blog.dominik-roth.eu

The Missing Physics

We have quantum physics for the microscopic and classical mechanics for the macroscopic. We define a line between these that is fictional. The physics for this in-between is missing. The mind sits at this precise location where physics fails.

There are 3 paths to getting at a theory that explains this gap; and thus solve physics (or at least make a huge jump towards that) and how the mind works:


Path 1: Gravitons

We can try to define gravitons into existence. Gravitons need to mediate both the geometric behavior of spacetime as well as the phenomenon of collapsing a quantum system onto a classical state.

Quantum systems have an intrinsic potential (energy or similar thing) that is gapped. Only quantum systems of sufficient size can tunnel into the lower energy classical state while emitting a graviton.

We are failing along this path. Physics is stuck. We made no actual progress for decades. The only progress was on irrelevant side expansions. Experience without theory is blind, but theory without experience is mere intellectual play.


Path 2: Entropy

We don’t need measurement to physically exist. Any measurement can be explained as just entangling the small system with a large system. Measurement and collapse onto a classical state is just rapid decoherence.

Alternatively we only need measurement. Any operation on a quantum system can be phrased as a rotation (may use additional dimensions as the system has) followed by measurement by standard basis on a subset of dimensions followed by rotating back. But this is equivalent to just applying a measurement using an orthogonal basis that is rotated itself. We must be allowed to rotate into additional dimensions, of which we may demand arbitrarily finitely many; or we demand infinitely many but initialize all above a cutoff as 0, and measurement is orthogonal to all those.

Both of these model classical physics as an edge case of quantum physics in the limit of n → ∞. There are constructions of systems in which this works beautifully; the quantum weirdness cancels out and we are left with classical physics. But also other systems for which this simply does not happen, and we predict macroscopic quantum weirdness. Or also those where a classical physical system emerges but not one that uses the physics of our universe but a different one.

If we go for ‘there is no measurement’, then that’s where entropy comes from. If the mind is the process of increasing entropy, then a consciousness would follow a quasi-classical path through branch space.

We are failing along this path too.


Path 3: Build It

It’s very easy to wrongly convince oneself to understand something one actually does not. We have mental blind spots where missing understanding hides. The only way to ensure complete understanding of something is to build it.

‘We will never understand ourselves until we have succeeded in constructing an equal.’ — George Hotz

We are failing along path 3 as well. On one side, current advanced AI is fundamentally different to us. On the other side, we don’t even understand the AIs we are building. We understand the model (basically the Ansatz for the intelligent system) and the method used to condition it. But the emergent properties that occur during training are beyond our understanding; similar to how we have the complete DNA of the brain and actual working brains, but still don’t understand what’s going on inside them.


Is the gap even comprehensible, or does Gödel’s incompleteness take that away from us too? Don’t worry, it does not. Gödel’s incompleteness theorem is wrong. Or at least named wrong. It does not prove incompleteness but only incomputibility. It’s just a proof over the impossibility of arbitrary infinite compression of logic. It still places certain questions about our universe out of reach, as we can’t leapfrog the universe with our Turing-limited minds; but it does not fundamentally break physics, math or logic. To claim otherwise is as silly as crashing out over whether the set that contains all sets that don’t contain themselves contains itself.

Roger Penrose’s The Emperor’s New Mind is a great resource on the relationship between consciousness, physics, and computation; and argues compellingly that the mind cannot be fully explained by classical computation alone.

Stephen Wolfram’s A New Kind of Science explores how simple computational rules can give rise to complex physical phenomena; relevant to the question of whether the universe and mind are fundamentally computational.

Liu Cixin’s The Three-Body Problem is fiction on the existential horror of living in a universe escaping our comprehension.

The Hole at the Bottom of Math is an amazing video about Goedels incompleteness theorem and the birth of computer science.