So was listening the other day to a live stream on @dsterncardinale channel wherein he talked to Paul Price. The topic of discussion was Sanford’s concept of “genetic entropy”. The discussion about 1:20 hours long, but there was (IMO) one critical moment. After 38:00 minutes, Dan points out an important point that Sanford completely ignores, which is a non-constant rate of beneficial mutations.
To explain this: We can describe a genome with a particular sequence being situated in one place within sequence space, which is enormous. Even modest genome sizes. A genome with the size of N base pairs can be in 1 of 4^N possible configuration. Although, that is assuming a constant genome size, which is also able to change thereby making the size of sequence space a variable as well. So a mutation can be taken as a ‘step’ within the sequence space, causing the genome to go from one place to another place. Furthermore, each position within this sequence space has a fitness associated with it, which is also not constant (it depends on the context).
But let’s simplify for the sake of the argument. Let’s say that the specific region within the sequence space where fitness is optimal is fixed. In other words, the shape of the fitness landscape doesn’t change. The subset of mutations that causes genomes to “walk” across the sequence space while remaining within this optimal region are neutral. Mutations that causes the genome to step outside and move further away from the optimal region are deleterious. Note that genome which is within the optimal region can only experience neutral and deleterious mutations. However, once a genome goes outside the optimal region, it leads to two things: (1) The number of possible deleterious mutations (and thus the rate of deleterious mutations) goes down. You can’t take the same step twice. More significantly, (2) beneficial mutations are now possible, such as a mutation that reverses the initial deleterious mutation. The genome moves back to the original spot in the sequence space. Alternatively (and more likely), there can be one of many compensatory beneficial mutations. The genome steps back into the optimal region, but not necessarily to the exact original starting point.
The key thing to bear in mind is that the number of potential beneficial mutations, and thus the rate at which they occur, increases immensely for every deleterious mutation. In other words, for every step away from the optimal region in sequence space, there are an increasingly greater number of possible steps to move back towards the optimal region; and the further away you are from the optimum, the greater the effects are of beneficial mutations. Eventually, the rate and effects of beneficial mutations reaches a point where it balances the rate and effects of deleterious mutations. The rates between the two also don’t have to be equal since selection puts a hand on the scale. Thus, even if beneficial mutations are very rare relative to those that are deleterious, a population can still be at an equilibrium point: where deleterious mutations fixed due to drift are continuously corrected and/or compensated for by beneficial mutations that are fixed due to positive selection and drift; such that fitness remains constant.
Sanford doesn’t take this into account. His model assumes that the rates of both beneficial and deleterious mutations remains more or less constant, such that deleterious mutations accumulate at a constant rate. However, when considering the fact that the rate and effects of beneficial mutations would increase (and the rate of deleterious ones decreases) correspondingly, the accumulation of deleterious mutations wouldn’t be linear. The line would curve downward and reach an asymptote at the equilibrium state. When this is pointed out, Sanford says that it doesn’t matter since he assumes that any population will go extinct long before it reaches this equilibrium state… but he doesn’t test this assumption. This also ignores the fact that this equilibrium state (called the mutation selection balance) is empirically observed in the lab… unlike genetic entropy.
Dan pointed this out, but in his own way (and more more concisely). Paul’s responded with the following at 42:10 minutes:
PAUL So, that’s one aspect of it, but the other aspect… and I think this is the more important aspect, is that you’re thinking in an overly reductionistic way about the genome itself. You’re thinking of the genome like it’s a big ocean of switches that can be flipped. In a sense that is true, but in another sense that is not true. Because what we’re really talking about again is information content. So, when you bring information into the picture, it becomes clear that if the surrounding context… the surrounding informational context has been lost… so in your example, you’re talking about reaching an equilibrium because you’re talking about a genome that is so saturated with deleterious mutations that now it is reaching a point where you’re probability distribution is starting to do like this [holds hands up at equal heights] and you’re starting to get this equilibrium that you’re talking about. By the time you get to that point, you’ve done an unbelievable amount of damage to information content of the genome. We are not talking about switches, we are talking about words and sentences in effect.
So, now genetic entropy is measured a as a ‘decline’ or ‘degradation’ of ‘information content’ in the genome. At this point I was just screaming in the chat… Please define INFORMATION for me and in what units are you measuring this. This is the one of my sticking points I have with ID-creationist who use the word “information”. They never define it in a way that can be measured objectively. And no, they don’t use Shannon Information theory. Thank goodness Dan asked Paul for the definition at 47:24 minutes:
DAN: What are you… How are you defining information? Because we need a quantitative definition of information in order to make this work.
