The conflict is in assumptions. The first difference between an ID analysis of the evidence and materialist scientists’ analysis is that the latter assumes that homologous proteins in different taxa evolved from a common ancestor through undirected processes. As a consequence, they compare the differences between different proteins or between the same protein in different taxa where the differences in sequences could be dramatic. They then assume that one protein could be gradually transformed into the other. We would not make that assumption since the two different proteins or the two versions of the same protein could represent separate isolated islands in sequence space. Instead, we focus on research which directly studies the limits of change or the actual rarity of functional sequences.
Another assumption of materialists is that observing any change generated by an evolutionary process, or at least an evolution-like process (e.g. abzyme research), justifies the claim that evolution could drive any change of any level of complexity. This logic closely matches that of creationists who study how the flood waters from the Mount St. Hellen explosion generated layering patterns in sedimentary deposits. They argue that the capacity of a violent flood to produce those geological patterns justifies the belief that a massive flood could produce all geological patterns. I mention this comparison not to demean either group but to point out similarities in thinking which could foster healthier dialogue between them.
The experiments I cited which study accumulating mutations resemble evolutionary narratives running in reverse. The structural and sequence differences between flagellar proteins and their hypothetical closest common ancestors are so great that the evolution of the former involves a far different protein sequence entering the vast sea of nonfunctional sequences and then finding some island of functional sequences corresponding to some flagellar protein. The first encounter would be with a barely functional protein surrounded by non-functional neighbors. Then, natural selection could assist in refining the protein, so a sequence would move toward a more optimized performance. This process is vastly more challenging than simply modifying the function of an already existing protein.
The cited experiments demonstrate that the negative impact of mutations and the percentage of harmful mutations increase with the number of accumulated mutations. Note that I am not focusing on what the authors imagine could be true but on what their hard data demonstrates to be true.
The authors discuss how compensatory mutations and buffering effects (e.g. chaperones) could increase the threshold before accumulating mutations destabilize the protein. For instance, the negative effect of one mutation could be undone to some extent by the following one, so the sequence might change significantly more than average with a less negative effect. However, such series of mutations maintaining significant fitness would represent a very narrow corridor in sequence space as judged by the dominance of harmful mutations over compensatory ones.
In general, compensatory mutations and buffering increase the threshold for stability by only a limited amount. After a certain number of mutations, the limit is reached, and the protein quickly loses function with most new mutations. After about 10 mutations in B-lactamase and HisA, the majority of the following mutations are lethal. After several more mutations, nearly every subsequent mutation is lethal. These results are from actual bacteria, so they take into account any buffering effects. The authors argue that the destabilizing trend is a general property of proteins.
As a consequence, the rarity of proteins in the region around any functional sequence is so great that no functional protein could ever be found. For instance, the B-lactamase studies indicate that after about 5 non-synonymous mutations, around 1 in 3 amino acids could be tolerated at each site, so the rarity is almost identical to Doug Axe’s results. The HisA results are even worse. After around 10% of the sequences change for either protein, all functionality, based on their numerical fitting, is permanently lost. These regions are completely devoid of functional sequences. As a consequence, most of sequence space is so sparsely populated with functional sequences that no search could ever find any protein in the entire history of the earth.