As I looked back yesterday on some 2016 threads in s.b.p. when it was at the height of what I call
"an oasis of civilization," I came across a long exchange between Ruben Safir
[who often went under the byline Popping mad, as he does now]
and John Harshman, which I joined near the end. John and Ruben
were unable to communicate fruitfully, because John was stuck
in a dichotomy between two ways of classifying organisms:
the cladistic and the phenetic.

There is a happy medium between these ways, and Popping mad seems to
have been aiming for it with his talk about distances, but he seemed to
have underestimated the difficulty of getting it across.

I will talk about a remedy for that in the next post but one.
[Spoiler: it uses the difficult concept of disparity.]
In the next one, I focus on another instance of the dichotomy,
in what is called "the cladist wars".

Evidently the vertebrate paleontologists, if any, were so badly outnumbered
by the classifiers of extant organisms that their point of view never came across,
and the cladists won out without any dilution of their ideas.

More about that in the next post. For now, I say a few things about the
two extremes, the cladistic and the phenetic.

The heart of cladistics is found in the phylogenetic trees, which put all
organisms at the branch tips. The taxa resulting from these trees
are clades: everything above a fixed node in a tree constitutes a clade.

Clades are exactly what cladistic systematists include in their classifications.
They bar all old taxa of the Linnean classification,
such as Amphibia, which do not include all descendants
(in this case, reptiles, birds and mammals are not included).

Phenetics is completely out of fashion, so I only mention that
it classified organisms on the basis of overall similarity.
Phylogeny was not taken into account; nevertheless, it gave
reasonably good results because all parts of the organisms
were taken into account; this was possible because only
extant organisms were classified.

Peter Nyikos
Professor, Dept. of Mathematics -- standard disclaimer--
University of South Carolina
http://people.math.sc.edu/nyikos

On 9/5/22 11:42 AM, Peter Nyikos wrote:
> As I looked back yesterday on some 2016 threads in s.b.p. when it was at the height of what I call
> "an oasis of civilization," I came across a long exchange between Ruben Safir
> [who often went under the byline Popping mad, as he does now]
> and John Harshman, which I joined near the end. John and Ruben
> were unable to communicate fruitfully, because John was stuck
> in a dichotomy between two ways of classifying organisms:
> the cladistic and the phenetic.

Needless to say, I reject your characterization of that argument.

> There is a happy medium between these ways, and Popping mad seems to
> have been aiming for it with his talk about distances, but he seemed to
> have underestimated the difficulty of getting it across.
>
> I will talk about a remedy for that in the next post but one.
> [Spoiler: it uses the difficult concept of disparity.]
> In the next one, I focus on another instance of the dichotomy,
> in what is called "the cladist wars".
>
> Evidently the vertebrate paleontologists, if any, were so badly outnumbered
> by the classifiers of extant organisms that their point of view never came across,
> and the cladists won out without any dilution of their ideas.

This is not, apparently, a subject you know much about. And in fact
several of the most prominent proponents of cladistic classification
were paleontologists. Nobody was drowned out.

> More about that in the next post. For now, I say a few things about the
> two extremes, the cladistic and the phenetic.
>
>
> The heart of cladistics is found in the phylogenetic trees, which put all
> organisms at the branch tips. The taxa resulting from these trees
> are clades: everything above a fixed node in a tree constitutes a clade.
>
> Clades are exactly what cladistic systematists include in their classifications.
> They bar all old taxa of the Linnean classification,
> such as Amphibia, which do not include all descendants
> (in this case, reptiles, birds and mammals are not included).

Not true, of course. "Amphibia" was merely re-ordered so as to exclude
everything except the crown group. All those "labryinthodonts" and such
are mrely primitive tetrapods, not ambphibians.

> Phenetics is completely out of fashion, so I only mention that
> it classified organisms on the basis of overall similarity.
> Phylogeny was not taken into account; nevertheless, it gave
> reasonably good results because all parts of the organisms
> were taken into account; this was possible because only
> extant organisms were classified.

How would you define "reasonably good results"? Matching the actual
cladistic relationships? It was of course the claim of pheneticists that
discovering phylogeny was impossible, so it was pointless to try. I
doubt you would agree.

On 9/5/22 11:42 AM, Peter Nyikos wrote:
> As I looked back yesterday on some 2016 threads in s.b.p. when it was at the height of what I call
> "an oasis of civilization," I came across a long exchange between Ruben Safir
> [who often went under the byline Popping mad, as he does now]
> and John Harshman, which I joined near the end. John and Ruben
> were unable to communicate fruitfully, because John was stuck
> in a dichotomy between two ways of classifying organisms:
> the cladistic and the phenetic.

Wait, are you going to introduce Mayr's "evolutionary systematics",
which tries to combine cladistic and phenetic approaches? If so, you are
being oddly coy in not starting out that way.