Dan did an excellent job of clarifying that he is asking for a definition that can be used to quantify whatever Paul is referring to. Paul’s immediate response to this question is astonishing:
PAUL: Yeah… so… I don’t think that is correct. Yeah, that’s not correct.
Not only does he not define what he means by “information”. He argues for why he doesn’t need to provide a definition. To me, that killed the conversation right there and then. If you can’t define the terms of your claim, then your claims hold (literally) no meaning. Such claims can neither be confirmed, nor disputed, nor even discussed in any rigorous way. An instance of not even wrong.
If find such an admission astonishing, since Paul is essentially stating: “I claim that information content in the genome is declining due to genetic entropy, but there is no way to measure it”
Again, it kills their own argument. Paul also mentions an article he wrote on the creation ministries website along with Robert Carter. Paul says that in this article they make the argument that they don’t need to define what they mean when the use the word “information”. I was curious so I started reading it. The article starts with dismissing examples of the evolution of new functions (such as yeast evolving a new ability to digest a type of sugar) because these don’t represent anything… quote… “genuinely new”… whatever that means. Next, the article states the following:
Information is impossible to quantify!
Skeptics often challenge creationists, “If information is decreasing, what is the rate of its decrease?” Another similar objection is, “Can you quantify the changes in the information content of the cell?” This line of questioning successfully cuts to the heart of the matter. They claim our inability to define information robustly means information does not exist.
I cannot speak for all the skeptics of course, but that shouldn’t be the claim. The claim is not “if you cannot define it, it doesn’t exist”, the claim should be "if you cannot define it in a quantifiable manner, you cannot determine objectively whether the thing in question is increasing, decreasing, or remains constant".
For example, if we observe two genomes of two generations, one person A claims the information content when down, person B claims information increases, and person C says it remained constant. How can we tell who is right? How can we test these hypotheses without defining “information” in a manner that is quantifiable at least in some way? We just can’t. There is no way to objectively resolve such disagreements. One might as well argue which color is “cooler”, blue or red?
Next, they discuss why we can’t use Shannon information as the definition. They bring up an example in languages, like how the German (Eichhörnchen) and English (Squirrel) words have the same ‘meaning’ in semantics… or as they put it… they “both ‘code for’ the same information content” (referring to the same animal)". But the information of these words would be different if we measured it in terms of Shannon Information.
However, there already are a few issues here. Tthese words aren’t necessarily referring to the same animals (or identical taxa). Germans use the word to refer to one specific genus Sciurus, whereas ‘squirrel’ in English is used to refer to members of the whole family Sciuridae. At least, that is the taxonomical usages. The family also includes “ground squirrels” which are often not referred to as ‘squirrels’ in common vernacular. Instead, they are more often called ‘chipmunks’, ‘prairie dogs’, or ‘marmots’. The taxonomic word for the Sciuridae family in German is Hörnchen, which literally translates to ‘little horn’, and this word is coincidentally also used to refer to various horn-shaped pastries in German. If you click on the previous link to the German wiki site and use a browser with a translator, sometimes the translator fails to recognize the context and “Hörnchen” is mistakenly translated into “croissant”.
It’s also fun to look a bit into the etymology. English ‘Squirrel’ and Latin “Sciurius” have the same origins in Greek ‘skiouros’ which literally translates from ‘skia-oura’ to ‘shadow-tail’, so basically “shadow-tailed”. The reason for why ancient Greek referred to (presumably) squirrels like this isn’t clear. It’s difficult to ask them to clarify, of course. One proposed explanation is that these people believed that the large bushy tails of squirrels were use to shade themselves. Whatever reason, when English speaking people say “Squirrel” they don’t think “shadow-tailed”. The original metaphorical meaning has been lost. It’s rather like how we have come to refer to computer mice as ‘mice’, but it is still easy to recognize the metaphor.
The point I am making is that words actually don’t possess “information content”, regardless what is meant by “information content”. Such a statement implies that words “contain” something that gives them meaning. No. Meaning is determined by how words are used, which changes drastically depending on the time, place, and context (as shown earlier). Meaning is determined by the individual, or more predominantly, at the sociological level. I can make up a new word on the spot, or use a words in a novel way. However, this matters little unless it catches on like a meme, or I have a friend group with our private slang. That’s how meaning is established in language. This makes “meaning” subjective, and un-quantifiable in an objective manner. So, it literally makes no sense to say that a word has an “amount” of meaning.
