V. Application to Biological Life
Now that we have a method to identify examples of intelligent design, we are now equipped to apply it to biological life to see what the likelihood is that it was designed.
We shall first discuss the relationship between natural selection and functional
information. We will then derive a generous estimate of I.nat. for an evolutionary search.
We shall then estimate I(E.x.) for several test cases and, applying the method suggested in the previous section, see if it is likely that biological life is an example of intelligent design.
Natural selection, fitness functions, and functional information
It is usually assumed that the origin and diversification of life is not a blind search.
Actual mutations, insertions, deletions, and genetic drift may be chance events, but natural selection essentially guides the search and, hence, the search is not blind. On the one hand, it is assumed that natural selection explains how life could appear and diversify without requiring any intelligence, but on the other hand, terms that that are usually applied to intelligence, such as 'design' and 'selecting' are commonly applied to natural selection. It is very common to read articles where the author marvels at what natural selection is capable of. Of course, this raises the question, does natural selection, itself, require intelligent design? The fatal mistake made by many who appeal to natural selection is the assumption that natural selection, itself, does not require intelligent design. It is bad science that does not test its assumptions, so we must apply intelligent design detection to natural selection itself.
Although natural selection is credited with somehow discovering the right combination of nucleotides to code for, say, proteins like SecY or RecA, there is a great deal of vagueness about how it actually is supposed to do this, and not just for two proteins, but for thousands. Not only must it somehow locate the proper sequences that are determined a priori by physics that will produce a stable 3-D structure, but it must also be able to assemble the information that will produce more impressive things like molecular machines, molecular computers and, ultimately, the cell and the organism itself.
Fortunately, the field of genetic algorithms or evolutionary algorithms can be used to introduce rigor to the concept of natural selection.
Every evolutionary search process, whether we are discussing natural selection, or a computational evolutionary algorithm, requires a fitness function. Without a fitness function, the search degenerates into a blind search, at best. The fitness function can be defined as follows:
Fitness function (evaluation function): represents "the requirements to adapt to. It forms the basis for selection, and thereby it facilitates improvements. More accurately, it defines what improvement means. From the problem-solving perspective, it represents the task to solve in the evolutionary context." (3)
Since the fitness function, whether it is found in nature, or in a genetic algorithm, must contain the requirements to adapt to, or that defines the desired outcome, it must contain at least as much functional information as the desired outcome. If the functional information contained in the fitness function is less than the functional information required for the desired outcome, then the deficit must be made up for in a blind search, which falls prey to the probability problems that emerge.
Natural selection requires a fitness function. If a given protein is a product of natural selection operating within a fitness landscape, then sufficient functional information required to find that protein in an evolutionary search must be encoded within the fitness function. If a few hundred, or several thousand proteins are required, then a great deal more functional information must be encoded within a much more complex fitness function. If molecular machines are also desired, then additional functional information must be included within the fitness function.
No one actually knows where this amazing fitness function is in nature such that we can measure the amount of functional information that it contains. However, if we assume that natural selection is responsible for, say, the origin of gene coding proteins, then we can estimate the amount of functional information the fitness function of nature contains by measuring how much functional information a given protein requires. We are then in a position to see if intelligent design is required for natural selection to produce the given protein by comparing the degree of functional information that must be encoded within the fitness function of nature and comparing it with I.nat.. Intelligent design detection methods must be applied to natural selection to see if intelligent design is required to encode the appropriate amount of functional information into the fitness function of natural selection. There is no escape; the functional information within a fitness function must be measured and evaluated, and a test performed to see if it requires intelligent design.
For example, if an evolutionary algorithm is attempting to produce even more complex software commands, then the fitness landscape includes the operating system within which those commands will survive or fail and the functional information required to produce that operating system must be measured. If the fitness function is outside the actual algorithm, say within a database, then the functional information contained in the database must also be included. To summarize; if natural selection or a fitness function are credited with producing a given amount of functional information, then if that functional information exceeds I.nat., by the method proposed in this article, ID is required to properly configure the fitness function.
