Philosophy of Systemic Computation

Systemic Computation is not just a computer architecture or a way of modelling. It's a way of thinking about things in the universe.

Let’s start with the basics. There appears to be stuff in our universe. Big stuff, small stuff. We have a lot of different names for all this stuff: planet, neuron, virus, water molecule, helium atom, neutron. But it’s hard to talk about everything at the same time if it’s all got different names. So let's call everything by the same name. From now on, everything is a “system.” It doesn’t matter how big or how small, what it’s made from, whether it’s alive or dead. Everything is a system.

We all know that the stuff (systems) in our universe seems to be made from other, smaller stuff (systems). You are an organism (a system) made from several major organs contained within another organ called the skin (systems within a system). Your organs are made from cells (systems), which are made from molecules (systems), which are made from atoms (systems), which are made from subatomic particles (systems), and so on. There appears to be a hierarchical arrangement of systems. Systems made from systems made from systems.

This hierarchical arrangement is not purely subjective. Each level of scale can be separated from the higher and lower level by degrees of interaction. So planets tend to interact at a planetary scale. While they are ultimately made from subatomic particles, their scale means that they affect vast numbers at a scale similar to them. (At this scale, the combined gravity is more significant compared to many of the smaller and larger scale interactions; at other scales, different interactions predominate.) Animals tend to interact mostly with animals, plants or other features around their scale – one animal is usually not able to modify individual atoms or entire planets in any controlled or significant way. Cells tend to interact mainly with cells; genes and proteins with other genes and proteins, and so on. So systems in the universe seem to form “clumps” at similar scales. These structures are self-organising and appear to be caused by the laws of physics, nothing more or less.

At each level of the hierarchy of organisation, systems are also all limited by boundaries or scopes. So the large bodies in our solar system interact when they are within each others’ gravitational fields. If our Moon were ever pushed significantly outside the scope of the Earth’s gravitational field, then it could no longer interact with the Earth in any significant way. Species on Earth are often limited by geographical constraints, so if one member of a species was to find itself on another continent with no way of returning, then it would be outside the boundary that contained its species and it would no longer be able to interact with other members of its species. The cells within our bodies can only interact if they remain within our bodies. Remove a cell and it cannot behave as a part of you any more.

If we were to draw a typical arrangement of systems at just one level of the hierarchy, then it might look something like this:

numberedsystems.jpg

Every system is a circle with two little arms. They're numbered so we can refer to them, but the numbers are not important. Also, don’t worry about the absolute positions of these systems. Think of them as like mathematical sets where all that matters is what is inside or outside. So systems 1, 2 and 3 are inside the scope of system 4, which is itself within the scope of system 12. System 9 is outside the scope of system 4, but within 12. System 15 is outside all scopes. The scope of a system is also a system, because that’s the first rule: everything is a system.

Perhaps in this picture we are looking at several cells contained within boundaries formed by the organs, which are within the skin of an organism. Or perhaps we are looking at molecules that are within range of each others’ electromagnetic charges. Or maybe we’re looking at probability distributions of subatomic particles. The exact level doesn’t matter – we can use the same language, and see similar structures.

Natural systems have many important properties as described on the other pages of this wiki so we already know how these systems behave. We know their behaviour has a significant stochastic element – there is randomness in whatever these things do. We know they behave asynchronously – there’s no inherent coordination underlying their function. They do things in parallel, so while they may not all exhibit a specific behaviour at the same time (which would then be synchronous), there’s nothing to stop them from doing their thing within the same period of time that others do something.

Clearly these hierarchies of systems are autonomous – they need no intelligence to organise them. They also seem to be homeostatic and fault tolerant. They “like” being in their hierarchies and in their scopes so much that they will restore themselves if damaged (even atoms “prefer” to have a certain number of electrons and neutrons and will grab more or discard extra if the number is not in balance). Systems are definitely robust, for example, you don’t normally see gravity “crash” or electromagnetism “break.” There are definitely a lot of systems in the universe, with no sign of a central controller, so we can call them distributed. There appears to be almost no limit to the number of hierarchies or the different arrangements of scopes, so our systems are open-ended and complex.

Every system is approximate. Although not shown in the picture, every system is usually made from other systems, so you could substitute each system with one or more other systems if they are exactly equivalent. So we could choose to talk about you being a single system, or we could talk about you being a collection of tens of systems (organs), or of trillions of systems (cells). This means whenever we use this language we must choose the level of abstraction (the level of the hierarchy) that we wish to focus upon, and we always recognise that the description is an approximation. There is probably always going to be a different way of expressing each system so each could always be replaced with a different set. The system is a way of talking about and summarising reality while recognising that the words are never absolutely right.1

Scopes are a little more complicated than we’ve discussed so far. Every system can act as a scope, where that scope behaves like a field – other systems may be partially or fully within that scope. Systems should only be able to interact if they are in the same scope as each other. (Or more correctly, the interaction of systems is a function of their membership of the same scope, so they might have a higher probability of interacting the more they share a greater membership.) Intuitively this means that things within the same region or boundary are more likely to interact, while those separated from each other are less likely. While systems are drawn as circles, the field may be any shape or have any properties. So we may wish to talk about gravitational field systems, or probability distribution systems, or human skin systems. In all cases we’re referring to a kind of boundary with some internal properties that affect the internal systems, external properties that effect external systems, and potentially fuzzy edges. This means every system may “overlap” every other system, even if it is to an infinitesimal degree (or to a very large degree), allowing our systems to merge with each other (or have partial memberships of each other). Systems can thus affect each other and be affected by each other to a large degree, making them very embodied. For example if a couple share a house then the scopes of the two people overlap, so they are likely to be able to interact with the same artefacts (that are in both their scopes) and with each other.

