Doesn’t matter with respect to FI. A function can improve (such as activity or specificity, or in this case fitness) without any “new” function being acquired, and yet FI can increase.
Also, technically fitness-improving mutations can of course also contribute new individual functions that increase the fitness-function.
A few responses to various people. I’ll get to the issue of the word “complex” in a separate comment.
Functional information as discussed by Szostak et al. is not some fantasy of creationists or ID advocates. Szostak et al. were attempting to define adaptive biological information. Their measure is a possible one to use, though we don’t usually have enough information to measure it. ID advocates and/or creationists have taken it up under the mistaken idea that some level of it is unachievable by natural selection. They have no evidence for using it as a reliable indicator of Design Intervention.
The point of the “coins” example is not tp persuade you, but to demonstrate just how silly the challenge is. I’m not going to change your mind, but maybe you will come to understand the misconceptions embedded in that challenge.
It matters with respect to ID theory, which state that high FI cannot be produced by blind natural processes from an unrelated state, not that blind natural processes such as RV+NS cannot increase FI from a related state.
I don’t see what that has to do with design detection and the 500 bits rule.
I know ID proponents like to assert there is always going to be a net loss of functions so evolution could never be expected to gain them (“devolution”), but as previously explained that is wrong too.
It’s a red herring here though.
Such arguments are generally a variation on the “2LoT” argument, which roughly states that the Second Law of Thermodynamics does not allow natural processes cannot create order. This is simply false, as natural processes create order all the time. Briefly, “order” is created by expending energy to expel disorder.
I see you changed your post and added this new qualifier after I responded to it. Suddenly this new and never before seen qualification has been added(I could not find the idea in any of Gpuccio’s posts on the 500 bits stuff that it has to be “from an unrelated state”). Moving the goalposts in an ad-hoc fashion are we?
Edit: Here’s Gpuccio saying he would infer design for a sufficient amount of changes in a single protein:
Okay, then explain the difference between a related and an unrelated state? By chance, any two randomly generated DNA sequences are expected to have 25% sequence similarity on average(5% for protein sequences). Is an unrelated state supposed to be one that has zero sequence similarity (to some arbitrarily decided distant ancestor)? Does it become a “related state” as soon as the similarity between two sequences rises above 0?
A question: how does one tell the difference between official “ID theory” and the assertion of some individual ID advocate?
And getting back to one of my OP questions:
Where has this tenet of ID theory been validated?
Has any ID advocate actually calculated the Functional Information (according to Szostak’s definition, NOT Gpuccio’s completely-fake faux-FI) for any organism? If not, then this claim would seem to be nothing but unsubstantiated speculation.
For that matter has any step in Gpuccio’s argument been validated?
It would seem more accurate to state:
It matters with respect to ID theory, which asserts a whole bunch of unsubstantiated nonsense, which nobody outside the ID echo chamber gives the proverbial pair of fetid dingo’s kidneys about.
It seems to be more the dogma of an ID cult. There is no theoretical basis for it at all.
I’ll also note that it has no relevance to Gpuccio’s argument which is based on the comparison of existing proteins. There is no ‘unrelated state” involved.
This thread reminds me of one of the most significant events in the history of the ID movement, the infamous “Mathgrrl thread” from Uncommon Descent. Thankfully, William Dembski did not take the opportunity to make this embarrassing (for him and all ID’ers) debacle vanish down the memory hole when archiving the UD files. As I recall, that discussion went for over 400 posts without a single ID’er providing the requesting calculations of CSI for the straightforward scenarios provided, even though the ID’ers nonetheless bizarrely insisted that the questions had been answered.
I wonder how long we can go here before @Giltil or any other ID enthusiast either answers the posed question or admits they have no answer. I suspect the limiting factor will be the time to heat death of the universe.
You’re right, after reviewing Gpuccio’s posts, I could see that the 500 bits stuff need not emerge from an « unrelated state » in order to infer design. However, Gpuccio uses this notion of « unrelated state » when he defines FI. I quote him: « FI is related to a function. The function has that value of FI, which is the minimal number of specific bits that are necessary to implement the function. It is also the probability of generating that function in a random system from an unrelated state »
My apologies for my confusion and thanks for pointing my error.
Thank you.
Yeah I noticed he wrote that. That isn’t the definition of FI either, so I really don’t see why he says that. There’s nothing about how Hazen & Szostak defines it that says FI must be defined in terms of an “unrelated state”. They show how FI can increase with increasing activity or specificity of obviously still-related ribozymes and aptamers.
That may be what Gpuccio claims. It doesn’t mean that it’s true.
Those figures needn’t be the same. Certainly not in this context, So Gpuccio is making another obvious error,. And again, it is irrelevant to the argument under discussion because Gpuccio doesn’t calculate FI in that sense either.
I am confused. Is Gpuccio using Szostak’s definition or his own? If the former, it cannot matter what Gpuccio states about FI, as his statements cannot alter Szostak’s definition.
He can of course just make up is own definition of FI, it’s just he doesn’t even seem to be doing that as when he goes on to apply it he clearly uses it for cases where (he says based either on similarity or conservation) FI increases in a protein where both ancestor and descendant are related. In any case it doesn’t really make sense to bring in this related/unrelated stuff, because in any case whether the sequence similarity has dropped to zero or not, in so far as one sequence is the ancestor of the other they are still technically related.
The whole inference to design stuff is supposed to be about a probability, and when it reaches a probabily of 10-150 then they claim a design inference is warranted. 500 bits of Functional Information in biological systems is just a roundabout way of saying the same thing. So what matters is the probability of a particular series of events, not whether there is any degree of smilarity between one state and the other.
But any sufficiently long series of fitness increasing mutations is going to be very improbable, but natural selection can fix and accumulate fitness increasing mutations affecting one or more individual functions, so natural selection can produce arbitrarily high FI.
To remind everyone: all Gpuccio is measuring is the degree of sequence identity between some other species’s protein and a homologous human protein. A supposed “jump in FI” is really an increase in sequence similarity. (And he has to grossly distort the time scale to make even that increase a “jump”.) The protein with the highest FI is the one most similar to a human protein. That’s all.
Every time I look at Gpuccio’s argument there seems to be less that is solid. It started off as a blancmange, it now feels like a whole warehouse full of blancmanges.
Are you saying that he means to win Wimbledon?
No, that’s not all, not at all. Gpuccio emphasizes that in his methodology, mere sequence similarity is not enough to estimate FI. Another crucial condition beside similarity is conservation through long evolutionary periods. Now I know you dispute Gpuccio’s idea that conserved sequence similarity can be used to estimate FI. And yet this idea is not his own, he owe it to Durston et al.
But that is all that he measures.
Which he does not measure. And even if he did it would not give him a measure of FI in the sense he claims to be using. If you’d actually paid attention to the previous thread you would know that.