# The Sciences of the Artificial Author: Herbert Simon #### Summary, Key Themes --- ### Rough Chapter Notes #### Chapter 1 - Understanding the Natural and Artificial Worlds * Artifacts* refer to anything of human creation, such as an airplane, a plowed field, or asphalt. Artifacts have no predisposition to ignore or violate the laws of nature, but at the same time they are *adapted* to human goals and purposes. * Generally we associate engineering as being concerned with *synthesis*, while science is concerned with *analysis*. The scientist is concerned with how things *are*. The *engineer* (more generally the designer), is concerned with how things *ought* to be (in order to *attain* their goals and to *function*). It is worth noting that natural science has no *external frame of reference*, so bringing in judgments and ought's is dangerous. However natural science *provides* the external frame of reference (the context) that the artificial inhabits. Given this context (ground truth about reality), it seems reasonable to think that there is an object *best* way things *ought* to be. In other words science tends to be purely descriptive, while the science of the artificial will likely be [normative and descriptive](Dichotomy-of-Normative-and-Descriptive.md). * The sciences of the artificial have four key boundaries that we must keep in mind: 1. Artificial things are synthesized (not always with forethought) by human beings. 2. Artificial things may imitate appearances in natural things while lacking the reality of the later. 3. Artificial things can be characterized in terms of *functions*, *goals* and *adaptations*. 4. Artificial things are often discussed in terms of *imperatives* as well as descriptives. * **Functional Aspect of Artificial Things** * Fulfillment of purpose or adaptation to a goal involves a relation among three terms: the *purpose/goal*, the *character* of the artifact, and the *environment* in which the artifact performs. These terms implicitly *characterize* the artifact. * As a an example we can consider a clock. It has: * **Purpose**: to tell time * **Character**: it's gears and internal mechanics. * **Environment**: on a ship, in direct sunlight, etc. * And, finally, it can have an **adaptation** to it's environment: if the environment is a ship, the adaptation may be specialized internals to accurately track time on rough waters. * Natural science impinges on an artifact through the character (structure) of the artifact itself and the environment within which it resides. Whether a clock will tell time depends on it's internal construction and where it is placed. Note: the portion of the artifact that natural science does not directly impinge upon is it's *goal*. * An artifact can be thought of as an **interface**, a meeting point between an inner environment (internal composition and structure) and an outer environment (where the artifact operates) * We define things from a *functional* perspective. For instance, the color of a rabbit in the arctic white because it allows it to hide more effectively. We have just defined the color of the rabbit functionally. Couple with the theory of evolution and natural selection, we have an [Explanations](Explanations.md). * So, a large advantage of separating the outer from inner environment when studying artificial systems is that we can often predict the behavior from knowledge of a system's goals and it's outer environment, with only *minimal assumptions* of the inner environment * Likewise, we often also encounter an advantage of this separation from the vantage point of the *inner environment*. Frequently inner systems are insulated from the outer environment, so it need not be as heavily considered. * In the best case scenario: > We hope to characterize the main properties of a system and it's behavior without discussing the detail of *either* environment (inner or outer). We would like to create a science of the artificial that would depend on the simplicity of the *interface* as it's primary source of abstraction and generality. * Central to the description of artifacts are the *goals* that link the inner to outer system. * The inner system is an organization of natural phenomena capable of attaining the goals in some range of environments. The outer environment determines the conditions for goal attainment. If the inner system was designed properly, it will be adapted to the outer environment. * Simulations can tell us things that we do not already know in two ways: 1. Given a set of initial assumptions/condtions that we are confident in, it can still be very hard to reason about their outcome. Simulation can help in this case. As an example consider weather systems. 2. If we do not know much initially about the natural laws governing the system we can *abstract* what we deem to be the essential properties. This makes it easier to simulate the phenomena. This is particularly appropriate in the case where we are interested mainly in the *organization* of the parts, and only a few properties of the individual components. This is most appropriately demonstrated in terms of the computer. We can learn a great deal about computers without actually knowing the internal components that are being utilized. Put another way, we are concerned with the *function* of the components, not the details; i.e. we are concerned with the abstraction. * Empirical theories are those that are based on observation. The key component to these theories is that they could not have *predicted* the outcomes of an experiment via a chain of *logical deduction*. Instead the systems must have been created and then *observed*. This is what makes them *empirical*. *