In BIO-Complexity, a New Model for Human Ancestry
More “cargo cult” science?
191 thoughts on “But Is It Science”
Leave a Reply
You must be logged in to post a comment.
The Skeptical Zone
"I beseech you, in the bowels of Christ, think it possible that you may be mistaken."
Mung,
All the species on which it has been done show common descent, by the same criteria by which you agree within your peculiarly-circumscribed taxonomic realm of applicability. You see a boundary, like colewd et al, but would be hard-pressed to identify its location, or why it has yet to be noticed by analysis.
The crux of the matter is that in paternity tests, forensics, or DNA-based taxonomy, the issue is that some actual organism is responsible for contributing a DNA sequence. The thing is that we actually know how that happens with organisms, via reproduction, or simply leaving behind DNA (often in forensics). And yet, while many who will accept the principle of paternity tests and forensics pointing to an organism that contributed a specific sequence, oddly enough, they won’t accept that an organism contributed a particular sequence to both humans and chimps, one that often will have been modified, but is still highly recognizable as sharing a common source.
I suppose that, logically, magic could do it, but why it doesn’t do so now yet we’re supposed to think that it did in the past escapes me. As well, the evidence for any magic is sparse, at best.
Glen Davidson
I’m sorry Glen, but humans are not descended from chimps. No paternity test result will tell you a chimp is your father.
Allan Miller,
Mung and I agree that common descent has occurred. The question we a struggling with is how much of the diversity we see is due to common descent? John Harshmans 2008 paper show evidence that points toward common descent but does not “show common descent” even among similar birds. This is what I think Ann is trying to do with one specie. We have a top down inference called common descent or universal common descent but no bottom up validation of even one transition.
Then what is your acceptance of common descent based on?
colewd,
Why do you think they do not show common descent? There are some nodes that are tangled, but that is simply a problem at that part of the tree. That group is still more closely clustered than an outgroup. It would be rather peculiar to argue that these birds are not, in fact, all birds, descended from a common bird ancestor, simply because we can’t tell in what order local branching occurred. If we pull back from the problem area to look at the wider clade, the problem does not come with us.
Same goes for whales. They are mammals. That is completely non-controversial, and the DNA evidence fully supports it. Which would be peculiar, if they weren’t commonly descended with mammals. Their DNA has more in common with other mammals than any outgroup – which is precisely what molecular genealogy and forensics is doing. But somehow, in whales ‘it’s different’. Sequence commonality, right down to bits of ‘fluff’, stops being evidence of descent, when elsewhere it was. But no idea as to where this happens in the taxonomic ranking.
And there’s much more yet going on with phylogenetic analysis than pairwise sequence. It’s vast swathes of data, multiple genes from multiple species. Look at the SINE data. It forms a nested hierarchy. What process can do that? I can think of one. ‘Common design from common need’ is not it.
You all put me in mind of someone looking at two photocopies of a page and concluding that the image of a squashed fly on one casts doubt upon their common source.
What does this mean? “Common descent has occurred”.
In the context of your second sentence it makes no sense whatsoever. If common descent “has occurred”, then all of the diversity we see is due to common descent. Otherwise it wouldn’t be common descent. It would be multiple independent lines of descent.
Common means shared. That implicitly means all species share descent. Meaning they descend from a shared ancestor.
So you’re willing to agree to a vacuous statement that “common descent has occurred”, though perhaps only within individual species. You make a distinction between “evidence that points toward” and “show”, but can you describe that distinction operationally? What sort of evidence, and how much, would be necessary to “show”? What would “bottom up” validation look like? I expect that as usual you will ignore these questions.
Incidentally, the singular is “species”, identical to the plural. “Specie” is a term referring to money.
Paternity tests aren’t limited to showing direct lines of descent(which admittedly makes the name a bit of a misnomer). They can tell you whether someone is your sibling, or cousin, too. Using the exact same principle that is used to show that chimps and humans are cousin species. If you reject the latter, you must reject the former too. Otherwise you are being a hypocrite with a double standard based on an arbitrary cutoff.
Perhaps he was thinking of specious, which refers to ID.
Rumraket, the original claim was about evidence that would be admissible in a court of law, more specifically, evidence of paternity.
First there was Glen:
Then there was Allan:
Why are paternity tests allowed in court? I bet the fact that we can actually test them against real fathers and their children helps.
I seriously doubt that a paternity test works like constructing a phylogenetic tree. What phylogenetics software package do they use when doing a paternity test?
And no, I don’t see any problem accepting paternity tests and not accepting that some LUCA is the common ancestor of us all because I don’t think the evidence and the reasoning is the same.
How many generations back to these human “paternity” tests work for? Can it tell if you and I have the same great-great-great grandfather?
No, but maybe a kissin’ cousin:
Once a broad sample of representative forms turn out to have a tree, does Mung insist that we need to first look at every last mite, every last fungus and every last jellyfish species? ‘Cause, if we don’t, one if the mites might turn out to be unrelated to the others?
Reslly?
And because new species are constantly being cataloged, there’s no way to ever complete the tree to the last node. Check mate atheists!
