way the latter is maximally specific. I would say that through the lens of quantum mechanics, the real world seems to be a rather good fit for our made-up notion of fictional worlds.


Thanks for that, Linus! To be honest, when I wrote the line about “maximally specific”, I hesitated for a moment. For I myself too do not believe that the real world is maximally specific, although for somewhat different reasons than yours; and indeed I even have an article coming out in which I briefly argue for this! (It’s an academic philosophy article in Dutch, so not much of a possible readership here.) But I decided that these issues were too arcane to get into, and that certainly none of my viewers would challenge this particular assertion.

I was wrong!

The reason I’m slightly hesitant to accept your reasoning is that it depends heavily on which interpretation of QM one chooses. For it could certainly be held that the quantum world is maximally specific: any system is always described by exactly one wave function. What may not be specific are the values of certain macro variables, like velocity and position – but that is because these values are not at the fundamental level of reality. Of course, whether this interpretation works will depend, among other things, on how one deals with the measurement problem.

(My own reason to reject maximal specificity is that I think a maximally specific world could not contain free will, and I think there is free will.)


I’ve also just posted episode 3, Introduction to Interactive Fiction. Perhaps of less interest to people on this forum, but I hope it can be a useful resource!