Just a note that I have put up a new post at Panda’s Thumb in response to a post by Eric Holloway at the Discovery Institute’s new blog Mind Matters. Holloway declares that critics have totally failed to refute William Dembski’s use of Complex Specified Information to diagnose Design. At PT, I argue in detail that this is an exactly backwards reading of the outcome of the argument.
Commenters can post there, or here — I will try to keep track of both.
There has been a discussion of Holloway’s argument by Holloway and others at Uncommon Descent as well (links in the PT post). gpuccio also comments there trying to get someone to call my attention to an argument about Complex Functional Information that gpuccio made in the discussion of that earlier. I will try to post a response on that here soon, separate from this thread.
Neil Rickert,
I think Eric would agree with this as he is trying to get his arms around functional information at this point.
I may be incorrigible. I’m always trying to make the math make sense in English. lol.
I’ve tried the Google language converters but I think some work still needs to be done there.
I guess that given that I have often argued that the genetic code is a code because it meets the mathematical definition of a code there must be such a thing as a mathematical definition. Hah! Caught myself.
Thank for your comments, esp the specific reference.
I hope that’s a sincere “lol.” In the association of plain language with mathematics, what the plain language “really means” is the math. It’s not the case that what the math “really means” is the plain language. We don’t adopt mathematical formalism just to be high and fancy: as a practical matter, we can’t say exactly what we mean and get our arguments right without it.
Scanning the comments at Panda’s Thumb, I didn’t find your response. Please give us a phrase to search for.
Click here for comment 564, which begins:
Howling lunacy. I’m assuming that we’re seeing cultural madness here — Holloway is a product of the Bible Institute of Los Angeles, now known as Biola University — and not individual pathology.
For those of you interested in algorithmic complexity, I’ll mention that Holloway has given an old definition of algorithmic mutual information. It’s possible to derive stronger results (tighter bounds on certain quantities) when we go instead with the definition
where
and 
are binary strings, and 
is a shortest (binary) program outputting 
[I’m assuming that you know the definitions of 
and 
]
More importantly, you can’t simply write
which indicates that 
and 
are binary strings, and then assert that 
and 
are the universe and mathematics, respectively. To say merely that it’s wrong is not to tell the story that needs to be told. It’s utterly bonkers.
It’s very common for Christians to see mathematics as coming directly from their God, and to see ID as the explanation of the effectiveness of mathematics.
So, yes, this is probably not individual pathology.
My responses to Holloway’s comments 539, 542, 561, and 562 will be found by going to PT, looking for the “Eric Holloway needs our help” thread, displaying it on its own page, setting the Disqus comments to Sort by Newest, and looking at the newest as of right now (11/13/18) (The PT Disqus comments don’t have individual links you can copy).
I didn’t try to respond to 564, as I was totally dumfounded.
He has mathematical statements that may work in terms of the variables he has in them, but he makes it totally unclear what they refer to in the biological models, so his statements are at a minimum unclear as justifications of Dembski’s arguments.
Isn’t it interesting that the PS thread, with all its polite, lengthy and math-filled comments, will not resolve for the same reason that our discussions haven’t? Kirk simply refuses to accept a definition of function that does not align with his views of how cells are supposed to behave, i.e. sustain the lofty goal of bringing Homo sapiens to walk on this earth. Of course, if we set “organismal well being” as our function, then any oncogenic mutation will be a loss-of-function. But why is that a “better” argument?
Some cancers are known to leave the organism and become contagious such as DFTD in Tasmanian Devils*. I could argue that the cells have acquired a novel function that allows them to prosper instead of die along with the organism, but of course they are still bad news to any Tasmanian devil that happens to get infected by them. Would you not agree that this is matter of mere perspective?
* Memo to self: ask Fifth on the other thread why this particular choice was made by the Selector?
ETA: corrections
This raises the philosophical issue. Namely, what does ‘properly’ mean in the above sentence? What is the difference between function and mis-function?
If I read you correctly, you think ‘properly’ means cancer cells should operate like non-cancerous cells. If they do not, then they are not functioning properly. And that may very well be what Durston thinks.
But it is not what Swamidass thinks. He thinks that “Common selective pressures [are the] best explanation for most of cancer FI*”. So that means the driver genes in cancer cells are functioning properly, because ‘proper’ function is that function selected by evolution which causes the cancer cells to outcompete non-cancerous cells.
That approach to understanding the proper function is called the teleosemantic approach to naturalizing biological function.
