Applying evolution
The key principles of evolutionary approaches, as in biological natural selection, are variation, selection and 'inheritance' – maintaining desired features.
Chemistry
One application might be the design of an enzyme catalyst. Suppose a researcher has isolated an enzyme but wants it to behave slightly differently – perhaps work with a different substrate or at a different temperature. Engineering these changes into the enzyme would be very difficult, as the required structure would probably not be known.
An alternative approach is to make many slight variations of the enzyme, essentially at random, and to see which variants best meet the new criteria. These can be pulled out (selected) and used to create a new set of variants. Over several generations, the enzyme will be gradually refined so that it has the desired qualities.
A variation is used in drug development. Through the process of combinatorial chemistry – creating a 'soup' of chemical compounds by randomly combining a mix of slightly different components – huge 'libraries' of compounds can be produced. These can then be screened to see which bind to drug targets.
Biology
A nice example is the phage display library. This is used to make antibodies very specific to a target molecule.
The phage, a virus of bacteria, carries genes for an antibody. When it infects a bacterium, the antibody protein is made and sits on the outside of the host (it is 'displayed'). The genetic code of the phage can easily be altered, so a huge range of phage can be produced, each making a slightly different antibody.
These can be screened to see which bind most tightly to the target molecule. Selected phage can be mutated again, and the selection cycle repeated, so binding strength increases with each generation.
Computing
A very similar principle can be used in computing. Code can be written to achieve some purpose, then random modifications introduced. If some changes make the algorithm work better, they are selected for. Again, multiple cycles of mutation and selection leads to optimum solutions.
These kinds of approaches are used in a wide variety of applications, in research but also to work out, for example, the best way of laying out a factory or to create complex timetables.
The key point of all these applications is that the end point, or solution, is unknown and would be difficult to predict on the basis of known principles. But the combination of variation and selection, in repeated cycles, can lead to the optimum solution.
This is exactly the same process as biological evolution. The eye, for example, is a solution to detecting the outside world – a huge competitive advantage. The eye probably began as a few light-sensitive cells. The endless cycles of natural selection then refined them into the range of eye structures seen today.
Solutions may look tailor made, just as the antibody produced by the phage display process is a perfect fit for its target molecule. But the antibody was never 'designed' to fit its target.