On Monday, September 5, 2022 at 2:57:07 PM UTC-4, John Harshman wrote:

> Wait, are you going to introduce Mayr's "evolutionary systematics",
> which tries to combine cladistic and phenetic approaches? If so, you are
> being oddly coy in not starting out that way.

From what I've read about it, it didn't use disparity sufficiently well
to make much of a difference. A number of researchers in various
fields are trying to work out a measure for disparity now, but
it is very slow going, because disparity is extraordinarily difficult
to quantify even under the best of conditions.

That's for my next post. Here I go with the exposition on the cladist wars.

The only specific event of those wars that I have ever read about (from two different sources,
one of which I own: Kenneth S. Thompson's LIVING FOSSIL: The Story of the Coelacanth)
is the 1978 event "the lungfish, the salmon, and the cow". It had to do with the radical
definition by cladists of the word "related". The phenetic side claimed that it was absurd
to regard a lungfish to be more closely related to a cow than to a salmon, but that was
a naive choice of taxa, and the outcome was an undeserved victory for the cladist side.

Had a Romer-savvy vertebrate paleontologist been involved, the choice of taxa could have been very different.
One choice readily available at that time was "Bos, Ichthyostega, Elpistostege."
[Nowadays, the third is better replaced by the more familiar Tiktaalik.]
It does violence to our ordinary idea of human relationships to claim that Ichthyostega
is more closely related to us human beings than it is to Elpistostege.
It's almost as bad as saying that Mitochondrial Eve is more closely related to
everyone alive today than she was to anyone in her family at the time she was born.

Had the victory gone the other way, we might be far advanced in a definition of "more related"
that combines phylogeny with measures of disparity. As it is, the theory of macroevolution,
to which disparity is an indispensable tool, has made comparatively little progress to date.

Peter Nyikos
Professor, Dept. of Mathematics -- standard disclaimer--
Univ. of South Carolina at Columbia
http://people.math.sc.edu/nyikos

On 9/5/22 12:35 PM, Peter Nyikos wrote:
> On Monday, September 5, 2022 at 2:57:07 PM UTC-4, John Harshman wrote:
>
>> Wait, are you going to introduce Mayr's "evolutionary systematics",
>> which tries to combine cladistic and phenetic approaches? If so, you are
>> being oddly coy in not starting out that way.
>
> From what I've read about it, it didn't use disparity sufficiently well
> to make much of a difference. A number of researchers in various
> fields are trying to work out a measure for disparity now, but
> it is very slow going, because disparity is extraordinarily difficult
> to quantify even under the best of conditions.

What have you read about it?

> That's for my next post. Here I go with the exposition on the cladist wars.
>
> The only specific event of those wars that I have ever read about (from two different sources,
> one of which I own: Kenneth S. Thompson's LIVING FOSSIL: The Story of the Coelacanth)
> is the 1978 event "the lungfish, the salmon, and the cow". It had to do with the radical
> definition by cladists of the word "related". The phenetic side claimed that it was absurd
> to regard a lungfish to be more closely related to a cow than to a salmon, but that was
> a naive choice of taxa, and the outcome was an undeserved victory for the cladist side.
>
> Had a Romer-savvy vertebrate paleontologist been involved, the choice of taxa could have been very different.
> One choice readily available at that time was "Bos, Ichthyostega, Elpistostege."
> [Nowadays, the third is better replaced by the more familiar Tiktaalik.]
> It does violence to our ordinary idea of human relationships to claim that Ichthyostega
> is more closely related to us human beings than it is to Elpistostege.
> It's almost as bad as saying that Mitochondrial Eve is more closely related to
> everyone alive today than she was to anyone in her family at the time she was born.

Depends on the definition of "more closely related" you want to pick.
Now, would you agree that by any definition you might be more closely
related to a person who looks different from you than to a person who
looks similar to you? If so, then why is disparity relevant even under
the ordinary human definition?

But of course the cladistic definition is simple and obvious: recency of
common ancestry. Your common ancestor with Ichthyostega is more recent
than your (or Ichthyostega's) common ancestor with Tiktaalik, thus you
are more closely related to Ichthyostega than either of you is to
Tiktaalik. And this can be objectively determined by phylogenetic
analysis, an advantage for any definition: it's operational.

> Had the victory gone the other way, we might be far advanced in a definition of "more related"
> that combines phylogeny with measures of disparity. As it is, the theory of macroevolution,
> to which disparity is an indispensable tool, has made comparatively little progress to date.

That lack of progress wouldn't seem to have anything to do with
cladistic classification. Why would you think so?

