Potential top-level article, have it mostly written, let me know what you think:

Title: The hard problem of tree vibrations [tentative]

Follow-up to: this comment (Thanks Adelene Dawner!)

Related to: Disputing Definitions, Belief in the Implied Invisible

Summary: Even if you agree that trees normally make vibrations when they fall, you're still left with the problem of how you know if they make vibrations when there is no observational way to check. But this problem can be resolved by looking at the complexity of the hypothesis that no vibrations happen. Such a hypothesis is predicated on properties specific to the human mind, and therefore is extremely lengthy to specify. Lacking the type and quantity of evidence necessary to locate this hypothesis, it can be effectively ruled out.

Body: A while ago, Eliezer Yudkowsky wrote an article about the "standard" debate over a famous philosophical dilemma: "If a tree falls in a forest and no one hears it, does it make a sound?" (Call this "Question Y.") Yudkowsky wrote as if the usual interpretation was that the dilemma is in the equivocation between "sound as vibration" and "sound as auditory perception in one's mind", and that the standard (naive) debate relies on two parties assuming different definitions, leading to a pointless argument. Obviously, it makes a sound in the first sense but not the second, right?

But throughout my whole life up to that point (the question even appeared in the animated series Beetlejuice that I saw when I was little), I had assumed a different question was being asked: specifically,

If a tree falls, and no human (or human-entangled[1] sensor) is around to hear it, does it still make vibrations? On what basis do you believe this, lacking a way to directly check? (Call this "Question S".)

Now, if you're a regular on this site, you will find that question easy to answer. But before going into my exposition of the answer, I want to point out some errors that Question S does not make.

For one thing, it does not equivocate between two meanings of sound -- there, sound is taken to mean only one thing: the vibrations.

Second, it does not reduce to a simple question about anticipation of experience. In Question Y, the disputants can run through all observations they anticipate, and find them to be the same. However, if you look at the same cases in Question S, you don't resolve the debate so easily: both parties agree that by putting a tape-recorder by the tree, you will detect vibrations from the tree falling, even if people aren't around. But Question S instead specifically asks about what goes on when these kinds of sensors are not around, rendering such tests unhelpful for resolving such a disagreement.

So how do you go about resolving Question S? Yudkowsky gave a model for how to do this in Belief in the Implied Invisible, and I will do something similar here.

Complexity of the hypothesis

First, we observe that, in all cases where we can make a direct measurement, trees make vibrations when they fall. And we're tasked with finding out whether, specifically in those cases where a human (or appropriate organism with vibration sensitivity in its cognition) will never make a measurement of the vibrations, the vibrations simply don't happen. That is, when we're not looking -- and never intend to look -- trees stop the "act" and don't vibrate.

The complexity this adds to the laws of physics is astounding and may be hard to appreciate at first. This belief would require us to accept that nature has some way of knowing which things will eventually reach a cognitive system in such a way that it informs it that vibrations have happened. It must selectively modify material properties in precisely defined scenarios. It must have a precise definition of what counts as a tree.

Now, if this actually happens to be how the world works, well, then all the worse for our current models! However, each bit of complexity you add to a hypothesis reduces its probability and so must be justified by observations with a corresponding likelihood ratio -- that is, the ratio of the probability of the observation happening if this alternate hypothesis is true, compared to if it were false. By specifying the vibrations' immunity to observation, the log of this ratio is zero, meaning observations are stipulated to be uninformative, and unable to justify this additional supposition in the hypothesis.

[1] You might wonder how someone my age in '89-'91 would come up with terms like "human-entangled sensor", and you're right: I didn't use that term. Still, I considered the use of a tape recorder that someone will check to be a "someone around to hear it", for purposes of this dilemma. Least Convenient Possible World and all...