Thursday, July 27, 2017

A New Theory of Life

This is a draft of Chapter 1 in the book outline


A New Theory of Life

1.1 Puzzles

Life defies the second law of thermodynamics. Or at least that defiance is suggested by a preliminary understanding of the second law. There is no perpetual-motion machine. Every system, including any machine or any living organism, when considered by itself alone and separate from the outside, must consume the usable energy with which it starts, must eventually run to a stop or die. Yet life carries on, for billions of years now so far as we know.

This puzzle was one of the many things I learned from my Ph.D.-engineer mentors during the most educational year of my life. In 1973, when at the completion of my B.S. in electrical engineering I had failed to gain admission to any medical school, momentum in that same med-school direction combined with luck got me a job as a Bioengineering Research Assistant at Harvard Medical School in Boston. John L. Lehr, the youngest of these Ph.D. mentors, told me about the second-law/life puzzle. The puzzle settled in my head as one of the things I wished I could understand better, as one of the questions guiding my curiosity.

Over the course of many years I formulated my own answer to the second-law/life puzzle, and that answer is the subject of this book. But there is much more here.Other deep and difficult questions have perplexed me. Two of these are:
  1. Why does planning succeed for some of our human organizations and not for others? To explain, most economists now believe that socialism on a national scale is doomed to fail. The failure of Soviet communism supports this belief. But planning and central control of smaller economic organizations, such as retail stores, seems to succeed. If planning can work for businesses, why does it seem doomed to fail on the national scale?
  2. In elections in the US, the residents of cities tend to vote for Democrats, or representatives from the left, while rural residents tend to vote for Republicans, representatives from the right. I do not believe that either side can fairly be dismissed as stupid or evil. So what explains this undeniably consistent trend?

Once again I have discovered workable answers to such questions, after such questions have resided for decades in the back of my thinking. The theory presented in this book shows a reader how to answer those questions, and how to explain the answers.

1.2 Hypotheses and computer loops

When I was about 30 I wrote down another question, or actually a tentative hypothesis, which promotes the idea of ‘hypothesis’ to center stage. I had been wondering, how does my mind work? How does any living thing choose its actions so as to survive rather than perish?

Extending from my own mental experience as well as I am able to observe it, I decided that maybe, probably, every one of my thoughts and actions was simply a hypothesis. I can never be totally sure of my thoughts, and any action which I make, even an action in which I had previously felt great confidence, might fail for some reason I had not anticipated.

Yet I succeed often enough to have survived thus far in spite of the refutations which life, and greater circumstance, deliver to me and my hypothetical thoughts and actions. And this survival in spite of possible error seems to characterize the existence of other now-living life forms. The explanation for this survival seems to lie in relationship between the living thing and its environment. If the good hypotheses are rewarded handsomely or frequently enough by the environment, and the bad hypotheses are punished mildly enough by the environment, then the living thing which lives by testing these hypotheses has a chance to survive. If the living thing has a way of remembering its successes and failures, and if the environment has some regularity, then a living thing given some mental or calculational capacity, in addition to memory, may improve its adaptation to its environment.

I wanted to test this theory, that the key to life might be simply a strategy of repeated trials guided by a growing memory, in computer models. I had loved computer programming since I had first experienced it, using Fortran, as an undergraduate in 1970. So it was easy for me to see how I could write a beginning try-and-remember loop, to model a very simplified life form. I was excited by this idea as I started a graduate program in computer science in 1982. After two and one half years in that program I passed the doctoral written exam, gaining admission into the dissertation phase. But try as I might I was unable to find, in that department at that time, enough faculty members who felt interest in my proposal and who would serve on my dissertation committee. And unfortunately I felt little excitement for any of the research projects going on in that department, projects which I might have joined to find a dissertation topic and faculty adviser. I took a leave of absence from that graduate program to take a job as a carpenter, thinking I would return to the program after one year.

But the environment provided an unanticipated positive reward for my shift to carpentry. I was soon answering many calls from friends and professional acquaintances who, learning that I was carpentering, wanted me to work on their houses. During the next few years I completed many remodeling jobs, which kept getting bigger, and I got two North Carolina licenses: as building contractor and plumber. I was able to start a business of designing and then building spec houses.

So I never finished the Ph.D. in computer science. Without planning I had stumbled into a way of making a living which, I realized, was better for me than if I had completed the Ph.D. I had more power over my own choice of direction than if I had become a professor. And my dissertation topic lives on! This book is my dissertation and you, reader, a member of my committee.

