Tag Archives: Optimization

The Fallacy of “Sub-Optimal” Design

Ever hear people like Richard Dawkins rant about the so-called “sub-optimal designs” in nature that must obviously disprove the existence of any omniscient Supreme Designer? As a practicing professional engineer, I find it a little annoying. Let me explain why.

What exactly is an “optimal design”? When I worked for a steel joist manufacturer, our designs were typically all about minimizing weight. It was often a very tight-margin business, and if we could save another pound of steel, that was a good thing. But, least weight doesn’t always equal least cost to produce. Sometimes, it was worth it to consolidate a bunch of different optimized least-weight designs into a big run of identical pieces, even if it meant some of them were a little heavier then needed. Just think about how much faster you could work at producing something if your instructions said that the next 1000 pieces would be made exactly like the first one, instead of having to look at the directions before every piece to see what had changed. The efficiency of repetition in our shop sometimes made a design that was not optimized for weight actually the most optimal design for us regarding least total cost (i.e. we traded a small material cost increase for a large labor cost decrease).

In my current role as a structural engineer, I’m reviewing shop drawings right now on a colleague’s project where I designed the seismic bracing for him. He unfortunately had some severe architectural constraints on his project with regard to permissible beam depths and flange widths in the walls these braces were in. After spending a couple of weeks trying to work out a solution with more conventional means, I finally came across an example of a different configuration in one of my reference books that we were able to make work in our situation. Would I call it an optimal design? Not hardly, but I was thrilled just to find anything that would meet those kinds of high demand loads with the restrictions we had.

Why do I bring up these two examples? To illustrate a couple of general points regarding optimum designs.

  • Optimization is always with respect to specific parameters. If you’re paying by the ton of steel, the most optimum design may very well be the one that weighs the least. If you’re the contractor erecting the building, the most optimum design might be the one that can be erected the fastest, or with the fewest jobsite workers. If you’re the owner of the new building, the best design may be the one that balances material costs, construction costs, and lifecycle costs for an overall lowest cost of ownership. Parameters like weight, cost, speed, strength, resilience, flexibility, lifespan, redundancy, etc. are always optimized at the expense of others. It is meaningless to talk of an optimal design without specifying what parameter is optimized. By the same measure, it is also meaningless to speak of something being a sub-optimal design without knowing what the original designer was trying to optimize for. I can say a military tank design is suboptimal for speed, and that may be true, but that isn’t where the tank was designed to excel: that heavy 4″ thick armor that slows it down so much also responds to incoming fire far better than trying to drive a race car into battle! Just because you would optimize for a particular parameter, doesn’t mean the original designer (or anyone else) would.
  • Constraints limit what is possible with regard to optimization. Looking at the end product of our seismic bracing design might appear to the fabricator to be a little odd when building it, not knowing the limits we were having to work within. Even a peer reviewer, knowledgeable of engineering design, might wonder why we didn’t simply use a much bigger beam, as is typical for these types of braced frames. But, if they’re like me, they’ve learned to ask why a puzzling design was chosen before they start throwing stones at it. In engineering, we deal with design every day – creating our own designs, reviewing the designs of colleagues, even sometimes having to try to guess the original design intent behind 100+ year-old buildings being renovated. And though it can be tempting to immediately deride some design that isn’t how I would design it, I’ve found an attitude of humility very appropriate when looking at the designs of others. For sometimes, the designs I thought were poor were actually quite innovative solutions to constraints I wasn’t aware of. But then I tried to run an alternate design that should’ve been “better,” and I ran into the same constraints the original designer did, and found the original design to be the only viable option after seeing the complete picture.

This is just a couple of reasons I think the bad design argument fails. It essentially reduces to saying “Because I, a person of limited knowledge, can’t comprehend some particular design chosen by an allegedly all-knowing Designer, He must not exist.” What hubris! Moreover, it seems to go even further by thinking that these odd cases outweigh the abundance of cases of brilliant-appearing designs in nature, many of which have spawned a whole field called biomimetics.  This field of study, which attempts to improve existing human designs or innovate new ones based on designs seen in nature, would not exist except that so many natural objects solve design problems we struggle with in ingenious ways. Indeed, I would say we have sufficient positive examples of exceptional design in nature to warrant an humble, inquisitive stance toward the supposedly “sub-optimal” cases we don’t fully comprehend yet. But really, isn’t that the attitude good science is founded on anyway?