————————————————–
*I did notice that Swamidass used FI abbreviation, ie Functional Information, but I don’t think that concept is necessary for my above post. But it is worth pointing out that the Szostak definition of functional information requires one to pre-specify the function which raises the same issues of how to do so. In the FI case, he just regresses to a human making the decision, which is enough for his definition.
My blunt rewording ot Tom’s* comment: If you take that approach, you are admitting that none of your critiques of the science have any merit.
———————
* I really do have to decide between using first or last names!
English doesn’t always make sense.
His biological model appears to be a computer with a random number generator. The biology isn’t there.
Corneel,
BruceS,
I my world Kirk is the rational player here. He does not refuse to accept anything. He shows how P53 functional specificity is less in cancer per the Hazen and Szostak method. Secondly if a child gets cancer and dies that is clearly a loss of fitness. No matter how you try and spin this cancer is usually the result of a loss of functional information along with most other microevolutionary adaptions.
You are being very imprecise with your terminology here Bill.
First you speak of functional specificity, then of fitness, then finally about functional information. It seems to me all three of these concepts means different things, and can’t be directly interchanged. A loss or gain of one is not necessarily the same as a loss or gain of the other.
Rumraket,
Fair enough: You can replace functional specificity with functional information because that is what Kirk showed. As far as fitness goes I used that term as Corneel measures function by fitness although I agree with you they are not precisely the same.
You are comparing apples and oranges.
The biological function I am referring to is that of cancer cells within the environmental niche of the body they have originated in and in the population of cells within that body..
The biological function you are referring to relates to the genome/development environment (eg parents smoking?). It is probably difficult to assign a specific cause to the child getting cancer so it would be difficult to apply genetic analysis alone to fitness. Not a good candidate for assessing biological fitness of child’s genome. And as I said, that is a different issue from the one I was describing.
Whether Durston has the right idea depends in the end on the idea’s usefulness for scientific prediction, explanation, control. If we are talking the science of cancer treatment, it is not as useful as Swamidass’s approach.
BruceS,
Why do you think Kirk’s method is not as effective for cancer treatment?
FI = Functional Imprecision?
Are you really asking about the consensus on the law of the conservation of quantum information in biology?
I don’t think you realize what you are asking about…
If this law were ever applied to biology, it would not only kill Darwinism. It would kill all biology that is based on classical information…
The law goes beyond anything you can imagine about the information theory that can be applied to DNA including genes… There is enough evidence that the great majority of DNA sequence contains 1 or more copies of quantum information that people like Paul Davies call “shadow information”.
Classical information plays little role in how cells differentiate and organisms develop or even change within kinds…That’s why experimental evidence with mutagenesis has failed to produce anything new…
No evolutionary biologist wants to even think about what the consequences of this fact would be if experiments proving that were ever funded… They will be but who is going to publish them and then who is going to acknowledge them?
Why would evolutionists allow “to slit their throats”?
Because it ignores what makes cells cancerous in the sense that they will outcompete and replace non-cancerous cells.
ETA: As far as I can tell, his approach would treat anychange from non-cancerous cell in the same way.
BruceS,
It does not ignore it as far as I can tell. In the case study p53 contributes to apoptosis when DNA is damaged beyond repair. When apoptosis is disabled then the cell has a survival advantage over other cells however if you compute this it is a loss of functional information as P53 no longer can perform the apoptosis function due to loss of binding due to mutation. In the cancer case little sequence specificity is required where in the non cancer case high sequence specificity is required indicating a high level of FI.
Loss of functional information caused cancer in this test case.
The problem is this: why would Durston’s idea lead alone lead a scientist to focus on those driver genes?
However, I think there is an issue I glossed over.
As I said, Functional Information according to Szostak depends on which function scientist choose to consider. In this case of cancer treatment, the scientist should focus on the functions which enable a cancerous cell to outcompete.
Biological information, on the other hand, depends on the interests of the organism by causing it to be evolutionary successful.
In this case, the two happen to coincide.
I don’t think I have anything to add, so I will leave the last word to you.
BruceS,
Thanks for the discussion 🙂
And what if a 90 year old develops cancer? Unlikely to affect reproductive success any more, don’t you agree?
But I suppose I musn’t grumble: you have embraced fitness at the organism level as a decent yardstick of biological function, and this I applaud. Even though, as BruceS rightly pointed out, that may not be the best choice in cancer research.
