Defining "complexity"

In discussions of “Intelligent Design”, the claim is often made that it has not been demonstrated that unguided processes are able to increase the “complexity” of a biological system.

However, lack of agreement over how “complexity” can be measured or defined often impedes meaningful discussion of this claim.

Here’s a question to helpfully resolve this ambiguity: Suppose an engineer has added a new component to an internal combustion engine. He has taken a fuel injection system and modified the arrangement of its components such that the engine can now operate with canola oil as a fuel, in addition to using gasoline.

The original fuel injection system remains intact. The engineer has merely duplicated and rearranged its components, then added the new device to the existing engine. The engine can still operate using gasoline as before, but now it can also use canola oil if necessary or desired.

Has the engine been made more complex with the addition of this new component? Why or why not?

I’ll leave this vexed question to others. But let me just note that the word “complexity” is used in the arguments about Intelligent Design in a non-obvious way. The phrase “Complex Specified Information” comes from Leslie Orgel (1973) via Richard Dawkins. Both of them were not using “complex” to mean complicated. Instead they were using it to mean that being this well specified (say, this high a fitness or higher) was very improbable. This “complexity” can be achieved simply by having enough simple adaptations that don’t particularly interact with each other. Further argument about whether Complex Specified Information can be achieved by natural selection and other ordinary evolutionary processes then does not at all involve any consideration of complicatedness. I think that, actually, this was a poor choice of words on Leslie Orgel’s part. Good luck in finding a clear definition of “complexity”, but keep in mind that for arguments about CSI, that will be more or less irrelevant.


Complexity science is a different approach entirely from the ID conversation, but one definition of ‘complex’ versus ‘complicated’ is ‘not susceptible to reductive analysis’. Complex behaviour is emergent, and not easily explained in terms of all the component parts.

I think in ID, on the other hand, complexity is defined almost entirely in terms of reductive analysis.

A messy room is very complex, is it not?

Arguably, just complicated.

I might not have said this clearly enough, so let me give it a try.

When Behe says ‘irreducibly complex’, he means ‘if you remove a piece it stops working’. But that is just complicated (in the complexity science language I’m using here). That is, we understand every single piece, reductively, in the engine of my motorcycle. Every piece has an explainable role and job in making the engine run and run efficiently. There are some single pieces that, if removed, will make the engine stop working entirely (either immediately or over some time span), while others will only reduce its efficiency.

That’s a little different from ‘complexity’ that is emergent, and that can’t readily be explained in terms of pieces of the system. A nephew and I have long afternoon arguments over a beer or two about whether consciousness is such a phenomenon, for example. Can we eventually, even if we can’t now, fully explain consciousness in terms of the operation of neurons and synapses?

I think this is a quite different sense of ‘complexity’, and even of ‘irreducible complexity’ compared to what Behe talks about.

And, I guess, for the discussions here, is life itself such an emergent phenomenon? Is there something about life that in-principle cannot be reductively analysed in terms of physics, chemistry and biology. For what it’s worth, I’m not talking about a ‘spiritual’ dimension, but rather about a phenomenon that arises only beyond certain limits of complexity.

1 Like

And I repeat, “complex” in Complex Specified Information simply means that a configuration that good or better is very improbable. Thus if there are 500 loci, each with two alleles, one of which is the fitter one, and the alleles are equiprobable (gene frequency 0.50 for each allele), then a “complex” specified haploid genotype is one with all 500 loci having the more fit allele. Similarly, a sequence of 500 tosses of the coin with 500 Heads has “complex” specified information, when the scale of relevance is how many Heads there are. Is there some other way in which the word “complex” comes into arguments about ID?


Point of clarification: Are you asking what we think ID people mean when they say ‘complex’, are you asking what we think complex should be understood to mean more broadly, or possibly something else entirely?

I don’t think ID people mean any single thing when they say ‘complex’, instead they mean several different and generally contradictory things simultaneously, and are hoping no one notices.


I’m mainly interested in what ID people actually mean. So ID people are probably the one’s most suitable to answer my question. Not that I don’t appreciate the answers already given.