On Monday, September 5, 2022 at 2:50:32 PM UTC-4, John Harshman wrote:
> On 9/5/22 11:42 AM, Peter Nyikos wrote:

> > As I looked back yesterday on some 2016 threads in s.b.p. when it was at the height of what I call
> > "an oasis of civilization," I came across a long exchange between Ruben Safir
> > [who often went under the byline Popping mad, as he does now]
> > and John Harshman, which I joined near the end. John and Ruben
> > were unable to communicate fruitfully, because John was stuck
> > in a dichotomy between two ways of classifying organisms:
> > the cladistic and the phenetic.

> Needless to say, I reject your characterization of that argument.

Automatic reflex reaction noted.

You obviously haven't looked at the thread in over six years,
not with Giganews archives only going back a month.

But I'll help you. Here is the url for the entire thread.
https://groups.google.com/g/sci.bio.paleontology/c/sFY6QxipSb4/m/T0QLNBF0AQAJ

Happy surfing up and down the 148 posts. Here is a note I made on one of them:

"Harshman is stuck in a false dichotomy between cladistics and phenetics"
Apr 14, 2016, 5:03:15 PM

You may want to start there.

See you tomorrow with the post that addresses this false dichotomy of yours.

Peter Nyikos

On 9/5/22 3:51 PM, Peter Nyikos wrote:
> On Monday, September 5, 2022 at 2:50:32 PM UTC-4, John Harshman wrote:
>> On 9/5/22 11:42 AM, Peter Nyikos wrote:
>
>>> As I looked back yesterday on some 2016 threads in s.b.p. when it was at the height of what I call
>>> "an oasis of civilization," I came across a long exchange between Ruben Safir
>>> [who often went under the byline Popping mad, as he does now]
>>> and John Harshman, which I joined near the end. John and Ruben
>>> were unable to communicate fruitfully, because John was stuck
>>> in a dichotomy between two ways of classifying organisms:
>>> the cladistic and the phenetic.
>
>> Needless to say, I reject your characterization of that argument.
>
> Automatic reflex reaction noted.

Try to engage with my argument rather than just characterizing it.

> You obviously haven't looked at the thread in over six years,
> not with Giganews archives only going back a month.

Of course I haven't. But I know my own opinions.

> But I'll help you. Here is the url for the entire thread.
> https://groups.google.com/g/sci.bio.paleontology/c/sFY6QxipSb4/m/T0QLNBF0AQAJ
>
> Happy surfing up and down the 148 posts. Here is a note I made on one of them:
>
> "Harshman is stuck in a false dichotomy between cladistics and phenetics"
> Apr 14, 2016, 5:03:15 PM
>
> You may want to start there.
>
>
> See you tomorrow with the post that addresses this false dichotomy of yours.

Why begin with a personal attack? What's the point?

On 9/5/22 3:51 PM, Peter Nyikos wrote:
> On Monday, September 5, 2022 at 2:50:32 PM UTC-4, John Harshman wrote:
>> On 9/5/22 11:42 AM, Peter Nyikos wrote:
>
>>> As I looked back yesterday on some 2016 threads in s.b.p. when it was at the height of what I call
>>> "an oasis of civilization," I came across a long exchange between Ruben Safir
>>> [who often went under the byline Popping mad, as he does now]
>>> and John Harshman, which I joined near the end. John and Ruben
>>> were unable to communicate fruitfully, because John was stuck
>>> in a dichotomy between two ways of classifying organisms:
>>> the cladistic and the phenetic.
>
>> Needless to say, I reject your characterization of that argument.
>
> Automatic reflex reaction noted.
>
> You obviously haven't looked at the thread in over six years,
> not with Giganews archives only going back a month.
>
> But I'll help you. Here is the url for the entire thread.
> https://groups.google.com/g/sci.bio.paleontology/c/sFY6QxipSb4/m/T0QLNBF0AQAJ
>
> Happy surfing up and down the 148 posts. Here is a note I made on one of them:
>
> "Harshman is stuck in a false dichotomy between cladistics and phenetics"
> Apr 14, 2016, 5:03:15 PM
>
> You may want to start there.

There is no post with that time stamp. Is there perhaps a typo? Or does
google try to match time zones? There is a post at 2:03:15 PM that might
be the one you're thinking of, but it doesn't say anything about
cladistics vs. phenetics, so seems to have nothing to do with your note.
Puzzling.

