This is the chapter I would have loved to have been able to write. My career was in digital computer design and fabrication and I am very interested in emergence so a chapter on computers as examples of emergence is right down my alley. Of course, Ellis does a far better job than I ever could.
Anyone who has written software programs or done circuit design for chips or any microprocessor architecture work will find this chapter easy to understand. Anyone who hasn’t done anything in this field is well advised to simply skip it. Ellis shows how a digital computer system exhibits top-down causation by emergent entities. With well-known concepts, he clarifies the concepts of emergence.
He describes computer systems as having two primary hierarchies. One is an implementational (vertical) hierarchy and the other is a logical (horizontal) hierarchy. Each hierarchy involves many levels of modular structures. For example, in the implementational hierarchy, one of the lowest levels is the atoms and molecules level. A higher level is the transistor and wiring level that form the physical layout. Above that, the transistors are combined into logic and memory circuits while above that, the circuits are combined into modules such as the arithmetic control unit, the central processing unit, etc. The highest level is the physical interface with humans. On the software side, the lowest level is the binary code that runs on the hardware. Next is assembly language and sequentially higher-order compilers and sophisticated programming languages. Modular units like subroutines are closed entities comprised of lower level commands.
Ellis explains “Each higher level behavior emerges from the lower level ones. But what ultimately determines what happens? The higher levels drive the lower levels.” This, he claims, is the essence of top-down causation. It is intrinsically coupled with bottom-up causation since the behavior of the lower levels causes the desired action at the higher levels. The higher levels are characterized by what can be considered abstract entities that interact and cause the appropriate desired action at the lower levels.
Then Ellis uses a term which I believe is potentially confusing and needs some discussion. He emphasizes that the entities in the higher level logical hierarchy are non-physical. He states that “Software is not a physical thing, neither is data. They are realized, or instantiated, as energetic states in computer memory. The essence of software does not reside in their physical nature: it is the patterns of states, instantiated by electrons being in particular places at a particular time, that matters.” What does it mean to be “non-physical?” As Christians we are very comfortable with non-physical beings and we quickly associate non-physical entities with spiritual ones. But I do not believe Ellis is using the term this way. By non-physical, I believe he means an entity that is not uniquely associated with a physical state. It may be the result of any number of physical states but its existence does not depend on any specific one.
While Rolf Landauer reported to me at IBM Research in the late 90’s, he told me in our personal discussions that “Information is physical, but independent of its embodiment.” I’m still struggling to understand this more fully. Information can be transferred from one physical embodiment to another without any necessary energy dissipation but deleting one bit of information necessarily dissipates kt ln 2 of energy. I’m also trying to figure out whether Landauer would agree with Ellis about software being non-physical because it is independent of its physical embodiment. I suppose being independent of physical embodiment may not be sufficient to warrant the term “non-physical.” In any case, we cannot assume that either Landauer or Ellis are thinking of spiritual-like entities when they talk about “non-physical” entities.
Finally, Ellis closes this chapter with this gem: “At a higher level, the existence of computers is an outcome of the human drive for meaning and purpose: it is an expression of the possibility space of meanings, the higher levels whereby we guide what actions take place.” As is true of so many sentences in this book, I have to read it many times to think it through. And maybe it wouldn’t make sense without reading the whole chapter. But I was in the business of designing and building computers all my career and never thought of it this way.
What he seems to be saying is that as we go up the physical and logical hierarchies, entities emerge from the lower levels with increasing abstraction. Finally, we get to the top levels where meaning and purpose reside. The closely linked relationships of bottom-up and top-down causation lead to the possibility of meaning and purpose, aspects that do not exist at the lower levels. We often say that science cannot address meaning and purpose but in Ellis’ view, emergence allows us to recognize meaning and purpose at the top levels even though it is absent at the lower levels. This I must contemplate for a while.