By definition, an adaptation increases fitness. Oopsie, you seem to have switched back to molecular function again. See how you switch back and forth? Please try to stick to a single concept. Using fitness as your biological function in FI is perfectly fine, but you need to be consistent about it.
Corneel,
Thanks for pointing this out. It appears that gain in FI and fitness are probably not correlated at least from the standpoint of random genetic change.
That depends. For example, if you choose molecular function to be the function in your FI, then you are correct. But then you’ll have to accept that FI is increased in certain types of cancer.
OTOH, if you choose fitness to be the function in your functional information, then a correlation is guaranteed! But then you’ll have to concede that natural selection can increase FI in a population .
Did I just have a deja-vu? I guess we have discussed this before, so I’ll leave it at that.
Is there a gain in FI every time someone goes down the rabbit hole? There for sure seems to be some sort of feedback mechanism operating.
Corneel,
How so?
A discussion of this very topic can be had at PS but:
https://www.nature.com/articles/nrc2013
Back to the Future: 2030
No Free Brunch Theorem
Nobody cares if the theorem is true of false.. If it is true/false, who cares? Everyone with his or her beliefs can and has to be accommodated…
BruceS,
Yes, I agree, Holloway is not understanding the interpretation correctly. Expect Information might not increase, but that doesn’t matter in biology. It’s not the average member of a population that encounters new/better/worse function, it’s going to be those on the fringes that might increase information. I want to call this something like “selected information”, or maybe “differentiating information”. Biologists probably already has a term for it.
I wonder if there is any scientific value in the various concepts of information.
I know there is lots of philosophy that attempts to re-conceptualize existing science using information*. And papers which are published in science journals which re-present existing science using information terms.
And of course ID arguments that show science cannot do what it claims to do regardless of the actual content of the science!
But are there any research programs that provide new models or theories or results that require some use of one of the approaches to formalizing information.
ETA: last sentence change.
————————————
* For comparison, I would say the discussions of information in statistical mechanics are mostly if not all philosophy of science, not science per se.
It doesn’t? How do you know that?
I’m sure you, and the experts here, are aware of the II law of thermodynamics…that states that in closed systems the information can’t increase spontaneously… If this is true, how can a lifesystem even develop, such as an embryo?
So, where does this information come from?
I thought adaptation was the consequence of being fit. Perhaps you guys should abandon English and stick to math.
😉
I wonder if there is any scientific value in the various concepts of entropy.
I would say there is value in it’s use as a heuristic.
OMagain,
As Kirk showed evolution of cancer cells can be the result of loss of information as in the P53 example. Where is the evidence for a gain of functional information in a cancer type according to Szostak and Hazen’s definitions? All cancers I am aware of require the loss of the apoptosis function. Also most cancers require the loss of the protein degradation function.
I do not think that Kirk “showed” any such thing. He produced a definition of ‘information’ under which cancer-causing mutations in p53 represent a reduction in ‘information’. Woop-de-doop.
Likewise, whether any given oncogenic mutation represents an increase or decrease in Hazen’s functional information would depend on how you, the investigator, have decided to define “function”.
Define “function” as “host fitness” and any oncogenic mutations decrease FI; OTOH define “function” as “cellular replication ability” and any oncogenic mutations increase FI. There is, pretty much, an infinite number of ways to define function. Hence the commenters hereabouts who have been wondering aloud about the utility of information theory in biology.
Also, what’s with “the” apoptosis function and “the” protein degradation function?
Vague, much?
Finally, “require” is the wrong verb: “are associated with” is more accurate.
DNA_Jock,
In the case of p53 it depended on how many different combinations of AA’s would perform the function. The fewer the higher FI no matter how you defined it. The no cancer function required higher specificity for this protein then the cancer function, requiring higher FI.
If I created the hypothesis that cancer was partially caused by a loss of functional information, how would you shoot it down?
Well, first you would have to support your hypothesis…
The first problem you have is with your terminology “caused by a loss of FI”. As I already explained to you, mutations that cause cancer may represent a loss of FI, for certain definitions of FI. That does not mean that the cancer is caused by the loss of FI. Imagine that I create a metric “hoobelliflip”, and I show you an example where a p53 mutation that causes cancer has greater hoobelliflip than wild-type p53. I cannot go around claiming that the cancer is caused by the increase in hoobelliflip. Correlation =/= causation.
