It is difficult to explain why more and more complex organisms have steadily appeared throughout evolution if every life form is supposed to be a result of accidental changes in the genetic material. The second law of thermodynamics demands that in any closed system entropy increases. This means that energy tends to go towards equilibrium, disintegrating into simpler forms, rather than integrating into more complex ones. In other words, a system inevitably moves towards the state of maximum randomness and disorganisation. Life, of course, is not a closed system, so an increase in complexity does not violate the second law. Nevertheless, it seems strange that at every level there is a tendency in evolution to produce something new and more complex, going persistently against that law - from relatively simple and crude forms to complex and refined ones. Polanyi and Prosch comment:
another unsolved problem arises from the continuous quantitative increase in DNA chains from those of bacteria to those of man - from about twenty million DNA alternatives to about twelve billion. DNA does not behave naturally. It moves from a lower energetic level to the higher, because it moves towards a higher complexity, which cannot be explained by DNA itself. There is no chemical model available to explain this enormous growth or the chemical explanation for this fundamental fact of the system, just as we have no chemical explanation for the historical origin of DNA or for its capacity to produce media that apparently anticipate the continued development of the embryo. (1975, p.167)
Materialists sometimes argue that all life could develop from a hypothetical first cell, as all new life develops from a single fertilised cell. However, a cell can develop into a complex organism only because all of the parts and instructions are in the original cell produced from conception. For large scale evolution, mutation must on average add information. It has been already demonstrated many times with detailed probabilistic analysis that this is extremely unlikely (most classic textbook cases of mutations cited in favour of neo-Darwinian evolution are, in fact, losses of information). So, it is incongruent to conclude that random mutations on their own can account for an increase of complexity.
Redundancy (two or more solutions for the same problem found in many species, such as the development of the vulva in the nematode) is a further challenge for the traditional view. Denton writes:
...the greater the degree of redundancy, the greater the need for simultaneous mutation to effect evolutionary change and the more difficult it is to believe that evolutionary change could have been engineered without intelligent direction. (1998, p.339)
Even if it is accepted that gradual incremental steps may in some cases accidentally lead to more complex structures, they could not do so in all. A comparison can be made with horse-drawn carts and motor cars. Carts and cars have some similarities (e.g. four wheels) and the same purpose, but cars did not gradually evolve from carts. Throughout centuries, carts had been steadily improved. However, in order to make a car, a leap that required the development and addition of several completely new components at the same time was needed. Even the simplest functional motor requires a few parts non-existent in the most advanced carts. And if just one of these components were missing, the motor would be nothing more but extra weight that the cart would be better off without. Similarly, the survival of a new species is dependent on all the necessary mechanisms (in at least a rudimentary form) being present to begin with. The problem is that obviously one gene mutation is not enough for more complex adaptation. But, if just two mutations are required at the same time to produce at least a slight advantage, the chance that this will happen accidentally decreases dramatically. One example from Dawkins' book The Blind Watchmaker may be a case in point (1986, p.97-99). Weakly electric fish use electric fields to navigate in muddy waters. However, this remarkable ability is of no use unless the body of the fish is absolutely rigid. To make up for this, the fish has developed one long fin, so that the rest of the body can remain still. Even with this fin, the movement of the fish is rather slow, but this is compensated for by its ability to detect electric fields in water. So, the navigation system is useless without the fin, and the fin is maladaptive without the navigation system. Their appearance had to be synchronous, but they require very different sets of genetic mutations (not to mention that these mechanisms must also be controlled by an appropriate nervous system and brain). Sometimes many simultaneous mutations are necessary, which makes chance, as their main cause, improbable. Considering that the vast majority of mutations are lethal anyway, it stretches belief that numerous beneficial mutations can occur at the same time accidentally.
This issue is even more striking in relation to macroevolution, the emergence of new (usually more complex) species from earlier ones, especially if it is taken into account that intermediary forms are often not found, and that no breeders have ever managed to produce one species from another. To quote Denton again:
There are innumerable examples of complex organs and adaptations which are not led up to by any known or even, in some cases, conceivable series of feasible intermediates. In the case, for example, of the flight feather of a bird, the amniotic egg, the bacterial flagellum, the avian lung, no convincing explanation of how they could have evolved gradually has ever been provided. (1998, p.275)
Let us take one of these examples. No transitional fossil structure between scale and feather is known. This is not surprising, considering that a half-feather is likely to be a disadvantage rather than an advantage. A feather has a quite complicated structure that is light, and yet wind-resistant. This is possible because of the complex system of barbs and barbules. Barbules on one side of the barb are rigged, and on the other have hooks. It is hard to imagine that chance mutations could produce this precise cross-linking of the barbules to make a connecting lattice. Even if the chance mutation of a ridge/hook occurs in two of the barbules, it also needs to be translated to the rest of the structure. Moreover, if the lattice structure was not lubricated, the sliding joint made by the hooked and ridged barbules would soon fray, which means that the wings would be useless. Many others adaptations are necessary to have birds that can fly (forward-facing elbow joints, navigating tail, strong wing muscles, hollow bones etc.). Even if all of them have developed gradually, each step had to be synchronised and the new must be so great an advantage that it compensates for the losses (of fully functional forelimbs or strong bones). Moreover, not only does each modification have to have sufficient survival value, but the related genes also have to be dominant in order to pass it on to successive generations. Of course, once this transformation has occurred, natural selection will select the better wings from the less workable wings. Evolution clearly operates in part by Darwinian natural selection, but this process simply selects those transformations that have already occurred by different mechanisms.
The problem of complexity for undirected evolution resides not only in the remarkable number of components that are sometimes necessary, but also in the fact that life forms are such highly integrated systems that their components cannot be changed independently. Any functional change would require specific compensatory changes in the interacting subsystems. For instance, a change in a protein structure would necessitate many complex simultaneous changes throughout the molecule to preserve any biological function. Denton concludes that ‘it is hard to envisage a reality less amenable to Darwinian change via a series of independent undirected mutations altering one component of the organism at a time' (ibid. p.342).
The above does not imply that complexity is irreducible, as proponents of Intelligent Design would like to present. For example, some components may have been adapted from existing structures that had a different faction, or more significantly, irreducibility may diminish on a molecular level. However, it still makes sense to challenge the claim that such complexities can be fully explained by the adaptive selection of purely accidental mutations.