Key idea: A complex system does not need to appear all at once, because its parts can start out doing other jobs and later be combined, and the famous examples of “irreducible” systems turn out to have simpler working relatives.
The argument
The idea of irreducible complexity was popularised by the biochemist Michael Behe in his 1996 book Darwin’s Black Box.
The claim goes like this. Some biological systems are made of several parts that work together. Take away any one part, and the whole system stops working. Behe compared such systems to a mousetrap: without the spring, or the hammer, or the base, it catches no mice. If every part is needed, then a half-built version would be useless. And natural selection cannot keep something useless. So, the argument says, these systems could not have evolved step by step.
The eye is often given as an example, and it is answered on its own page about how the eye evolved. Behe’s own favourite examples were tiny ones inside cells: the bacterial flagellum, the blood-clotting cascade, and the immune system.
The argument has a hidden assumption. It assumes that the only way to build a system is to add its parts one at a time, each one doing the same job it does now. Biology does not have to work that way.
Parts can change jobs
Biologists call it co-option or exaptation. A part evolves because it is useful for one job. Later, it gets borrowed for a different job.
The bones in your middle ear are a clear case. Mammals have three tiny ear bones that pass sound to the inner ear. Two of them evolved from bones that formed part of the jaw joint in earlier animals. The fossil record shows the change in steps, with some early mammal relatives using the same bones for both chewing and hearing. Those bones were never “half an ear” waiting for a purpose. They were working jaw bones the whole time.
So the right question is not “what use is half a flagellum?” It is “could some of these parts have been useful for something else first?”
Scaffolding: when helpful parts become essential
Think about a stone arch. You cannot build one by placing stones in mid-air. Builders first put up a wooden frame, lay the stones on top of it, and then take the frame away. Now every stone depends on every other stone. Pull one out, and the arch falls down.
Someone who saw only the finished arch might say it could never have been built in steps. They would be wrong. It was built in steps, with help from a part that is no longer there.
Living systems can work the same way. A part might first be merely helpful. Later, other parts change to rely on it, and it becomes essential. Or an early part that helped get things going is lost once the rest can manage without it. Either way, the result can look “irreducible” today even though it was assembled gradually.
The bacterial flagellum
Many bacteria swim using a flagellum: a long whip-like tail driven by a rotating motor at its base. It really is impressive, and it is built from dozens of different proteins.
But it has relatives. Some disease-causing bacteria, including the ones that cause plague and some kinds of food poisoning, use a structure called the type III secretion system. It is a tiny needle that injects bacterial proteins into the cells of the host. It cannot swim. It does a different job, and it does it well.
Several of the core proteins in the base of the flagellum are closely related to proteins in this secretion system. In other words, a subset of the flagellum’s parts can work together, without the motor and the tail, to do something useful.
In 2006, the microbiologist Mark Pallen and the researcher Nicholas Matzke reviewed the evidence in the journal Nature Reviews Microbiology. They pointed out that many of the flagellum’s other proteins also have relatives elsewhere in bacteria that do other jobs. They argued that no great leaps are needed to build a model of how the flagellum evolved.
One honest caveat. Scientists still debate the order of events. The secretion system may have evolved from the flagellum, rather than the other way around, or both may come from a shared ancestor. That debate does not rescue the argument, though. The claim was that no subset of the parts could do anything useful. A working subset exists.
The blood-clotting cascade
When you cut yourself, a chain of proteins switches on one after another, ending with a sticky mesh called fibrin that plugs the wound. Behe argued that this chain, called the clotting cascade, needs all of its links.
The biochemist Russell Doolittle spent decades studying how this cascade evolved. The main tool is gene duplication. Sometimes a gene is copied by mistake, so a genome ends up with two copies. One copy keeps doing the old job. The spare copy is free to change, and occasionally it takes on a new role. Many clotting proteins look like modified copies of one another, which is what you would expect from this process.
The comparisons with other animals are telling:
- In 2003, Yong Jiang and Russell Doolittle searched the genome of the puffer fish. They found genes for 21 of the 26 clotting-related proteins they looked for. Genes for three factors that humans have (factor XI, factor XII, and prekallikrein) were missing. Yet puffer fish blood clots.
- The same study searched the sea squirt, a simple relative of animals with backbones. It had none of the clotting factors themselves, but it did have most of the smaller building blocks, called domains, that those factors are made from. That is what you would expect if the cascade was assembled by copying and reshuffling older parts.
- Dolphins and whales lack a working factor XII. Their blood still clots.
So the clotting system works with fewer parts in some animals. It is not irreducible in the way the argument claims.
The question goes to court: Kitzmiller v. Dover (2005)
In October 2004, the school board in Dover, Pennsylvania, voted to change its ninth-grade biology lessons. Students were to hear a statement presenting “intelligent design” as an alternative to evolution, and a book called Of Pandas and People was offered as a reference. Eleven parents sued the school district.
The case, Kitzmiller v. Dover Area School District, was heard by Judge John E. Jones III of the US District Court for the Middle District of Pennsylvania. The trial ran from September to November 2005. Michael Behe testified for the school board. The biologist Kenneth Miller testified for the parents.
On 20 December 2005, Judge Jones issued a 139-page ruling. Among his findings:
- The school board’s policy was unconstitutional, because it promoted a religious view in a public school science class.
- Intelligent design is not science. The judge gave three reasons: it relies on supernatural causes, its central argument of irreducible complexity rests on a false choice between design and evolution, and its criticisms of evolution have been answered by the scientific community.
- On the flagellum, the court noted the peer-reviewed work on the type III secretion system. On clotting, it noted that dolphins, whales, and puffer fish clot their blood with parts of the cascade missing.
- On the immune system, the court recorded that Behe was shown 58 peer-reviewed publications, nine books, and several textbook chapters on how the immune system evolved, and said this was still not enough evidence for him.
The judge also wrote that the court took no position on whether intelligent design arguments might be true. He added that intelligent design could continue to be studied, debated, and discussed. His ruling was only that it could not be taught as an alternative to evolution in a public school science class.
Two other facts are worth knowing. In November 2005, before the ruling, voters replaced all eight board members who were up for re-election, and the new board did not appeal. And because this was a district court ruling, it is not binding on courts elsewhere, even though it has been widely discussed.
Why “I can’t imagine how” is not evidence
The deepest problem with the argument is its logic. It starts from a gap: “nobody has shown how this evolved”. It then treats the gap as proof that it could not have evolved.
But gaps close. In 1996, Behe wrote that the immune system had no natural explanation. Within a few years, a stream of studies had traced how much of it arose. The flagellum and clotting stories have also filled in since the argument was first made.
Science can never prove that every possible system has a known history. What it can do is test specific claims. When the specific claims were tested, simpler working versions and borrowed parts turned up again and again.
Where to go next
For the classic example answered in full, read how the eye evolved. To see how new genes and gene copies arise, read mutation. To see how useful variations spread, read natural selection.