You might think so, but you would be WRONG. As I have already explained, I can arbitrarily invert my metric of FI by changing the definition of function. Hazen’s FI is [minus log2] of the proportion of sequences that perform function X this well or better. So I can define my function X as “promotes cancer”, or “increases cellular replication ability” and all the oncogenic mutations have INCREASED FI.
It is a little depressing that you replied to my comment without understanding this.
I’ll just add here that the reason that the specific example of p53 has some biological plausability going for it, is because it is a tumor suppressor. I bet that as soon as we draw your attention to oncogenes that have acquired novel or increased function, such as mutated SRC (which is constitutively or over-expressed in a fair number of tumors), you’ll just change your story again.
Can’t you see that you are constantly updating your definition?
Yeah, Corneel,
I have not done a census recently, but I get the distinct impression that gain-of-function mutations such as SRC over-expression or one of the many other Tyrosine kinase up-regulations (ALK-P, BCR-ABL, etc, etc) are the first step in the majority of cancers. Retinoblastoma being the most famous counter-example…
DNA_Jock,
Do you discount that “increasing cellular replication ability” can be due to a loss of functional information?
I’d say the reverse is true: A high fitness is the consequence of being well adapted, but I see what you mean: there may be several definitions out there.
Won’t help. Neither English nor math is my mother tongue.
And there is another deja-vu. I am pretty sure that I once mentioned deleterious gain-of-function mutations to Bill. I guess the idea doesn’t combine well with the concept of designed organisms. If some protein is doing stuff it isn’t supposed to do, then it’s broken. No function.
I will spell this out one last time.
Consider 2 cancer patients.
Jimmy suffers a deletion that renders one of their cells Rb -/-. Molecular biologists would refer to this as a loss-of-function mutation that causes cancer.
Rupert suffers a rearrangement that pairs BCR with ABL. Molecular biologists would refer to this as a gain-of-function mutation that causes cancer.
Using the definition of function A “keeps host cancer-free”, both of these mutations can be described as decreases in FI.
Using the definition of function B “promotes cellular replication”, both of these mutations can be described as INcreases in FI.
If an informational biologist, or IDist, were to say the cancer is the result of (or the cancer is due to) the decrease in FI, then they are talking rubbish; equally rubbish would be to say that the cancer is the result of the INcrease in FI.
OTOH the IDist could say that the cancer is associated with a decrease in FI, but they could EQUALLY say that the cancer is associated with a INcrease in FI, depending on which definition of “function” that they have arbitrarily chosen, A or B.
Hence the commenters hereabouts who have been wondering aloud about the utility of information theory in biology.
Very well put.
Corneel,
Sure. Gain of function loss of functional information just like cancer.
DNA_Jock,
Unless sequence specificity is decreasing. You are not increasing FI unless sequence specificity is increasing. Continued function after a random mutation indicates a lack of sequence specificity. You always make interesting points even when I don’t agree with them they stimulate thought.
A eukaryotic cells normal function is the ability to change states depending on where it is in its life cycle. It must behave differently from conception to birth then it does in mature animals. Cancer is often caused by the cell reverting to its embryonic state in a mature animal. The embryonic pathway WNT/Beta catenin is one I have studied that is active in many cancers. It’s embryonic activity is involved in heart development. Here is a short video explaining its control mechanism.https://youtu.beNGVP4J9jpgs
colewd,
Bill,
Can you please double check the link to the video?
Thanks
“Sequence specificity” is increasing. It has to be. Thus FI is increasing.
Now granted, the increase may not be terribly impressive, but it is there. BY DEFINITION.
Gad, I’ll put numbers on it for you.
Imagine a sequence space.
One million sequences do fuck all (Subset A)
One hundred sequences inhibit cell growth, but poorly (Subset B).
One hundred sequences inhibit cell growth, rather well (Subset C).
A mutation moves us from the least active in subset C to the least active in subset B.
Kirk has defined function as “prevents cancer”.
He notes that the FI has decreased from -log2(100/1,000,200) to -log2(200/1,000,200), i.e. from 13.3 to 12.3
Alice has defined function as “promotes cell growth”.
She notes that the FI has INCREASED from -log2(1,000,101/1,000,200) to -log2(1,000,001/1,000,200), i.e. from 0.000144 to 0.000289. Hey, it’s doubled!
FI is merely log base 2 of the percent rank. If you invert the ranking order (and you always can, by changing how you define “function”) then increases in FI become decreases, and vice versa, by definition.
This is all High School math.