A Systems View of Markets
Markets as networks.
Complex systems
Life, machines and marketplaces are complex open systems with mechanisms to optimise flows.
There are no truly closed systems or truly open ones, so all ‘systems’ are areas of localised order (the mechanics of the system) with a semi permeable boundary that allows for controlled inputs and outputs while still being enough of a boundary to distinguish the system from its environment .
This order of the mechanism itself is low entropy relative to surroundings (embodied low entropy), it self-configures to maximise flows and therefore net increase of entropy of the mechanism and its surrounding environment, so as not to contravene the 2nd Law of Thermodynamics (entropy increases over time).
This is not an active process as much as a de facto statement of fact, ie large complex systems (turbulent flows, living things, businesses, markets) can only exist if they have large flows and if you have large flows, you can have large complex systems.
A business with a large balance sheet and no cash flow may look fine financially, but contravenes the laws of physics and so will eventually fail. Similarly, relatively low balance sheet companies in the midst of large flows will eventually become larger entities to control the flows in the ecosystem they sit in. This is what is happening with some of the platform Internet companies, most notably Netflix. The idea that Internet platforms don’t necessarily have lower balance sheets was debunked in this joint Anthemis/Temenos paper. In other words, the saying “Uber, the world’s largest taxi company, owns no vehicles. Facebook, the world’s most popular media owner, creates no content…” merely reflects a temporary phenomenon.
If you can actually prove maximisation of rate of production of entropy vs merely its increase over time (which nobody has) then all sorts of things fall out in the wash (including a description of natural selection as a by product of physics and information theory). It would also possibly allow for a genuine science of economics that doesn’t just say people are irrational when it doesn’t work!
Flow triggers & regulators: liquidity and noise (non-transparency)
Using the entropy model to describe flows in markets, the rate of flow is determined by potential value gain, where this is perceived value minus price. The measure of potential value by one participant is different from another’s depending on its state and past history, just like the perceived information in a book is often different for a 20 year old and a 2 year old.
There is a difference between price and value, however, just as there is a difference between information & meaning and energy & entropy. Value, meaning and entropy are all relative measures unlike price, information and energy.
Energy flows across an entropy gradient, meaning is transferred across an information gradient and transactions happen across a perceived value difference. Liquidity in a marketplace is a function of (a) this gradient and (b) the resistance to flow through it (time for settlement etc).
The liquidity trigger (a) is potential value minus price which increases as there is a information disparity between systems (if I offer you a gold bar for a dollar because I believe its worth that, you’ll probably buy it). NB Price is equivalent to average value between 3 or more systems (not 2 because bartering works for that), in information terms.
The information disparity in turn is due to (b) imperfect information flow ie noise (one of us has got some wrong information in the $1 gold bar scenario). Noise is merely information that cannot be processed by a particular system but perhaps could by another (at the extreme, high entropy messages have lots of ‘meaning’/value).
Noise can both increase liquidity via perceived difference in value and reduce transactional flow due to imperfect information flow. In other words, noise increases potential transactions but reduces flows (An analogy here would be electric power where noise is equivalent to an increased voltage (potential transaction) but reduced current (slow settlement) with power remaining constant).
Noise can be as a result of lack of informational transparency but also other factors (ignorance, too much distracting information from other channels etc.)
Overall, the self dampening effects of noise and flows are balanced and self regulating (my guess, 50% efficiency, based on natural digital to analog effects such as music sampling, angular resolution of eyes, paradoxical liquidity reduction due to high frequency trading etc.)
ie rate of production of entropy is limited to a percentage of maximum theoretical rate.
NB its possible that increase in information flows due to the development of the Internet created inevitable reduction in transparency elsewhere (eg dodgy mortgage securities).
Flow predictability: fungibility and diversity of risk appetite (price spreads)
Fungibility
Fungibility is not a feature of currencies, it is the thing that defines them. A currency automatically emerges when everyone agrees something is valuable and it’s not useful, a combination which makes something desirable for everyone regardless of utility and therefore eminently swappable.
As measures of entropy between multiple systems converge (the difference between price and perceived value reduces), money is the by-product when that consensus spreads.
Money is a single, huge, uncollapsed multi-party, chained transaction that exists over a long period of time — until an economy or civilization collapses!
Risk diversity
This is the most important aspect of systems for people whose business is financial services, but is a second order effect and is key to understanding actual dynamics of marketplaces.
Semi open systems are, by definition, far from equilibrium, and markets that are far from equilibrium are ones where there are large second order effects like money creation.
The principal difference, from my limited experience, between financial services and other disciplines is the pivotal role of the measure and role of risk (ie certainty about a measure rather than the measure itself).
Overall model: networks of systems are networks of networks.
You can go slightly further that the typical systems model of nodes and connections and eliminate the difference between the network and the nodes (channels/transactions [network] vs receivers/market participants [nodes]).
The reason you can do this is that the nodes are not ‘things’ with features that are independent of the network itself, but curled up bits of network wiring themselves, just like a whirlpool looks like a circular node in a channel of water like a river but is, in fact, a channel itself.
The structure of a system itself is permanent over multiple cycles, a bit like the water in a whirlpool where the molecules its made of changes but the shape is still there. In other words, systems are actually just structures within channels rather than things at the end of channels such as the senders and receivers in the Shannon model of information exchange. You could in theory create a model of information exchange purely based on the configuration of channels with no sender or receiver.
This means that you don’t need to have a model for how the internal mechanisms within nodes in a network of systems works — they work the same as the connections between the systems. Marketplaces AND their participant are just wiring.
With buyers/sellers & marketplaces, living things & ecosystems and application & networks, the things and the landscape they sit in are made of channels that self configure to maximise flows. You can potentially create models of all of their components that just consist of networks.
