Here's a quote from Richard Dawkins:

"Biology is the study of complicated things that have the appearance of having been designed with a purpose."

My question is this. What types of phenomena can give the appearance of design?

I know of 3:

  1. Design itself (eg: human beings designing cars etc)
  2. Evolution (mutation and natural selection)
  3. Coincidence... no systematic reason for the appearance of design, just luck

Would these 3 encompass all possibilities?

  • 1
    Self-organization? There is a host of various mechanisms that "design" nice patterns, evolution is just one. And "give the appearance" to whom? If one is determined to find an appearance of design (or of the supernatural, or of destiny) they'll find it anywhere. Humans are psychologically predisposed to trace faces in the clouds, see Apophenia.
    – Conifold
    Commented Mar 22, 2021 at 6:06

2 Answers 2


What does it mean to give the "appearance of design"?

Bear in mind that some religions consider everything - clouds, earth, water, stars, etc. - to have been directly designed by some god or gods. This would indicate that any physical process can give the "appearance of design," depending on the mindset of the person looking.

So let's change the question a little. The relevance of evolution to design is that evolution is an optimization process. Evolution optimizes the reproductive fitness of individuals (or of genes). An intelligent designer is also executing some mental optimization process: he has some idea of what would be fit for his purpose, and tries to optimize what he's making so that it is more fit for his purpose.

So let's change the question to, "What are some optimization processes that can be used to create complex structures?"

A list of optimization algorithms can be found here. Many variations of evolutionary algorithms are in that list. It does no good to just repeat all these algorithms here, but I would like to call out stochastic gradient descent in particular, due to its importance.

Stochastic gradient descent is the main method used to train large artificial neural networks, which are behind most of the recent advances in machine learning. The behavior of these networks is not specifically designed by humans, and in fact is not very well understood by humans, but they solve tasks as if they had some skill.

Note that evolutionary algorithms can also be used to train large artificial neural networks; stochastic gradient descent and evolutionary algorithms are similar in this way.

See also the principle of least action and the free energy principle for examples of how minimization and maximization are inherent in physics. Soap bubbles are minimal surfaces. Pendulums come to rest in a minimum-energy state. Maximization and minimization principles are everywhere in nature.

  • Re your "An intelligent designer is also executing some mental optimization process". Are you suggesting there's no actual principle difference between evolutionary naturalism and creational theism/monoism? Since evolution is an optimization process like you said above, and optimization always has some particular goal, so can we always attribute such goal(s) to an intelligent metaphysical creator? Commented Mar 21, 2021 at 21:29
  • @DoubleKnot well, an intelligent being as a creator would be following some mental optimization process just as evolution is, so they have that in common. To say there is no difference is going too far. Perhaps evolution can in some sense be thought to be "intelligent," but not in the same way as the typical picture of a God, and not in the same way a human is. Also, biological evolution did not create the universe.
    – causative
    Commented Mar 21, 2021 at 21:52
  • I think it's a bit messy to describe theism as implying an optimization process. Typically, an optimization process implies iterative "steps" which move towards some local or global maxima. On the other hand, theists often frame creation as having no sub-optimal states. Note: here, "unfinished" and "suboptimal" are not the same thing. Many creation stories do have multiple steps going from "unfinished" to "finished." Baking a cookie to fulfill my goal to eat a cookie is not an optimization process.
    – Brian
    Commented Apr 15, 2021 at 19:49
  • @Brian certainly baking a cookie to fulfill your goal to eat a cookie is an optimization process. Consider the space of all possible actions you could take. Some of them help fulfill your goal to eat a cookie, most do not. Your brain has structures that guide you to choose actions within this space that optimize that goal. Moving on, an optimization process usually involves multiple steps - but not necessarily always. If we algebraically solve for the minimum of a quadratic function, this is an optimization process even though we only took one step, right to the best solution.
    – causative
    Commented Apr 15, 2021 at 22:37
  • @Brian consider if you wanted to make a robot that could bake a cookie. You could frame this as an optimization problem within reinforcement learning, seeking to find a neural network that maximizes a certain reward signal you define, rewarding it for successfully baked cookies.
    – causative
    Commented Apr 15, 2021 at 22:42

One point: I think Dawkins makes the point that pure coincidence actually cannot give the appearance of design (at least not on a frequent basis). In fact, he more or less defines the appearance of design in contrast with arrangements that come up in nature with little effort, by pure coincidence. For example, we don't see design in a bunch of tree branches or logs laying around in a forest in any random way, but we do see design in those logs being put together to form a cabin. The likelihood of a bunch of trees in a forest growing in such a way to look like a cabin is so minute that I would not call that kind of coincidence a source of appearance of design (at least not on a reliable basis).

As far as evolution goes, he actually talks of "sieves" as a more general concept. As an example, imagine there is a small hole on the ground. Given enough time and movement (rain, wind, etc.), what you will come to find in the hole are very small pebbles and leaves, etc. This isn't because anyone picked small things to place there, but because the only things that would have made it there had to be small enough. He argues that when that sieve has to do with conditions for surviving and thriving in an environment, we get natural selection.

The way I see it, you can classify the sieving as happening by human intervention or by purely natural processes, and while these categories sometimes overlap, they are exhaustive. In the first case we speak of design by humans, in the second, design by nature or by God.

  • Yes, so sieving is natural selection without mutation. My question is... is there some other natural process that can produce the appearance of design other than natural selection or natural selection+mutation? Commented Mar 21, 2021 at 18:03
  • The way I understand him, Dawkins is saying sieving can also happen by human means. For example, when humans design a faucet, they are sieving among all potential arrangements of pieces of metal. The sieving is happening in the space of all potential solutions, but this is still sieving. If you believe Dawkins' sieving idea, then the "binary" distinction is "with human intervention/without". You could make other categories too, like "with intervention by other mammals", etc. But this is still sieving.
    – Fox Mulder
    Commented Mar 21, 2021 at 18:13
  • 1
    Are you asking whether there are other processes that lead to appearance of design that cannot be understood as instances of sieving?
    – Fox Mulder
    Commented Mar 21, 2021 at 18:14
  • yes exactly. processes other than sieving. Commented Mar 21, 2021 at 18:16

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .