In Why Information Grows, Cesar Hidalgo discusses the origin and growth of information.

Why should we even care? The second law of thermodynamics implies that the universe tends toward disorder over time. However, we live in a pocket of not just order but of increasing order. Order exists at the atomic level, and humans have increased order through our societies. How is this not a violation of the the second law of thermodynamics? This is the mystery to explain.

Hidalgo starts with the physical foundations of order. This part of the book caused a bit of a mindquake for me. First, he establishes the idea that information is a physical property of the universe. Information is a physical property in the way of temperature not in the way of mass. You cannot feel or touch information. Rather, it's a property you can measure. It's a property of how the physical state of the system relates to the state you would expect it to be in given the entropy of the system. In this sense, information and order are physical properties. Information does not imply meaning. Meaning is something created by the decoder when information is decoded (e.g., a human reading a book).

Where does this physical information come from? While the universe as a whole tends towards disorder, it turns out that the spontaneous emergence of order is quite common. Within dynamic systems that are out of equilibrium, the emergence of steady states—of information—is quite common. In systems where there is an excess of energy, the system will tend to have emergent steady states—e.g., the whirlpool in a drain. However, such steady states tend to be transient: dye dripped into a glass will create beautiful swirls of emergent order and then dissipate away into uniformity.

This leads to the second property of the universe that information needs to grow: solids. Solids can encode information in ways that persist over longer periods of time. So information emerges because of excess energy in pockets of the universe. It persists because it is encoded in objects. Why does it grow? Why do the outputs of these dynamic, out of equilibrium systems seem to become more elaborate over time?

Matter computes. Even simple chemical systems can compute in a rudimentary manner. If a system can take multiple steady states, then which of that steady states it takes on is dependent on the specific inputs to that system. In the most basic system, this might just be the amount of energy. Because matter computes, because it can react in ways that reflect its history, we have the potential for information to build on itself and grow.

From that atomic level understanding of why information grows, Hidalgo jumps all the way up to human societies and economies. In human society, it may seem like information is not solid. Information is the ephemeral stuff in our head, right? Hidalgo argues that information embedded in physical objects. They are the "crystallized imagination" which encodes something about the world. However, the intuition that human minds are key is not wrong. The crystalized imagination encoded in physical objects can only be decoded through the knowledge and knowhow of an entity who is using those objects. However, these same objects can also encode knowledge and knowhow. For example, it is not obvious how one uses toothpaste. However, using it effectively does not require knowing about how to make toothpaste, package it, ship it, etc. Physical artifacts thus lead to the growth of information by allowing us to utilize the knowledge and knowhow of others without having to possess it all ourselves.

However, physical artifacts are not sufficient for the growth of information in human society. There is only so much knowledge and knowhow that a single human can possess. Hidalgo calls the maximum theoretical amount that a person can know a personbyte. The point of a personbyte isn't to quantify it. Rather, it's to point to the idea that information is quantized. Without information being crystalized in physical artifacts and without creating networks of people to create larger bodies of knowledge, our knowledge and knowhow will be limited. However, creating a network of knowledge is not easy. It requires both that individuals acquire knowledge and that they can coordinate their knowledge. Hidalgo uses the example of a band: putting together a bunch of people who can play instruments does not create a band. They need to learn how to play together.

These networks of individuals also have their limits. Hidalgo points to theories of firm size described by Ronald Coase and Oliver Williamson. When links between people are cheap, then they can share knowledge and knowhow transactionally, e.g., by selling a tool. However, when links are expensive, then the transaction costs associated with exchanging knowledge and knowhow are high enough that it makes sense to create persistent networks. The theories of Coase and Williamson say that these networks—firms, in economic contexts—will tend to grow to the point where they have absorbed all the transactions that are cheaper to internalize than externalize. Or, if they cannot absorb these expensive transactions, those systems will be hindered by inefficiencies.

Hidalgo calls the theoretical limit to the information which a firm can absorb a firmbyte. Like with personbytes, the key idea of a firmbyte is that there is a limit to the knowledge and knowhow that can be contained within a single firm. Acquiring knowledge and knowhow beyond those limits requires firms themselves to network. Firms may also choose to network if the transactions they need to accomplish their goals are relatively cheap. Even if they could hold that knowledge, it may be more efficient for them not to.

A key factor to the network structure of individuals and of firms is the cost of links. One way to reduce the cost of links is standardization. When two firms can work to a common standard, they don't have to negotiate their interface. The cost of links is also reduced through trust. In fact, trust is critical to information growth and innovation. However, unlike earlier economic theories which hypothesized that trust arose because of the need to facilitate economic transactions, modern economic theories see business as embedded in a broader social context. It is that broader social context of families, friendships, the human relationships at work, etc. which creates the trust that economic transactions need to be successful. Higher trust systems can store more information because they have lower transaction costs.

The rest of the book discusses some empirical support for the idea that knowledge and knowhow are hard to transfer because they are embedded in networks of people and firms. For example, the book explores how industries in a country are more likely to grow by branching out to related industries rather than to ones dissimilar from what is already there. What this implies is that spurring economic growth requires thinking about how to grow individual knowledge and industry based on what is already there. There is also a fun comparison of knowledge and knowhow as means to unpack information in human systems to how information is represented and unpacked in biology.

Overall, this book was a fascinating look into the origin and growth of information. It is one of those books that is less about presenting new facts and more about giving familiar facts an illuminating new frame. Also, it's fairly short! Despite sometimes belaboring its points, it's a worthwhile read! ( )

The first few chapters are a terrific exposition on the natural (that is, not anthropogenic) origins of information using the models of Boltzmann, Shannon, and another guy whose name escapes me. I had to stop reading at around page 83 because I lost interest at the same time that I lost the book. ( )