Well I ain’t a smart guy, but I am an ID-proponent, and had my Wheaties for breakfast. I’ll give it a shot since no one else has:

Evolutionist: But what exactly to you mean by “complexity”
Jeff: How ‘bout the kind of complexity that turns a microbe into a microbe-biologist?
Evolutionist: Well that’s not a rigorous enough definition. After all corn could be considered more complex than a micro-biologist!
Jeff: Oh…ok…. Well let’s go with corn then. The kind that turns a microbe into corn.

Wow, a complete evasion of the question. What a pointless thing to have said.


Ah man, I worked so hard on that…


Is that your idea of being someone worth dialoging with?



I found Pross’ “dynamic kinetic stability” view of life an interesting way of viewing things.

Normally when we think of states of matter we can think of solids, liquids, gases. Pross visualises life as a “dynamic” state of matter as opposed to a “static” one, where “more fit life” is more “stable” in such a kinetic system than those that are “less fit”.

Pross then engages in a discussion of chemical stability and instability and introduces a concept he calls “dynamic kinetic stability,” which applies to replicative systems. Dynamic kinetic stability, or DKS, is central to Pross’s attempt to develop a chemical explanation of life. “In the context of chemical systems,” he writes, “static and dynamic forms of stability are very different. In the ‘regular’ chemical world a system is stable if it does not react. … In the world of replicating systems, however, a system is stable (in the sense of being persistent and maintaining a presence) if it does react—to make more of itself, and those replicating entities that are more reactive, in that they are better at making more of themselves, are more stable (in the sense of being persistent) than those that aren’t. This is almost a paradox—greater stability is associated with greater reactivity.”

Throughout the remainder of “What Is Life?” Pross makes the case that DKS, although entirely consistent with the second law of thermodynamics, is sort of the dodge that allowed living systems to arise from inanimate matter. Replicating entities are, he maintains, a different sort of matter governed by a different sort of chemistry from nonreplicating entities. In the long, penultimate chapter of the book, entitled “Biology Is Chemistry,” Pross insists that “Darwinian theory can be integrated into a more general chemical theory of matter, and that biology is just chemistry, or to be more precise, a sub-branch of chemistry—replicative chemistry.”

Pross explicitly equates evolutionary concepts with concepts drawn from systems chemistry. Natural selection, for example, is “kinetic selection”; fitness is “dynamic kinetic stability”; and maximizing fitness is “maximizing DKS.” I’m not sure I buy all of these connections. Pross doesn’t provide a chemical mechanism that underpins natural selection, for example. His argument that natural selection equals kinetic selection seems to be based entirely on the fact that the two forms of selection lead to the same result, but that doesn’t make them equivalent.

1 Like

It’s a tough crowd :roll_eyes:


If you observe the machine after being tested by the engineer himself:

  1. The gasoline engine can still operate normally
  2. The engine with gasoline + canola oil can operate normally

If the engineer’s modification is observed in the community:

  1. The gasoline engine can still operate normally, it is used by the public
  2. The engine with gasoline + canola oil can operate normally, public use, provided there is a refueling point for canola oil
  3. Engines with gasoline + canola oil, are used by the community, provided that there is a new service station in the event of damage caused by the effects of canola oil
  4. Other new effects that will arise due to the use of gasoline + canola oil that have not been previously known
  5. And others such as the price of canola oil

So it looks like engines with gasoline + canola oil are more complex than engines with gasoline alone

In the ID framework, complex simply means improbable.

The answer is yes, for the addition of this item makes the new engine configuration more unlikely than the initial configuration.

This is vague. More details if possible.

1 Like

It also would appear to give a number of false positives. I can think of a fair number of events that are improbable, that don’t seem to meet any meaningful definition of “complex”. Being hit by a stray bullet fired into the air immediately comes to mind.


Exactly why I wanted further details from him. There is nothing complex about lightning striking me now, but its very improbable.

He needs to narrow down exactly what he means.

1 Like