Mung,
As noted, you accept evolution in it’s entirety. Your are a TE. You have nothing.
Joe misses the point.
I have never personally done DNA comparisons or used any phylogenetics software. So if I accept common descent it has nothing to do with my having looked at DNA sequences or comparing DNA sequences. That’s probably the case for the vast majority of people who accept common descent.
Here I have to ask, in what way is diversity due to common descent? All our body cells are descended from the original zygote. Would you say that this common descent causes us to have liver cells, dkin cells and all the rest? That their appearance is due to common descent? Or would you say that common descent is the means by which they appear but their form is not due to this descent but due to something else?
In 1994 Ron Brady wrote a piece which I would say is still relevant today. Here is the abstract:
IMO common descent is a description not an explanation. You seem to think that it is an explanation.
I have both purified, sequenced and compared DNA in my work. And I have used phylogenetics software as part of a course I took a few years ago.
That’s not why I accept common descent though, I accept common descent because I read and understood the case for it. And I understand it is now, today, based largely on comparing DNA sequences.
Just like the case for elucidating your relationships to another person, in a kinship test (see, I used a more correct word there), is based on comparing DNA sequences. In both cases, the relationships are computed by counting and comparing the shared and differing traits. These in turn determine where you are put in the family tree.
My link to the Ron Brady piece hasn’t come up, here it is:
http://natureinstitute.org/txt/rb/pdf/1994a_explanation_description.pdf
Rumraket,
So I take this to mean that common descent = universal common descent. Does everyone agree with this?
Mung,
Yes, I assume there is a standard of DNA comparison that determines whether there is a paternal relationship.
What is the standard that determines that species A and B descended from C? If this does not exist the claim of universal common descent is a house of cards.
True that is what I took it to mean. If you don’t mean universal common descent when you say common descent, you should say who’s common desent you are speaking of.
No, I agree with you there. Common descent does not explain why livers (for example) evolved in the first place. It does explain why all subsequent lineages have livers, because they inherited them.
People keep explaining things to you and you keep asking the same questions. First, it is seldom possible to determine that species A and B descended from C, since ancestral species are seldom available for inspection, and even when they are (fossils, e.g.), we have no means of distinguishing them from uncles and cousins. So in fact what we can determine is that species A, B, and C descended from an unknown common ancestor, and we do this by showing that the data are compatible with that hypothesis and no other. The standard for paternity tests is arbitrary; by assessing more an more data one can make the uncertainty smaller and smaller, but never zero. The same is true for phylogeny. The concordance of more and more data can make us more and more nearly certain of the pattern of the tree, but never absolutely certain. Science doesn’t deal in certainty. Still, for all practical purposes we can be sure of much.
Perhaps you should look at that paleognath paper with a view toward looking at the various standards mentioned there. To be specific, the most common standard of phylogenetic inference is the bootstrap. I believe I’ve recently referenced Felsenstein’s paper that introduced the practice. When you look at trees, the branches will most commonly be attended by bootstrap values. Anything much over 70% is quite strong, and anything close to 100% is as near certain as we can be. There are various caveats to mention if you really wanted to get into it, but that’s good enough for now.
Mung,
You are confusing two issues. I have been talking about simple pairwise comparison. You don’t build trees when you do that. You can do pairwise comparison on anything from siblings to a cat and a duck; it’s the same basic techniques available.
Trees are the next step along, when you have a set of such pairs and want to see how they relate to each other, in that remarkable way these supposedly completely unrelated datasets do. As far as pairwise comparison goes, the pair can be anything. DNA is DNA. Any two pieces have both similarities and differences. The dispute appears to be on the origin of the similarities. Two random DNA sequences would not align very much at all, even if we did have 25% bit-by-bit identity in a 4-letter alphabet. What’s remarkable are the runs of sequence held in common, should descent prove not to be the cause.
On the Cole view of tree-building, a useful way of increasing one’s confidence levels would be to render extinct all but one of the species at problem nodes. As if by magic, the tree for the remainder would become robust, due to the removal of these nodal uncertainties. The strange thing is, it wouldn’t matter which ones you killed … .
John Harshman,
How does the confidence of phylogeny accuracy compare to paternity accuracy? I would assume that paternity accuracy is at least 99% or paternity suits would not exist.
So according to your explanation we cannot identify single common ancestors.
Can you explain how bootstrap values are calculated? What would you expect the bootstrap value of my children to be?
We have lots of evidence of ancestry relationships to draw from. Where we know the relatives and compare the DNA. Is there any paleognaths in your paper where the DNA comparisons would meet the same criteria of relationships like canine DNA which I think we determined varied by about .5%?
And I don’t think that they use just any old sequences of DNA for paternity testing.
The confidence of either varies with the amount of data assessed. How much you need for a given level of confidence can vary quite a bit too. But we can be more than 99% certain of a great many phylogenetic conclusions, if that’s your standard.
We can’t, and we don’t try, and it doesn’t matter.