1.3 Other questions answered by this theory

In the section which follows this I will finally get around to stating my theory in concise terms. Unfortunately those concise terms may not prove evocative for you without the coaching offered in later chapters. So before we take the step of introducing the theory in concise terms, I will offer more encouragement for you to stay the course by listing additional important issues which this theory can enlighten. Stick with me until you learn how to think within this theory and you will be rewarded with deep and powerful ways to answer these questions.
  • Why do people talk past each other?
  • If anger is bad, something that mature people learn to suppress, why is it so common, so instinctive?
  • If humor or, more specifically, laughing together is good, why are some people so commonly injured by someone else’s joke?
  • If, as Darwin suggests, evolution of life as we know it proceeds by variation followed by natural selection, how does nature select its survivors?
  • What hope exists for the future of our human race?

1.4 Assumptions of the theory

Now, having laid out the promise of fruitful new understanding which a reader of this book may gain, we will steer into the presentation of the model. What follows attempts to give a formal and systematic presentation of the model which, for a reason which should soon become clear, I have named the Resource-Patterns Model of Life. We start with an outline of the assumptions which underlie the model.

Our basic assumption is that Living Things exist in a Universe. So now we will review our assumptions about that Universe, and then about Living Things.

1.4.1 Properties of the Universe

space and time
The universe has one or more dimensions. Time passes in the universe.
living things
The universe contains living things.
resources for present living things
The universe contains raw materials and energy of the sorts required by living things.  Most of these resources are distributed in concentrations, i.e. in patterns. Such patterns of resources give rise to the prospect that living things may discover and exploit these patterns.
resources for future living things
Patterns of resources vary widely in size, from tiny (perhaps atomic or subatomic) to huge (galactic or larger). Smaller patterns support microscopic life which we can see, while the larger patterns lie beyond the reach of present humans. 

1.4.2 Properties of Living Things

Living things can detect certain aspects of their surroundings.
ability to act
Living things can attempt to act in particular ways.  Many such acts involve motor or muscular movement.  But other possible acts might be to wait idly or to calculate without moving. Living things can act to imbibe resources or to reproduce themselves. Note however that a living thing's choice to act in a particular way does not guarantee that the attempted action will succeed.  Each attempt by a living thing to act might succeed or fail, depending upon circumstances.
Living things have goals.  Typical goals might be to imbibe the resources necessary for life, to reproduce, or to gain security.
Living things have some ability to store a record of their experiences.
calculating capacity
Living things can “think” about how to act.  Typically this calculation might consider: (1) the present state of the environment as determined through senses; (2) present goals; (3) present store of resources; (4) memory of prior similar experiences.
resource consumption
Living things necessarily use up some of their store of resources in each increment of time.  The amount of consumption may depend upon the action undertaken.
resource storage
Living things can store some of the resources which are necessary for their lives.  Typically living things can store an amount of each resource sufficient for multiple time increments, so that living things can spend some of their time in actions other than imbibing.
nondeterministic choice of actions
Living things will employ their memories and calculating capacities to guide their choices of actions with as much "wisdom" as they can muster. But they will commonly find themselves with no definite knowledge about how to act. So, in order to avoid starvation which will certainly come if they remain idle, living things will sometimes guess how to act, selecting an act at random if need be.
life in levels
Living things are usually composed of a large number of smaller entities which in turn appear to be individual living things on a lower level. But we can look in the other direction as well, to higher levels. As we humans go about coordinating our affairs with others on our level we are testing organizations which, when these organizations become successful, acquire some of the properties of living things as listed above. Through such coordinating action we humans may eventually create organizations with all the properties of living things on a higher level. Such organizations would in fact be individual living things on that higher level, in view of this model.

Obviously these assumptions have been designed in sympathy with humans.  The living things could be us humans.  The universe could be the Earth.  But the model allows us to look at other implementations, at other “living things” and other “universes”.  These other implementations will ring with suggestions about our existence as humans, about our social orders.

Here I have listed many assumptions.  But I highlight one of the assumptions in the name which I have given to the entire model: the Resource-Patterns Model of Life. This one assumption about the distribution of resources leads to valuable and useful suggestions. Our social orders often reflect patterns of resources in the universe. And the direction which resource patterns impart upon our social orders has been overlooked by social science to date, as far as I have been able to determine.

Hereafter I will commonly abbreviate: Living Thing as LT; Resource Pattern as RP; and Resource-Patterns Model of Life as RPM.