But how does this apply to genomes? It doesn’t. This is another common mistake that actually most people make, not just creationist. We often describe nucleotide and amino acid sequences as ‘words’ and that DNA sequences ‘code’ for proteins. However, these are metaphors (much like the aforementioned ‘computer mice’). A genome isn’t a “book”, or a “blueprint”, or a “recipe”. And this is where creationist are running their own arguments into the ground through equivocating the semantic meaning of words and sentences with sequences in genomes and proteins. That’s exactly what Price does when he said “We are not talking about switches, we are talking about words and sentences in effect.” It’s one big confusion resulting from a category error and/or mistaken the map for the place.
Moving on. In one sentence, Carter and Price admit that they cannot quantify their concept of “information”. Yet, in another sentence, they insist it can be quantified using simple examples.
So, on the one hand, the answer is no. When considering the decay of biological information over time, we cannot quantify the rate of decrease, because information, at its base, is an immaterial concept which does not lend itself to that kind of mathematical treatment. On the other hand, the answer is yes, we can sometimes quantify information when we have something simple to measure.
Alright. We may not get a method to measure, but we at least get an example of a measurement… right? No, sorry. We don’t. The examples they provide actually don’t provide any measurements. Here, they just ask you which of two things have more information, and they simply assert that one has more information. Why exactly? Because it’s just “clearly obvious”… that’s why. No explanation. No measurements. Just use your “intuition”.
Let’s illustrate that information can increase and decrease
Example 1:
A man in a coma, existing in a dreamless unconscious state, compared to a man who is conscious
During a 24-hour period, which of these two men will have had more information, or ideas, go through their minds? The answer is clearly the second man. The first man will not have had any information in his mind during that period of time.
Example 2:
A 30-page children’s book compared to a 1000-page encyclopedia
Which of these two books contains more information? Clearly the second.
These examples don’t illustrate anything. The “measurement” here is nothing but asking one to make a good guess. However, suppose someone gives the opposite answer than they give. Their intuition is wrong? How would they demonstrate that? How can you demonstrate who is right? Once again, still no answer.
Next up, they do something that is so completely backwards, it’s infuriating.
Information is carried in so many complex ways (syntax, grammar, contextual clues, etc.) that it staggers the mind to contemplate actually trying to quantify it in an objective way. Yet this is what the skeptic asks us to do. This is an attempt at obfuscation to avoid grappling with the obvious fact that life is built upon the foundation of information. In fact, life is information.
Yup, that’s right. They say that asking them to define their terms is “obfuscation”. No… That’s not obfuscation. It’s literally the opposite. Obfuscating is when one is making claims based on terms that remains ambiguous or undefined, obscuring what the content of claims even entails… which is exactly what Carter and Price are doing. ZERO self-awareness.
I find this very irritating since it feels like gaslighting. If you ask them to define “information”, they are basically saying that you’re being dishonest. They will act as if you already know full well what “information” means, that you’re just playing dumb, and that your question to define “information” is simply asked in bad faith.
Next, they respond to skeptics who doubt that DNA contains any information. Once again, I can’t speak for all “skeptics”, but until we know what they mean by “information” then we just can’t tell either way. If you ask whether DNA has semantic meaning like words and sentences, then I submit that the question commits a category error. However, the argument they make to claim that DNA contains “information” (in their undefined sense) is also fallacious in another way:
Some skeptics will resort to simply denying that the DNA truly carries any information, claiming this is just a creationist mental construct. The fact that DNA data storage technology is now being implemented on a massive scale is sufficient to prove that DNA stores data (information).4 In fact, information can be stored more densely in a drop of DNA-containing water than it can on any computer hard drive. To allow that humans may use DNA to store our own digital information, yet to disallow that our genomes contain ‘information’, would be a blatant instance of special pleading.
The link they provide goes to a Scientific American article mentioning the potential to use DNA as a medium of information storage. The “information” in this context is measured in bits, i.e. Shannon Information. So hod up… previously they state that Shannon Information is “not truly a measure of information”, but in this section they use Shannon Information to argue that DNA does have “Information content”, presumably in the sense of “Information” that they are arguing for. So which is it? Are they the same or are they different? Pick one. Stop the equivocation fallacy.
