Hatching a conviction

For the past few months, I have been hatching a conviction. It’s too early to call this reasoning scientific, and it may never get there, but I’m going to share it anyway, because I think it’s interesting.

My conviction is this: everything in our universe can ultimately be explained by the logic of information copying, from conservation laws, to universal expansion, to the fact that we inhabit three spatial dimensions. 

What I mean by the logic of information copying is that algorithms that produce stable, complex output tend towards patterns in which information self-copies. Take Langton’s Ant, for example. At first, it produces a huge amount of apparently random behavior. Then, abruptly, it settles onto a pattern that copies itself ad-infinitum. One can view this process as a system exploring a search space of binary patterns while it seeks out one that can be stably reproduced. The highway pattern is stable in a way that the preceding randomness is not, and so the highway is maintained. The system has fallen into a more stable state.

Other examples are less obvious, but effectively equivalent. Consider cellular automata that produce gliders or other moving forms. These repeating motifs are effectively patterns that are copied forward, at the cost of the original. The only things more stable than gliders in automata like Conway’s Life are inert patterns that cannot reproduce at all.

These examples might seem somewhat detached from physics, and, in truth, they are. However, it seems increasingly to me that the imperative that self-copying creates both ubiquity and stability can serve as a bridge of understanding between very simple and abstract patterns like Langton’s ant, and those we see around us in nature.

Take the fact that we inhabit a three-dimensional universe. Why is this? Why, even, do we inhabit a universe with dimensions at all, instead of some other kind of associative structure. Mainstream physics usually requires that we bake the number of physical dimensions in as a prerequisite for any model. It’s a bold quantum gravity researcher who proposes that our set of dimensions is an emergent property. Similarly, cellular automaton enthusiasts tend to propose a lattice with fixed properties from the outset.

However, closer examination of dimensions as entities in their own right reveals them to be hugely specific in nature. The fact that sets of points, whether you consider them as infinitesimal or otherwise, should associate themselves in such a way is massively unlikely, given the vast space of alternatives. Mathematics is replete with different kinds of sets of elements, compact and otherwise, which don’t look remotely like spacetime. We’re forced to impose a global symmetry as if from outside physics itself, and to refuse to look closely at why it’s there.

However, arranging things using dimensions ensures that the resulting system has certain properties. Putting things in a one dimensional loop, for instance, maximizes the distance between connected elements while ensuring a homogeneous structure. Arranging things in a 3D space retains these nice, homogeneous properties to some extent, but adds the property that paths between any two points are almost never crossed by paths between two others.

Why would this feature of non-intersection be important? The best reason I can think of so far is that non-intersection implies non-isolation. In other words, while it’s easy to isolate a given sequence of elements on a 1D ring by blocking either end, blocking a set of elements in a 3D space is, by comparison, almost impossible. You have to define an enclosing surface, which is a complex structure in its own right.

That which cannot isolate itself cannot define its own boundary. And that which cannot define its own boundary is much less likely to be able to copy itself. This makes sense in the context of the following, highly conjectural scenario:

1. The universe starts from a very small initial condition
2. The initial state of the universe appears highly random and not remotely physical in structure
3. Stable information patterns emerge from the noise which reproduce, just as in Langton’s ant
4. Imperfections in the copying process caused by competition between patterns results in the creation of a new, even more stable pattern
5. This pattern obstructs the reproduction of other variants by copying itself in an arrangement that prevents other patterns from self-isolating
6. The reproduction of all other patterns stops, and the uninterrupted production of the ultimate pattern continues unchecked

In this story, the final dominant pattern is the fabric of space. It interacts with almost nothing, except via the tenuous medium of gravity–a process of interference that moderates the creation and placement of new spatial instances. The preceding, almost perfect, patterns constitute dark matter, which is almost as tenuous. The earlier self-reproducing patterns are baryonic matter–now trapped into fixed quantities and hence subject to conservation laws. Dark energy is, unsurprisingly, the process by which new spatial instances are continually created.

In this story, the patterns themselves aren’t made of some ‘stuff’. Rather, they are the information that defines what stuff is.

While I can’t prove this story, and can’t even join most of the dots yet, it smells right to me. This is because it tackles a lot of surprising aspects of the physical universe using a single explanation, and does so by proceeding from what we know to be true about the behavior of information.

I’m going to try to test this story by building some models that explore the emergent property of pattern reproduction in algorithms, but sadly, we may simply never know whether this picture is true or not. Nevertheless, it gives you something to think about next time you find yourself copying a piece of text, music, or video content. Copyright or not, you’re enforcing arguably the most fundamental law in Nature: that successful patterns want to be everywhere.