Photo credit: By Bundesarchiv, Bild 183-23805-1665 / CC-BY-SA 3.0, CC BY-SA 3.0 de, https://commons.wikimedia.org/w/index.php?curid=5349654

Deconstructing Dawkins 3 – Optimal Design Overview

Richard DawkinsRichard Dawkins has made much of the “appearance of design” in biology being a false positive, and the notion that living creatures actually exhibit bad design that negates the idea of an omniscient Creator. After all, why would God, if He existed, and if He was all-knowing, do things like wire the human eye “backwards”? This is, according to Dawkins, a sub-optimal design that any engineer would reject out of hand, or get fired if he submitted a design like this to his company. In fact, regarding the “backwards wiring” of the vertebrate eye, he admits that it doesn’t actually have much effect on vision, but “it is the principle of the thing that would offend any tidy-minded engineer!”[1]

Oh really? Since he decided to drag us engineers into this, I’d like to ask one question: what exactly do you mean when you talk about an optimal design? I can tell you most engineering designs end up being sub-optimal, regardless of how “tidy-minded” we may be. That’s because we routinely have to make trade-offs between competing goals. I have a book on wood-framed shearwalls that humorously highlights this issue with a side-by-side photo of an “engineer’s dream wall” and an “architect’s dream wall”. The engineer’s wall is very stout and very solid. The architect’s  preference (and most owner’s) is one completely filled with beautiful expansive windows. Which one is the “optimal” wall? Neither one of us is getting what we would call the optimum. Us engineers need some minimum amount of strength that the windows aren’t providing, and the architect needs some minimum amount of holes in our solid wall so the owner doesn’t feel like he’s living in a dungeon! Factor in things like cost and meeting building code constraints and “optimal” becomes a very subjective term with different meaning to different stakeholders. But this is the way most design goes. You can’t maximize one parameter without minimizing another, and at some point, you’ll have 2 (or more) parameters that conflict. Do you focus entirely on the first, or the 2nd? Do you balance them equally? Maybe a weighted average based on your best guess as to which one will govern more often? Unfortunately, no matter which route you choose, someone will come along later, with the benefit of hindsight, and ask why you didn’t do it some other way. But God, being omniscient, has perfect foresight, so that shouldn’t be an issue for Him, right? True, He won’t make a mistake in design due to lack of knowledge or not anticipating future conditions, but the aspect of competing design parameters still applies.

Versatility and specialization are two such competing parameters. Specialized designs seek to maximize a positive parameter like speed or strength, or to minimize some negative parameter like weight or waste, at the expense of other factors. This is evident in animals like peregrine falcons whose hollow bones minimize weight, while their aerodynamics maximize speed. Versatile designs, on the other hand, seek to balance the most parameters at one time to achieve adequate performance over a wide range of conditions. This allows the object to fulfill many roles, or to survive in a variety of unpredictable conditions and possibly even excel over more specialized objects if conditions are constantly changing. Humans, for example, are extremely versatile. We may not thrive as well on our own as more specialized animals in arctic or desert or tropical environments, but unlike most of them, the same human can generally still survive in all of them. And, besides this highly versatile body design, we have the brains to make tools, and shelters, and transportation to overcome our bodily limitations, such that we can even survive in places like outer space where no animals, however optimized, can survive.

So is God required to maximize all parameters that go into a design? No. Some may fall into the category of “square circles” where the parameters are simply mutually exclusive. Is He required to maximize the particular parameter we favor over another that He deems more important? No. As professional engineers, we can seek the input of peers if desired, but nothing says we have to take their advice. The engineer signing off on the design and taking full responsibility decides the direction of the design. Is so-called “bad design” evidence against God? No. It simply means we likely aren’t seeing the whole picture. My own peer reviews of other engineers’ designs have raised questions as to why they chose a particular route, but then they proved quite reasonable after getting those questions answered. It was typically my lack of knowledge of the background of that particular project, or my unfamiliarity with some certain condition they’d been burned by before that made me think they’d missed something “obvious” when they had actually thought through their design better than I might have if I’d been in their position.

Engineers must approach peer reviews with an attitude of humility, but even more so if the design being reviewed is God’s. If I can overlook the good reason a fellow human engineer made the design choices he did, then I should be all the more open to the possibility that I’ve missed something an omniscient Designer did consider. And this is where I would encourage people like Dawkins not to arrogantly assume that there is no good reason for something just because they can’t see it. Tune in next week as we focus on a couple of specific examples where the atheist claims of “sub-optimal” and “bad” designs in nature have actually turned out to be engineering masterpieces.


[1] Richard Dawkins, The Blind Watchmaker (London: Penguin Books), p94.