Evolution Reconsidered - 1

 Are we the DNA of evolving machines?  This thought has been the seed of some exciting insights.

Machines evolve, but we downplay this by saying humans drive evolution.  It is not the machines themselves that "evolve." But perhaps we have missed the "essential analogy." For animals, DNA is the evolving blueprint for organisms such as us.  For machines, we have literal blueprints, and humans are the ones that create them, change them and share them with an underlying human goal in mind.  x made a stir by turning our idea of evolution inside out in his "Selfish Gene." According to this story, genes, not animals, compete and evolve.  The analogy I am proposing here is the same: the machine "evolves due to the evolution of designs created and maintained by humans.

Elon Musk (36:00) talks about self-driving cars this way.  He talks about autopilot "evolving" when of course, evolution is being driven by human creativity.  Is this "real" evolution?  Or, on a more fundamental level, does Musk better grasp the "meta" idea of evolution than the biologists dismiss such a claim outright?

This idea has preoccupied me for a few days.  The "concept" of evolution can be used in powerful analogies.  In fact, these analogies are so powerful that evolutionary ideas tend to "slop over" into public speech.  Let's make the core evolutionary concept a bit more formal.

A "frame" is a set of concepts -- perhaps a formal "theory." The "frame" of biological evolution is called  "Biology.".  

The "frame" of biological evolution is a particular case of a category of evolutionary frames, including the evolution of things humans design.  After all, Darwin didn't even use the term "evolution." Sometimes, as in the case of cities and corporations, humans play a crucial role but not the only role in the "design" process.  It is particularly instructive to consider the case where no living agent is present.  Modern "deep learning" systems "evolve" with massive "crowdsourced" human input and are judged "successful" by human beings. Still, the reason why one "AI" performs better than another can be deeply mysterious.  In fact, the creation of effective AIs, such as AphaGo, depends on playing one version against another using a strategy that is tellingly called a "genetic algorithm." 

This is important when we try to imagine how life began in the first place.  What is absolutely essential for evolution to take place?  My guess is this:

• Some energy source allows the process to go "uphill" against entropy.  Sunlight and geothermal energy are examples.
• The richness of "components" in the environment.  Evolution in the game of life is perhaps an extreme example of evolution in the simplest possible environment, but "interesting" evolution seems to need something richer, such as a chemical "soup."
• Variation and inheritance.  "Variation" is assumed by the very term "evolution." and "inheritance" can take place by a single organism varying over time (such as in the Gia hypothesis). In the case of Gia, hundreds of millions of years passed while the Earth "evolved" into a situation where life was possible.

We consider organisms as entities that "evolve." In the biological frame, we speak of a population of organisms evolving, which seems to be part of the particular mechanism of biological evolution.  We will see if it's essential to the phenomenon.  One prominent biologist remarked that evolution could not start with one organism, nor could we think of evolution taking place in the life of a single organism.

It is sometimes fruitful to think of a sequence of re-designed single organisms as a "population." We will look into this when we see if a city can be squeezed into an evolutionary frame.

In biological evolution, DNA and "genes" play a crucial role.  We freely speak of this level of the process as the "blueprint" of life.  This gives us a clue that the biological frame is straining to escape.  Do actual "blueprints" play a role in some other kind of evolutionary frame?

In biological evolution, some mechanisms selectively drive the process.  The most well-known is "survival of the fittest." This idea is famously vague and circular - a problem that won't trouble here.  Other mechanisms can be more precisely specified, such as a shortage of resources leading to the selective success of more efficient organisms.  For the moment, all we need to note is that, in biological evolution, such influence exists.  Less known is that, without such pressure, "evolution" can stagnate for hundreds of billions of years.  "Evolution" itself is not inevitable as in the popular imagination, where evolution "leads" to more and more complex organisms (us).

When you look closely, it is really the blueprints that are evolving.  The organisms (or populations) "evolve" only as a consequence of revised blueprints.  There is presumed to be some kind of feedback from the "environment" of the organisms to the blueprints.

The blueprints are popularly assumed to vary randomly in the biological evolutionary frame.  This idea tends to break down when closely examined.  Only specific changes are possible; some are far more likely than others.  In general, one might say that blueprint changes are constrained by the properties of the blueprint medium - the physical substrate that stores the blueprint - physics and chemistry in the case of DNA, or even more specifically, protein chemistry and architecture.

The idea that biological evolution is "random" seems to come up to contrast with some kind of teleological "purpose" behind the design process (such as God).  Perhaps this goes too far, even in the case of biological evolution.  In other cases, human beings have an evident teleological element.

It is interesting to compare this with, say, the "evolution" of jet transport aircraft, where the constraints of physics and chemistry are paramount.  Constraints on the blueprints (mostly computer models) are hardly limited by the nature of the medium (computer memory).

Do we need something "alive" somewhere in the process, either as the organism or the designer?  I hark back to some advice I got years ago when I treated human organisms as "alive." It's a matter of definition.  Along the way, I have gathered an increasing appreciation of the importance of language - its power and limitation.  Is it intelligible to say that a frame acts "as if" it is alive?  This boils down to the usage of the language.  Who is to be the master?  It is not quite right to ask if Elon Musk is "right" about saying his cars "evolve." The question is, is this intelligible?  Is he using an "essential analogy"?  What makes the analogy "essential" (or useful) is the ability to map back and forth from one frame to another in a way that we can learn something about frame A from considering "similar" frame B.

Language is a vast tangle of analogies (perhaps "alive" itself, but that is a topic for another day).  It plays a crucial role in understanding the world, especially in collectively exploring it.  But it is not the world itself.  On the other hand, as we use more fruitful (essential) analogies, our ability to understand the world improves.  I think of this like the "world" of mathematics - a universe of pure imagination governed only by human logic.  Even in cases where human reason must be left behind (such as in the math of Quantum Mechanics), new notations and a consistent set of rules allow us to make stunning discoveries about the "real" world. 

The universe of language - what can be said - is perhaps a similar case.  The more we can intelligibly say, the better chance we have of saying something that "is the case." Stretching what can be intelligibly said boils down to finding essential analogies.

As an aside, I think it is a grave mistake to assume that "real" knowledge consists only of what can be said - a conceit of philosophers.  My dog knows a lot about the world.  She doesn't understand words the way we do (to her, they are sounds that command special attention or response).  Of course, she doesn't speak at all.  It is also evident that many perfectly intelligible statements are nonsense.  Figuring out which statements are "true" (referring to that which is the case in the world) turns out to be a thorny problem.

I am also skeptical even of the claim that language is the predominant way humans share information.  To spot this mistake, just look around you.  Who told you you could see through windows or open doors?  I maintain that the world's most important "facts" are shown to us, not "told." Further discussion of this insight would lead me far from the topic here, except to say that "evolution" happens in the world all the time whether we know it or not and no matter what we call it.