Next up is a section by which they attempt to explain what a “real, genuine” increase of information look like?
What would a real, genuine increase look like?
To get back to the skeptics’ main question: what would real increases in information look like? I submit that to answer this, just sit at a computer and watch yourself type out a paragraph in a word processor. Mutations are incremental; they are small changes that happen in a stepwise fashion as cells divide and generations multiply. The genetic code consists of letters (A,T,C,G), just like our own English language has an alphabet [NOTE: though the A, T, C and G in the DNA molecule aren’t actually letters. The letters are our way to abbreviate the sequences of bases. Don’t confuse the map for the place]. But here is the central problem—it takes hindsight to recognize whether function or meaning is really present. Watch this transformation:
At what point in that series did you understand the meaning? Perhaps you guessed it at step 4, but you would have been lucky, for you did not know if a word like *housing* or *household* was about to appear. It didn’t become totally clear until step 5, when a full word was spelled and the program ended. There’s no real way to say, before you’ve already reached step 5, that ‘genuine information’ is being added.
- H
- HO
- HOU
- HOUS
- HOUSE
This is completely asinine. At what point did I understand the meaning? My answer: At every point AND at no point, depending on the context. Each and every step can be (and are) used to convey meaning. “H” is the eighth letter of the Latin Alphabet, and can be used as the symbol for the element hydrogen. “HO” may refer to many things: A village in Denmark, Santa’s laugh, or a derogatory term for a sexually promiscuous woman. “HOU” can refer to a common Chinese surname. or the AITA airport code for the William P. Hobby Airport in Houston. Even with “HOUSE” it may refer to a structure for habitation, or it could refer to the medical drama series with Hugh Laurie playing the main character. The answer can be YES at every step if the context provides the meaning, or NO if there is no context provided.
So how can they possibly assert that “real, genuine information” has only appeared at step 5? They even point out that “household” was a possibility. So there is no way to infer what the intended word is, let alone the intended meaning of the word. Not even reaching step 5 would be enough to know. And that’s the key point. The only way to know is if Price or Carter tell us what they intend to write down. However, this would end up in a circular argument. Step 5 is where “genuine information” is being added, because they have preemptively defined this as the step where “genuine information” is being added.
How would they apply this thinking to a genome… or a smaller segment like a gene? Even if they did, how would this not be circular as well? Well, here they try to apply this:
Mutations suffer from this same problem. But there’s an even bigger problem: in order to achieve a meaningful word in a stepwise fashion (let alone sentences or paragraphs), it requires foresight. I have to already know I want to say “house” before I begin typing the word. But in Neo-Darwinism, that is disallowed. Mutations must be random and unguided. Due to the sheer number of possible nonsense words, you cannot expect to achieve any meaningful results from a series of random mutations. What if you were told that each letter in the above example were being added at random? Would you believe it? Probably not, for this is, statistically and by all appearances, an entirely non random set of letters. This illustrates yet another issue: any series of mutations that produced a meaningful and functional outcome would then be rightly suspected, due to the issue of foresight, of not being random. Any instance of such a series of mutations producing something that is both genetically coherent as well as functional in the context of already existing code, would count as evidence of design, and against the idea that mutations are random.
We end up with the crux of the problem. Their concept of “information” of a genome is in reference to some unspecified goal that requires some foresight. A specific spot or region within the vast sequence space that is presumably the intention of a designer, and the only way for that goal to be reached is by a designer with foresight. However, this relies on the assumption that the sequences we see, or the sequences of our ancestors (going back to creation), were the goal. This is committing the Texas Sharpshooter fallacy… the name is in reference to someone seeing random bullet holes and draws a bullseye after the fact. That’s what they were previously doing with the step 5 of the HOUSE example (they drew a bullseye around step 5). They are also drawing a figurative bullseye around the original genomic sequence(s) which they believe Adam and Eve would have possessed, and that any deviation from this genome is missing the “goal”, and thereby a decline of “information”.
This is my best attempt at inferring what they mean by “information”. It is basically the distance of a sequence relative to the original spot in sequence space that was intended by their creator. The closer the sequence of a genome is to that spot, the more information this genome has. The further you move away from it, the less. But then again, this relies on the presumption of drawing this bullseye after the fact. It’s useless in a scientific discussion.
