One thing that has been confusing me about Creationist arguments about mutation is that even though they agree that retrovirus insertion and replication errors can produce new sequences in the DNA, they do not produce new "information." Some of them, the more easily dismissed, argue that there are only four different "letters" in the DNA "alphabet," (there are actually 64 "letters," but that does not affect this argument or its refutation, so I'll address this point in a later paragraph), and the insertions do not add any new letters. That is like saying that all of our collected human wisdom can be found in the sentence "the quick brown fox jumps over the lazy dog," since it contains all 26 letters used to write in English. Instead, we use those 26 letters to form hundreds of thousands of words, many with unique meanings, and combine those words to convey ideas that even the combined resources of the Library of Congress and the British Museum Library can't contain more than a fraction of.
Others have an objection that is harder to understand, mainly because they do not explain what would constitute "new information." I believe that at last I have considered a way of looking at information -- I have no way of knowing if it is their way of looking at it -- that makes sense, at least as long as you are working with the DNA alphabet of four "letters." With organic molecules, especially proteins and nuclaeic acids, close is good enough. If a diabetic cannot get human insulin, pig insulin will do the job. Anabolic steroids work the way they do because they are similar to human testosterone (which is, itself, an anabolic steroid). There is an enzyme in chocolate (phenylethylamine) that minics endorphins released in our brains when we are in love. So even if a section of a gene, or the protein it generates is different, it is still just a variant which either does the same job (albeit it might do it slightly better, or slightly worse) and is therefore a neutral mutation, or does not do the job and is a detrimental mutation. The chances that it does a completely new job (and that is not even requiring that it does it well) are neglegible (and therefore easily dismissed with a handwave if you are a Creationist).
But as I said, that makes sense if you treat DNA sequences as words written in a four-letter alphabet. But DNA (using RNA as an intermediary) builds protein out of amino acids, and it takes three bases in DNA/RNA to select one amino acid. The DNA "alphabet" can be compared to the ASCII alphabet. Pure machine code is a sequence of "bits." Each bit can be either on or off. Machine code can be thought of as having a two letter "alphabet." But even in the earliest computers, the sequence was broken up into eight-bit segments called "bytes." A one-byte register could be in one of 2^8 = 256 different states. ASCII assigned the first 128of them to correspond with various characters, including the ten digits of the decimal system, the letters of the alphabet in both upper case and lower case, and various punctuation marks. Modern computers can work with longer registers than 8 bits, but the byte is so useful and so entrenched that the registers are always multiples of 8 and are partitioned into bytes.
So it takes a DNA "byte," called a "codon," of three bases to code for each amino acid. An insertion, or deletion which is not divisible by three does not just lengthen or shorten the gene, it changes it completely from that point to the end. In fact, it does worse than that. A gene has to begin with a "start" codon and end with a "stop" codon. But the bits of the "stop" sequence are no longer all in the same codon. The gene has been destroyed. Because there are four or five different codon sequences that can serve as "stops," it is likely that one will turn up in the DNA sequence, but the new resulting gene (and related protein) will look nothing like the original. It is completely new.
Also, in man-made computers every bit is always in the same position in the byte (We never begin a byte on the third bit and end on the second bit of the next register), but this is an artificial rule that does not apply to DNA. So if the old "stop" codon can be connected to a "start" codon (and there are four or five different codons that can act as "starts"), it, too can create a new gene. Altogether, a single mutation which upsets the synchronization of the codons could destroy up to three genes and create up to three brand new, unrelated genes. This kind of mutation is far from common, as most mutagens work on whole codons, but it only takes a few in a thousand generations to produce big changes.
Others have an objection that is harder to understand, mainly because they do not explain what would constitute "new information." I believe that at last I have considered a way of looking at information -- I have no way of knowing if it is their way of looking at it -- that makes sense, at least as long as you are working with the DNA alphabet of four "letters." With organic molecules, especially proteins and nuclaeic acids, close is good enough. If a diabetic cannot get human insulin, pig insulin will do the job. Anabolic steroids work the way they do because they are similar to human testosterone (which is, itself, an anabolic steroid). There is an enzyme in chocolate (phenylethylamine) that minics endorphins released in our brains when we are in love. So even if a section of a gene, or the protein it generates is different, it is still just a variant which either does the same job (albeit it might do it slightly better, or slightly worse) and is therefore a neutral mutation, or does not do the job and is a detrimental mutation. The chances that it does a completely new job (and that is not even requiring that it does it well) are neglegible (and therefore easily dismissed with a handwave if you are a Creationist).
But as I said, that makes sense if you treat DNA sequences as words written in a four-letter alphabet. But DNA (using RNA as an intermediary) builds protein out of amino acids, and it takes three bases in DNA/RNA to select one amino acid. The DNA "alphabet" can be compared to the ASCII alphabet. Pure machine code is a sequence of "bits." Each bit can be either on or off. Machine code can be thought of as having a two letter "alphabet." But even in the earliest computers, the sequence was broken up into eight-bit segments called "bytes." A one-byte register could be in one of 2^8 = 256 different states. ASCII assigned the first 128of them to correspond with various characters, including the ten digits of the decimal system, the letters of the alphabet in both upper case and lower case, and various punctuation marks. Modern computers can work with longer registers than 8 bits, but the byte is so useful and so entrenched that the registers are always multiples of 8 and are partitioned into bytes.
So it takes a DNA "byte," called a "codon," of three bases to code for each amino acid. An insertion, or deletion which is not divisible by three does not just lengthen or shorten the gene, it changes it completely from that point to the end. In fact, it does worse than that. A gene has to begin with a "start" codon and end with a "stop" codon. But the bits of the "stop" sequence are no longer all in the same codon. The gene has been destroyed. Because there are four or five different codon sequences that can serve as "stops," it is likely that one will turn up in the DNA sequence, but the new resulting gene (and related protein) will look nothing like the original. It is completely new.
Also, in man-made computers every bit is always in the same position in the byte (We never begin a byte on the third bit and end on the second bit of the next register), but this is an artificial rule that does not apply to DNA. So if the old "stop" codon can be connected to a "start" codon (and there are four or five different codons that can act as "starts"), it, too can create a new gene. Altogether, a single mutation which upsets the synchronization of the codons could destroy up to three genes and create up to three brand new, unrelated genes. This kind of mutation is far from common, as most mutagens work on whole codons, but it only takes a few in a thousand generations to produce big changes.