It is estimated that there may be somewhere between 500 and 900 different protein folds,[4,5] that form roughly 4,000 to 7,000 different protein families.The stable folds are determined by physics, not biology. This requires that any evolutionary process must perform a search of sequence space to locate those areas where physics produces a stable, 3-D structure. Origin of life theorists are not decided as to what processes could lead to the minimal genome. Regardless of whether one prefers a genetic approach or a metabolic approach, we do know that at some point, proteins must be produced, or at least the information coding for stable, folded proteins must be achieved. We can, therefore, take all origin of life scenarios and put them into a 'black box' which performs an evolutionary search and outputs the stable folded proteins that are permitted by physics. It is not necessary to know what the processes within this black box do, all we need to know is the output. The output can be evaluated two ways, one way is to assume that the black box is performing a blind search which, of course, requires no intelligent design, and the other way is to assume that some sort of fitness function is operating within the black box which may or may not require intelligent design, depending upon how much functional information is required for the output. To estimate I.nat. for a prebiotic, origin of life search, we must estimate the number of trials available for a blind search. We will then be in a position to estimate I.nat. and compare it with the functional information required to produced a minimal genome to see if a fitness function would be necessary that would require intelligent design. Since we do not know what processes could perform the search, let us be extremely generous.
Taylor et al. have estimated that the mass of the earth would equal about 10.47. proteins, of 100 amino acids each. If we suppose that the entire set of 10.47. proteins reorganized once per year over a 500 million year interval (about the estimated time period for pre-biotic evolution), then that search permits about 10.55. options to be tried. Using Eqn. (3), I.nat. ≈185 bits of functional information. Of course, this scenario is much more generous than any scenario under consideration, but at least we will not be underestimating I.nat.. If I(E.x.) requires more than I.nat., then we can assume that either a fitness function requiring intelligent design must be included in the black box, or intelligent design is operating in some other fashion to properly encode the functional information.
We are now ready to examine four test cases.
Case One: the Venter Institute's synthetic genome for M. genitalium:
The five 'watermarks' in the synthetic Venter genome are formed by choosing base pairs that, when translated into amino acids and using the amino acid single letter symbols, spell out the following five words:
Hazen et al. point out that the number of functional options can vary according to the degree of efficiency required by the system. This is true for both human languages and biopolymer sequences. In this case, however, we will assume that the Venter Institute wants their watermarks correctly spelled according to the above sequences. Given that there are 20 options for each site in each word, using Eqn. (1), I(Ex) = 259 bits of functional information. Since I.nat. has been estimated at 185 bits of functional information, I(E.x.) > I.nat.. These results indicate that it is about 10.22. times more probable that the watermarks required ID than that they could be produced by mindless natural processes. Therefore, by the method proposed here, we can conclude that the 'watermarks' are likely produced by ID, in this case, the Venter Institute.
Case Two: a folded, functional protein domain:
Axe has estimated that the frequency of occurrence of stable, folded functional protein domains, a structurally independent component of a protein, is somewhere between 10.-64. to 10.-77.. These values correspond to M(E.x.)/N in Eqn. (1). The functional information required, therefore, to code for a stable, folded protein domain is 213 to 256 bits. Since we have estimated I.nat. at a generous 185 bits, which is much too low to achieve the amount of functional information required to produce a folded, functional protein domain, I(E.x.) > I.nat. and it is at least 10.19. times more probable that ID can produce a folded functional domain than mindless natural processes. The method of ID detection proposed in this article, therefore, reveals that ID is highly likely to be required to produce folded, functional protein domains. If the sequences coding for a stable fold are the product of a pre-biotic black box that contains a fitness function, then the fitness function will require intelligent design.
Case Three: an average 300 amino acid protein:
The functional information required to produce an average, 300-amino acid protein, can be estimated by analyzing the set of aligned sequences for SecY and RecA. These two proteins are particularly interesting because they are also universal proteins, found throughout organic life. It is inferred, therefore, that they would be required in a minimal genome. Analyzing a set of 1,553 aligned sequences for RecA and 469 aligned sequences for SecY reveals that 832 bits of functional information are required for RecA and 688 bits for SecY. It is reasonable, therefore, to estimate the functional information required for the average 300 amino acid protein to be around 700 bits of information. I(E.x.) > I.nat. and ID is 10.155. times more probable than mindless natural processes to produce the average protein. Again, if natural selection is invoked to explain the origin of proteins, a fitness function will be necessary that requires intelligent design.
Case Four: the simplest life form:
It is estimated that the simplest life form would require at least 382 protein-coding genes. Using our estimate in Case Four of 700 bits of functional information required for the average protein, we obtain an estimate of about 267,000 bits for the simplest life form. Again, this is well above I.nat. and it is about 10.80,000. times more likely that ID could produce the minimal genome than mindless natural processes. Again, if one wishes to explain the origin of the simplest life form by natural selection, a fitness function will be required that is capable of generating 267,000 bits of functional information, well into the area that requires intelligent design.
Next: VI. Conclusion
Labels: intelligent design detection