Browse Search Feedback Other Links Home Home The Talk.Origins Archive: Exploring the Creation/Evolution Controversy

Index to Creationist Claims,  edited by Mark Isaak,    Copyright © 2004
Previous Claim: CF011.1   |   List of Claims   |   Next Claim: CF101

Claim CF011.2:

The No Free Lunch (NFL) theorems (Wolpert and Macready 1997) prove that evolutionary algorithms, when averaged across fitness functions, cannot outperform blind search. This means that an evolutionary algorithm can find a specified target only if complex specified information already resides in the fitness function. Evolutionary algorithms cannot account for the complex specified information we see in life; that information has to come from design.


Dembski, William A., 2002b. No Free Lunch, Lanham, MD: Rowman & Littlefield, pp. xii, 199-212.


  1. The NFL theorems do not apply to biological evolution. The NFL theorems apply only when the fitness function is independent of the algorithm, but in evolution, evolving populations affect the environment and each other and therefore the fitness functions.

    It should also be noted that the NFL theorems do not refer to finding a target. They can apply to problems such as finding which of several algorithms performs best; such application differs from Dembski's concept of a target.

    Dembski himself later wrote that the NFL theorems are not important to his point, which is about displacement and conservation of information (Dembski 2002a).

  2. The NFL theorems consider the average of all fitness functions. Finding an above-average fitness function is not complicated and is often trivial. If you want a solution that performs well according to some metric, then a fitness function that measures that metric will usually work better than blind search. For example, if you are interested in survival and reproduction in a certain environment, then survival and reproduction in that environment is a good choice for a fitness function.

  3. The ultimate test of a concept is whether it works. Evolutionary algorithms work. They find solutions to many problems that are intractable with other methods. If mathematics contradicts reliable observation, the math is misapplied, irrelevant, or wrong.

  4. Complex specified information does not signify design.

  5. No design theorist has ever shown that complex specified information exists in life.

  6. That evolution uses information from the environment (via the fitness function) is nothing new. The process is called adaptation. Darwin wrote something about the general subject (Darwin 1859). It does not imply design.


Perakh, Mark, 2003. The No Free Lunch theorems and their application to evolutionary algorithms.


  1. Darwin, C., 1872. The Origin of Species, 1st Edition. Senate, London.
  2. Dembski, William A., 2002a (Nov. 6). The ARN Design Forum: What genetic algorithms can do.;f=13;t=000428;p=1
  3. Wolpert, D. H. and W. G. Macready, 1997. No Free Lunch theorems for optimization. IEEE Transactions on Evolutionary Computation 1(1): 67-82.

Further Reading:

Wein, Richard, 2002. Not a free lunch but a box of chocolates: A critique of William Dembski's book No Free Lunch.

Wolpert, David, 2002. William Dembski's treatment of the No Free Lunch theorems is written in jello. Mathematical Reviews, Feb. 2003, review 2003b:00012. or
Previous Claim: CF011.1   |   List of Claims   |   Next Claim: CF101

created 2003-10-22, modified 2003-10-25