Saturday, June 2, 2012

Critique of Bejan's "Design in Nature", Part II

So let’s get to it.  Bejan’s constructal law is this:
For a finite-size flow system to persist in time (to live), it’s configuration must evolve in such a way that provides easier access to the currents that flow through it.
He then elaborates a little:
…Everything that moves, whether animate or inanimate, is a flow system.  All flow systems generate shape and structure in time in order to facilitate this movement across a landscape filled with resistance (for example, friction).  The designs in nature are not the result of chance.  They arise naturally, spontaneously, because they enhance access to flow in time.
My initial concern here is his equivocation of the word “design”, which he uses for both inanimate phenomena like lightning, snowflakes and river deltas, and for actual engineered systems, such as airplanes, cars and modern infrastructure.  This is a common tactic used by the Darwin lobby.  Life is best described as  engineered systems, in that the appearance of design is breathtaking, especially as we learn more about it.  The similarities between life and systems engineered by humans is so striking, in fact, that Atheists and/or Naturalists must engage in wordplay in an effort to detract from the painfully obvious.  And it appears that this is exactly what Bejan is attempting to do.

But the basic premise is this: flow systems tend to find the path of least resistance.  When a river makes its way to the ocean and begins to branch, the water cuts its way through the landscape according to whatever dirt and rock gives way first, thereby forming channels in those areas, which forces continued removal of dirt and rock along those channels. Also, water tends to flow downhill, not uphill, so we can expect to see the water to also flow with gravity, which is also a way in which it finds the path of least resistance (duh).  It is the same with lightning and snowflake formation.

He then lumps life and engineered systems into the flow system category by saying that they, too, tend to find the path of least resistance.  And he points to the branching of trees, lungs, vascular systems and cooling fins.  He also points to animals and says that they have evolved to move mass across the landscape in a very efficient manner, i.e. their designs have been configured to minimize the work required to move them across the landscape.  He is correct that engineered systems and life are flow systems, but to say that they are no different than simple law-like physical phenomenon is way off the mark.  In fact, I would say that his theory is sophomoric.  So then what are the differences between unbridled energy flows and engineered systems? I can identify at least three dramatic differences off the top of my head: 

Engineered systems utilize energy flow, they are not a manifestation of such flows due to pure physicality.

The river flows downhill due to gravity alone, and the channels carved into the landscape are a result of the loosest materials giving way first, which, again, is a manifestation of pure physicality.  However, the hydroelectric dam harnesses this energy flow, or redirects it to serve a purpose.  The dam, or barrier, is put in place to maximize the potential, and the generator transforms the potential energy into electrical potential, and the power lines transmit that potential to the end user, thereby fulfilling the purpose of the dam itself.  Likewise, there exists a proton gradient across certain membranes in the cell.  ATP synthase is a protein complex that utilizes this potential to achieve an end, instead of just allowing the gradient to flow across the boundary without doing anything useful.  The membrane creates a barrier to maximize the potential, then the ATP synthase complex not only channels the flow, but transforms the potential across the membrane into chemical potential (ATP) which is then utilized to power most of the motorized goings-on in the cell.

There must be a choice contingent arrangement of matter that cannot be accounted for by pure physicality in order to accomplish this harnessing of available potential.

There is no physical law that can explain the composite concrete and steel arch needed to dam the river.  And there is no physical law that would cause the windings within the generators to be drawn out to a precise gage thickness, or to be coated with resin or plastic, or to be arranged into a specific number of loops and orientations about the armature.  These qualities are instantiated into the matter by a designing mind, which is what I mean by ‘choice contingent’. Likewise, the ATP synthase proteins are arranged in such a way as to have an actual generator, complete with ion channels, a rotor and a stator, configured and oriented in such a way to allow for one degree of freedom to rotate and capture the energy provided by the proton gradient.  

The best way to see the difference between natural phenomena and engineered systems is to picture a hurricane.  A hurricane is a heat engine that forms because of the heat energy flowing from the surface of warm tropical waters to the upper atmosphere.  A very predictive structure results, along with very destructive winds which perform ‘work’.  But it must be noted that the energy flow itself is what builds and maintains the structure.  If the energy is removed, the structure dissipates.  This is not so with engineered systems which only channel the energy and are not maintained by it.  When the energy flow is removed, the structure remains until the energy flow resumes.