Your children don’t have bootstrap values. Here’s how a phylogenetic bootstrap works: you resample the data matrix many times, with replacement, do a phylogenetic analysis on each resampled matrix. The bootstrap value of a group is the percentage of resampled trees on which that group appears. It’s an index of the inner consistency of signal in the data.
We have evidence of ancestry of individuals, because we can get DNA sequence out of the ancestors; but in phylogeny we’re talking about species, the ancestors of which are dead. I have no idea what you think the criteria of relationships for canine DNA would be, or why .5% should be considered a relevant number.
I notice that I answered all your questions and you never answered any of mine. But I have said that I expected that.
John Harshman,
My thinking here is that due to breeding we know that dogs are genetically related so their divergence of DNA is due to ancestry. If the divergence of the DNA of two species of paleognaths was .5% that would be within a benchmark number and would be additional evidence of common descent. The .5% is from a paper that you and I discussed in the past.
My apologies. I have been very busy recently with business and personal issues.
In science that’s considered to be the finding of the cause of genetic relatedness.
Science builds on such discoveries.
Non-science finds excuses to doubt what has been established.
Glen Davidson
So you’re saying we can’t infer common ancestry unless we either saw it happen or there is no more divergence than within a group where we saw it happen. As I suspected, you’re going with the “Were you there?” trope. Do you reject the past existence of Pangaea because we’ve only seen continents move a few inches? Do you reject radiometric dating because we’ve only seen uranium decay by a tiny amount? This is a seriously screwed-up view of science.
The inference of common descent doesn’t come from the percentage of difference. It comes from the pattern of similarities and differences, the nested hierarchy of life, or more generally the superior fit to data of common descent over other hypotheses. There is no other explanation for that pattern. “Common design” means nothing. “Similar designs for similar purposes” explains neither the nested hierarchy nor the differences. Now you may not like the word “inference”. You may even equate it with “wild guess”. But it’s how all scientific knowledge is gained. Your dichotomy between “direct observation” and mere inference is a false one.
Mung,
So what? This just seems like another attempt to argue yourself right retrospectively. Phylogeny doesn’t use ‘any old DNA’ either, if that is a category now. Any pairwise comparison must use a part or parts of the genome that is informative for the particular comparison being performed. Neither 100% identity nor 100% difference is fully informative (although 100% identity is pretty good evidence of common descent, I would say).
Different sequences will be informative at different levels. The objection ‘they use different things’ is not particularly devastating. One would have to wonder, yet again, what other mechanism may be at play in the sequences used for inter-species comparison, vs those used in genealogy, if one is going for the ‘but … it’s not the same DNA!’ gambit. .
colewd,
Why would interbreeding be a criterion? Rather few genes are responsible for fundamental reproductive isolation, where this is constitutional rather than circumstantial.
If 2 genomes are 99% identical, why would we agree they were related if we could achieve a successful F1 meiosis between them, non-artificially, but regard it as an unsupportable inference if we couldn’t?
Surely, in a world with evolution, separated populations would eventually become truly incompatible? Further divergence would progressively erode the signal of commonality. So we should not be so hasty as to declare compatible individuals as commonly descended and all other potential pairs as not. That seems far too dichotomous a view. And it is not, indeed, a view that anyone seems to be pushing here. Their dichotomy exists somewhere else, usually much further up the taxonomic hierarchy (‘s long as you’re not a chimp!). But no justification is offered, just vague critiques of the general inference, which is agreed with up to a point.
So the fact that paternity tests are allowed in a court of law is not a compelling argument for common descent of all life and people can reasonably accept one without accepting the other without being inconsistent.
I propose that even with paternity testing at some point the signal gets lost in the noise, and that’s all within just a single species. No one has yet to answer that challenge.
How many generations back to these human “paternity” tests work for? Can it tell if you and I have the same great-great-great grandfather?
On New Model for Human Ancestry, Mathematician Ola Hössjer Responds to Critics – Evolution News & Views
Mung,
You are leaping from paternity tests to ‘all life’ again. I told you I wasn’t doing that. What is it with you people and dichotomies? 😉 Let’s just stick to closer relationships for now.- within-genus, within-(taxonomic-)family, say. On what basis do you accept the commonality of a particular genome stretch in a paternity test, but reject identity on DNA stretches at a higher taxonomic level? That seems wholly inconsistent to me. And which taxonomic level are you settling on?
Yes.
Paternity testing is perhaps not the best term, but thhis kind of test can resolve on any scale, if you have a marker that changes on the scale of interest. Cytochrome c would be no good, but active Alus would be fine. Like radioactive isotopes, it’s horses for courses.
Allan Miller,
The first can be observed and actually tested. And the latter can neither be observed nor tested. And a common design easily explains then latter- see Linnaean taxonomy which lays out the common design.
Frankie,
I repeat, please stop with “there is no theory of evolution” nonsense.
Allan Miller,
Of course common genes works. It follows directly from Linnaean classification. No need to reinvent genes to produce similar tissues and organs.
Until you can link to it I will stay the course. And you can’t link to it, Alan. Why is that?
I can also quote biologists who have said there isn’t a theory of evolution.
So it looks like the nonsense is all yours.