1.4.3 Operation of the Resource-Patterns Model of Life

The resource-patterns model gives its user a way to think about certain sorts of problems. After we accept the above-listed assumptions, we may deduce from those assumptions a set of guiding principles which I outline here under four points.
  1. Living Things survive by finding and imbibing resources.  If LTs don’t find enough resources their numbers will decrease.  If LTs find abundant resources their numbers can and probably will increase.
  2. In each increment of time each LT has a range of choices about how to act.  Probably most of these possible actions will be useless in that these actions will not contribute to the effort to imbibe resources.  So a LT needs to narrow its range of choices.  This focusing of choices is the principal requirement of the LT's calculating capacity.
  3. Any particular supply of a necessary resource must be finite, assuming that this supply has been discovered by LTs at a particular place and time.  This supply can be exploited only until it runs out.  Ongoing life therefore requires an ongoing discovery of new supplies of necessary resources.
  4. Cooperation may help LTs to exploit some RPs.  Consider three cases:
    1. Some resources are abundant but far away, too far away for a single LT to exploit.  But such resources might be exploitable if a number of LTs combine in a linked network of trade.
    2. Other resources are near at hand but too difficult to extract without specialized tools or knowledge.  Such resources might be exploited if specialized LTs cooperate.
    3. Some resources may be extracted only through an effort which continues during a span of time. It makes sense for individual LTs to participate throughout that span of time only if the environment is stable and predictable.  The environment can become more predictable if the future behavior of other LTs becomes more predictable, if the LTs can somehow form rules of cooperation.

Thus, if a set of LTs can discover modes of cooperation, that set of LTs may flourish in an environment where a similar set of LTs, but without cooperation, would perish.

1.5 RPM uses Agent-Based Modeling

In the past few generations, with the increase of availability of computers, a new method of modeling for social science has grown called agent-based modeling ABM. The approach we will use in RPM can be called ABM.

ABM uses a number of small-scale entities called agents which interact with each other and their environment for the study of larger-scale consequences. The modeler adjusts the capabilities of the agents, either individually or in whole groups, and then models the passage of time by dividing time into moments or what we will sometimes call cycles. In each cycle each agent responds to its circumstances and acts within its capabilities. All together, with numerous agents acting during numerous cycles, group- or society-wide consequences often become evident to an observer with a society-wide viewpoint, to us modelers that is.

As you have probably anticipated, the LTs, which we have described as parts of RPM, are agents, typically, in our modeling.

Now we will compare two ways to implement agent-based models: computerized agent-based modeling (CABM); and thought experiment agent-based modeling (TEABM). Both of these methods of modeling have been used by me in developing my current understanding of RPM, and both methods influence my assertions in this text.

For some people, such as I who love computer programming, computerized ABM is very inviting and enthralling. RPM simply begs for CABM since it lays open many tantalizing problems which may be modeled by application of a little programming wizardry. But experience has shown that these computer models routinely get tangled in complexities which are not evident, for the most part, until a modeling project is underway. Nonetheless that difficulty can be a blessing: CABM promises to teach its practitioners a great deal about the assumptions which our human minds tend to make as we think about events. Thus CABM promises maximal education for human modelers – but at an expense in time and effort which is probably prohibitive for most aims of research.

The thought-experiment method of modeling, on the other hand, allows its practitioners to leap into any thicket imaginable. With minimal preparation the thought-experimenter can start to explain – at least to the satisfaction of the thought experimenter – what has happened and what will happen in that thicket.

Almost all of what I present in this book may be described as the results of thought experiments. So I should not discredit this method too severely. Einstein, after all, used thought experiment with great success. I believe thought experiment is appropriate and perhaps necessary in the early development of a subject. Eventually, if my RPM subject remains viable, terms will be defined and ways of measuring will be discovered. More exact science should ensue, including much fruitful CABM.

Given that the subjects we will discuss in this text will employ the method of thought experiment almost exclusively, I must warn, both the reader as newcomer to RPM and myself as leading modeler, of the limitations which anthropomorphism brings to our modeling. We will be modeling both the internal “nervous-system” workings of individual living things and the social interactions and thereby the achievements of groups of LTs. It will be too easy and tempting for us modelers, when setting a LT/agent in a problem situation, to assume that the LT/agent can sense and think as one of us humans can sense and think. But when we do that we rob ourselves of the power in RPM – to learn bit by bit about how our human minds and social interactions grow of necessity in the problem situations which we will model. We need to learn to notice carefully when we give extra powers to the LTs in our though experiments; probably we should write down these powers in a tabulation which we keep. This discipline will regain for us some of the teaching power which we set aside when we decided to model in thought experiments rather than in CABM.

1.6 About the Approach Taken in this Book

Here we will pause to consider the approach I have decided to take in writing this book for you.

Compare two styles of book: a college textbook for an introductory course in a subject; and a page-turner which keeps the reader engaged in development of its subject. Most of my previous experience with writing has emphasized the second style: I have tried to catch the reader’s interest, perhaps with reference to some issue on everybody’s mind today, and then to make the flow of development hold the reader’s attention through to the end.