Sophisticated engineered systems are algorithmic, whereas the behavior of natural phenomena is governed by pure physicality.

Hurricanes, rivers, stars, etc. are predictive because they are governed by laws, and can be expressed as a set of equations.  Projectiles trace a path through the air in accordance with an initial velocity vector and the pull of gravity.  Hurricanes rotate in accordance with the rotation of the Earth and their strength waxes and wanes in some proportion according to the magnitude of such variables as the heat gradient, pressure, humidity, etc.  So we can describe the behavior of natural phenomena with a set of equations with variables representing aspects of the environment.

But integrated engineered systems perform in accordance with a set of instructions, execute logic and looping, and calculate outputs based on inputs.  Consider an automated machining cell that produces widgets: a box a raw material is loaded onto a platform, which satisfies one or more on/off conditions, such as whether or not the box holds open a switch, or if the operator initiates a program by pressing the ‘go’ button.  If these conditions are satisfied, a robotic pick-and-place device begins execution of a program which picks up each individual raw widget from the box (which requires step-over looping), and places the raw widget into the first machine.  If another series of on/off conditions is satisfied after loading, such as a bank of sensors that detects whether or not the part is loaded against the datums, the first machining center initiates another program that turns on motors and pumps for spindles and hydraulic slides and such.  Once these programs have completed execution, and another series of on/off conditions is satisfied, such as sensors that detect the end position of the slides, a signal is sent back to the robot to unload the first machine, and then carry the load to the next machine.  The automated robot cell is executing a plan, and making ‘decisions’ based on inputs received from the environment or components of the system itself. It is not a set of equations that governs the system, it is a set of instructions, or an algorithm (fig 1).



fig 1

But to which of these does life compare?  Life is without question algorithmic.  Consider the assembly of the flagellum, which must be assembled according to a very exacting plan; a procedure which if not followed perfectly will result in the failure of the flagellum’s construction.  The plan executes when the cell senses a signal to start building a flagellum, which initiates a cascade of controls, starting with the manufacture of a protein that binds to and flags the section of the DNA that codes for the base of the flagellum, effectively turning it ‘on’.  So we start with an ‘on’ signal, which initiates a program that executes the construction of a marker that when bound to a particular section of the DNA acts as another ‘on’ signal, which initiates another program that starts production of the flagellum proteins that form the base of the structure.  Each stage of production is initiated in this way and will not begin until the previous phase is completed and sends an ‘on’ signal to initiate the next stage, just like the robot cell.

[D'Onofrio, Abel and Johnson have illustrated the algorithmic properties of protien synthesis in this article. And here is the flowchart.]

It should be noted that the algorithmic nature of engineered and living systems is only made possible because of the special arrangement of matter, not the governance of the laws of physics.  If the sensor in the robot cell is not positioned in just the right spot, it will not be able to detect whether or not the part is loaded correctly, and will not ever register an ‘on’ signal to initiate the next step in the process.  It is the laws of physics that makes the sensor work the way it does, but it is the choice contingent positioning of the sensor that enables the algorithmic quality of the system.  Likewise, the protein that binds to the DNA that codes for the first stage flagellum construction must bind in the right spot, or else the wrong portion of code will be switched ‘on’, or it will not switch on anything at all.  It is the laws of physics that causes the protein to bind to the DNA, but it is the special arrangement of amino acids that causes the protein to bind where it does.  But rivers and hurricanes and such are manifestations of pure physicality.  They do not execute a plan, they only behave according to a set of governing equations.  These phenomena will follow the equations regardless of any special arrangement of matter.  The arrangement of matter for natural phenomena serves as variables for the equations that describe its behavior, but the arrangement of matter for the algorithmic system only provides an ‘on’ or ‘off’ condition, and there is a very narrow range in which an ‘on’ signal can be realized, but only because it corresponds to another aspect of the system that is ‘looking’ for an ‘on’ signal at that precise arrangement.  Intelligent Design proponents would call this a ‘specification’.

In part three, I will use these distinctions to further critique Bejan’s concept.