It’s clear that all systems do something. Maybe they just move. Maybe they pull or distort. Maybe they alter other systems. So in our systems-based Laws of Physics we have to have a notion of behaviour. If two systems interact then they will transform each other at the same time. This means that the first system causes a change to the second and the second causes a change to the first – circular causality. We only need to think about two systems interacting at a time. (Even if more than two interact at precisely the same instant that’s the same as multiple sets of two systems interacting in parallel.) When you think about it, that’s all behaviour can ever be in the universe: two systems interacting. Nothing can change until some form of interaction occurs, whether we call that interaction a collision, a crash, a conversation, an attraction, or a reaction.

As soon as we have a concept of behaviour causing change, then we have to think about what is being changed. Clearly systems have properties in the universe: they may have specific spatial locations and shapes. These properties may be changing over time so they may have speed, acceleration, compression, stretching or deformation. Looking at higher in the hierarchy and systems corresponding to cells may have considerably more complex properties; systems corresponding to organisms may have massively complex properties. Whatever those properties are, they all derive from and comprise systems.

Their properties also help determine the result arising from interactions. So a baseball moving at 50 mph that hits a stationary glass window will transform the properties of both ball and window. After their interaction, the ball will no longer be travelling at 50mph and the window may have a rather more complex shape and many types of movement. (After the interaction it may be preferable to move down in the hierarchy of description and talk about the systems that make up the glass, as it may no longer make so much sense to regard it as one system with a very complex broken shape.) But a foam ball moving at 50mph that hits the same glass window would result in a very different transformation of properties. So in our systems, information appears to be intimately linked with behaviour. Both information and behaviour affect and are affected by all interactions. There’s no binary information stored and manipulated in one place, separate from everything else. We cannot separate data and function; we cannot distinguish between message and morphology. Shape, behaviour and information are the same thing.2

Returning to the example of the broken window, there’s one other factor we need to consider. The environment plays a key part in all interactions. If we were to move our window underwater and then throw the ball at 50mph towards the window, the resultant interaction would be hugely different. So the result of any interaction depends on the two systems that are interacting, and also upon the context in which that interaction takes place. Since everything is a system, we can express the context as a system as well.

We all intuitively recognise that the real world is context-sensitive. Even the fundamental law in most human societies, “thou shalt not kill,” sadly depends on the context in which this interaction occurs. While the context of normal society tries to ensure that the killer is significantly penalised as a result of such a deadly interaction, in the context of warfare the killer is rewarded after an identical interaction. When children ask, “what will happen?” we may answer, “it depends,” because it usually does. The result of any interaction depends on its context.

While it could be argued that there is no need to describe context explicitly as a system (we could replace the context with other interactions occurring in parallel), we are limited by what we can express, so the context is like a summary of those parts of the environment that significantly affect the current interacting pair of systems, allowing us to ensure we do not omit anything important. It also means that now any system can potentially take the role of a context and affect how two other systems interact (that’s why we draw systems with two little arms).3

Finally, we should ensure our systems are consistent with existing knowledge. All behaviours should follow those observed in the universe. Systems cannot be destroyed or created from nothing, only transformed. (We may interpret systems differently perhaps by substituting one set of systems with another set, but they would still describe the same feature, so this is not a transformation of any system, just of our representation of it). Any system can be a context, a scope, an interacting system or any combination at once.

These form a Language of Physics - a way of writing down and talking about the Laws of Physics. Admittedly they are vague, but it’s hard to be precise when describing everything in the universe. But what has any of this to do with computation?

The answer comes with the final rule. Computation is transformation.

At a single stroke, we turn everything in the universe into an entity that can represent information and be instrumental in the manipulation of that information. Everything. We can control how the computation occurs by modifying the nature of the interacting systems, the context in which that interaction occurs, and the scopes that constrain the interactions.

So we can talk about neurons interacting in the context of the chemical and cellular structure of the brain. By interacting, the neurons change each other. Change the properties of the neurons, and the results of their interaction changes. Change the context (introduce a different chemical environment) and the results of their interaction changes. The “wires” that connect the neurons (axons and dendrites) are the scopes – they determine which neuron can interact with which other neuron. Change the scopes and you change the wiring (you alter the organisation of the neural network).

Or we can talk about ants interacting in the context of their environment. The scopes of an ant include how far it can sense (see or smell) or move. Two ants may greet each other, help each other carry an object, or many other interactions depending on the current state of the ants and the context of their interaction.

Or we can talk about pebbles interacting on a beach, transforming their relative positions in the context of a wave. The scope of the pebbles is determined by their physical proximity – move them apart and they can no longer interact. Place them adjacent to each other, but change the properties of the pebbles or of the wave and the result of the interaction will be different.

We have a lot of different words for objects, interactions, organisations, and environments, but they can all be expressed using this single overarching language of systems, which we call systemic computation. When we construct a machine that operates according to these systemic computation rules, the resulting systemic computer has all of the natural properties we saw earlier. The right kinds of systems are together also provably Turing Complete. The reason why having such a language is important is because for the first time we can integrate the knowledge from computer science and complex systems with all the other sciences. Physics, Chemistry, Biology all become compatible and can be described using the same language of computation. The universe and everything in it can legitimately be called a computer and we can understand exactly how it calculates reality. We can also gain insights into how we might exploit its power.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License