Unfortunately, since I assume you would rather read a page-turner, I believe that my first book on this subject should cover the ground once. I aim to lay down the canon of RPM; I believe this should be my priority at this time. After I have completed this responsibility I will be able to imagine that my readers have a copy of this reference text on their shelf. Then I expect I may spill out many pieces of writing which are both more engaging, because related to an issue of immediate interest, and enjoyable for both reader and writer. In these subsequent texts I may refer often, for fuller explanation of some concept, to this canon on your shelf.

1.6.1 Choice of Language: Everyday Language vs. Specialized Terminology

Again and again I have faced a choice as I have attempted to communicate the essence of this new scientific model. Should I employ the everyday term, a word which will be widely understood, to name a specific concept within RPM, or should I invent a new model-specific term, the meaning of which must be learned by a newcomer?

As  you might expect I have tried to take the easiest path. This is to use the familiar term – that term with which indeed I first labeled the concept in my own thinking – in a narrowed way, in a new way with particular meaning within RPM.

This use of everyday English terms has a downside of course. A reader who has not labored to learn the RPM-specific meaning of an everyday term found in my text probably has no chance of understanding my meaning, but this newcoming reader may quickly and easily gain an erroneous impression of my meaning.

There is I believe no shortcut. A newcomer to a discipline must learn the particular ways of thinking, ways employed by seasoned practitioners, associated with the discipline-specific terminology. This remains true whether the terms are reemployed familiar words or newly-contrived for RPM purpose.

To help the reader with this learning of model-specific terms I have added a glossary at the end.

1.7 Outline of Contents to Come

Chapter 2, Tabletop Critters and Other Examples
Here we will see a few easily grasped examples of how RPM can explain and predict group-based behavior. Then we will look in more detail at a model of very simple LTs living in a difficult – but still very promising – environment. I will call these LTs “tabletop critters”. Our modeling here is done entirely in thought-experiment mode. Tabletop critters are the workhorses in my modeling to date.

Chapter 3, Activity and Abilities of the Critters
Here we will take a closer look at the individual agents, or critters, in our agent-based modeling. We will list more particularly what a critter can sense and do. We will consider critters at three stages of development
1) with enough abilities to probably survive in the initial condition
2) with enough additional abilities to exploit a first simple RP which we offer
3) with further abilities which – as we need to study – may enable a set of critters to spontaneously discover and exploit a RP.

Chapter 4, Life in Levels
We will review the evidence that life on Earth as seen by biologists has grown from level to level, for example single-cellular to multi-cellular. Then I argue that a similar growth, to a next higher level, goes on among us humans as we organize ourselves into families, businesses, and states. Our modes of social cooperation are complex and seemingly related to RPs.

Chapter 5, The Learning of Rules
LTs succeed in their lives, both as individuals and as members of organizations which succeed, by behaving under the influence of rules – rules which must somehow be acquired. While in early and simple cases we modelers can implant rules into our agents, this chapter starts to look at the complications which naturally arise as we modelers try to empower our agents to learn their own rules. One subject which arises here is preprogrammed or instinctive ethics as expressed among prospective groupings of LTs.

Chapter 6, Philosophy in RPM
RPM gives us a new way of looking at many longstanding questions of philosophy. The agent based modeling we do in the context of RPM opens the workings and motivations of modeled LTs, on a workbench so to speak. We see the stages of development for a need for ‘truth’. We see how agent-to-agent signals become a working language. In the agents of advanced thought experiments we see a combination of computable routines which, taken together produce effects which we would be challenged to distinguish from consciousness.

Chapter 7, Public Psychology
In RPM we see survival and reproductive fecundity awarded to whole organizations of LTs which succeed in cooperating among themselves to exploit RPs. So we naturally expect most LTs, being descended from these populous organizations, to have biases and instincts of their forebears which helped to develop and maintain those organizations. Our thought-experiments lead us to expect many forms of group-think, anger, deception, suspicion, and even humor.

Chapter 8, Public Policy
A particular category of organization of LTs, which we call “government” or “a state”, can be distinguished. Restraints or commands, which we humans call “law” are gathered under control of such an organization, and are often considered to be “public policy”. In RPM we can see public policy as a particular application of public psychology.

Chapter 9, Conclusion
We review the new contribution to social science made by RPM, with RPs which need to be discovered by as-yet-unknown methods. Cooperation, in as-yet-unknown ways, may be discovered without planning or foresight by any of us LTs. This way of modeling social life gives us great challenges and opportunities for understanding ourselves better.

No comments:

Post a Comment