On Monday, September 5, 2022 at 5:14:48 PM UTC-7, John Harshman wrote:
> On 9/5/22 3:51 PM, Peter Nyikos wrote:
> > On Monday, September 5, 2022 at 2:50:32 PM UTC-4, John Harshman wrote:
> >> On 9/5/22 11:42 AM, Peter Nyikos wrote:
> >
> >>> As I looked back yesterday on some 2016 threads in s.b.p. when it was at the height of what I call
> >>> "an oasis of civilization," I came across a long exchange between Ruben Safir
> >>> [who often went under the byline Popping mad, as he does now]
> >>> and John Harshman, which I joined near the end. John and Ruben
> >>> were unable to communicate fruitfully, because John was stuck
> >>> in a dichotomy between two ways of classifying organisms:
> >>> the cladistic and the phenetic.
> >
> >> Needless to say, I reject your characterization of that argument.
> >
> > Automatic reflex reaction noted.
> >
> > You obviously haven't looked at the thread in over six years,
> > not with Giganews archives only going back a month.
> >
> > But I'll help you. Here is the url for the entire thread.
> > https://groups.google.com/g/sci.bio.paleontology/c/sFY6QxipSb4/m/T0QLNBF0AQAJ
> >
> > Happy surfing up and down the 148 posts. Here is a note I made on one of them:
> >
> > "Harshman is stuck in a false dichotomy between cladistics and phenetics"
> > Apr 14, 2016, 5:03:15 PM
> >
> > You may want to start there.
> There is no post with that time stamp. Is there perhaps a typo? Or does
> google try to match time zones? There is a post at 2:03:15 PM that might
> be the one you're thinking of, but it doesn't say anything about
> cladistics vs. phenetics, so seems to have nothing to do with your note.
> Puzzling.

Note that you made two posts, one stamped 5:00 and another 5:03 in that thread on that day, according to my server.
I think you misunderstand what Peter said and meant above.

On 9/6/22 8:29 AM, Glenn wrote:
> On Monday, September 5, 2022 at 5:14:48 PM UTC-7, John Harshman wrote:
>> On 9/5/22 3:51 PM, Peter Nyikos wrote:
>>> On Monday, September 5, 2022 at 2:50:32 PM UTC-4, John Harshman wrote:
>>>> On 9/5/22 11:42 AM, Peter Nyikos wrote:
>>>
>>>>> As I looked back yesterday on some 2016 threads in s.b.p. when it was at the height of what I call
>>>>> "an oasis of civilization," I came across a long exchange between Ruben Safir
>>>>> [who often went under the byline Popping mad, as he does now]
>>>>> and John Harshman, which I joined near the end. John and Ruben
>>>>> were unable to communicate fruitfully, because John was stuck
>>>>> in a dichotomy between two ways of classifying organisms:
>>>>> the cladistic and the phenetic.
>>>
>>>> Needless to say, I reject your characterization of that argument.
>>>
>>> Automatic reflex reaction noted.
>>>
>>> You obviously haven't looked at the thread in over six years,
>>> not with Giganews archives only going back a month.
>>>
>>> But I'll help you. Here is the url for the entire thread.
>>> https://groups.google.com/g/sci.bio.paleontology/c/sFY6QxipSb4/m/T0QLNBF0AQAJ
>>>
>>> Happy surfing up and down the 148 posts. Here is a note I made on one of them:
>>>
>>> "Harshman is stuck in a false dichotomy between cladistics and phenetics"
>>> Apr 14, 2016, 5:03:15 PM
>>>
>>> You may want to start there.
>> There is no post with that time stamp. Is there perhaps a typo? Or does
>> google try to match time zones? There is a post at 2:03:15 PM that might
>> be the one you're thinking of, but it doesn't say anything about
>> cladistics vs. phenetics, so seems to have nothing to do with your note.
>> Puzzling.
>
> Note that you made two posts, one stamped 5:00 and another 5:03 in that thread on that day, according to my server.
> I think you misunderstand what Peter said and meant above.

Excellent. Since you understood what he said and meant, could you
explain it to me?

On Monday, September 5, 2022 at 8:14:48 PM UTC-4, John Harshman wrote:
> On 9/5/22 3:51 PM, Peter Nyikos wrote:

> > But I'll help you. Here is the url for the entire thread.
> > https://groups.google.com/g/sci.bio.paleontology/c/sFY6QxipSb4/m/T0QLNBF0AQAJ
> >
> > Happy surfing up and down the 148 posts. Here is a note I made on one of them:
> >
> > "Harshman is stuck in a false dichotomy between cladistics and phenetics"
> > Apr 14, 2016, 5:03:15 PM
> >
> > You may want to start there.

> There is no post with that time stamp. Is there perhaps a typo? No. Or does
> google try to match time zones?

Well, it obviously gave it to you in Pacific Daylight time, to me in Eastern Daylight time.

>There is a post at 2:03:15 PM that might
> be the one you're thinking of, but it doesn't say anything about
> cladistics vs. phenetics, so seems to have nothing to do with your note.
> Puzzling.

Take another look at the post:

On 4/14/16 1:28 PM, ruben safir wrote:
> On 04/14/2016 12:09 AM, John Harshman wrote:
>> I'm sorry, but this is incorrect in regard to species. It would be true
>> if evolution were perfectly clocklike, b
>
> no that is not true. time is not a measured characteristic in closest
> path analysis. It might be that two species are a billion years
> separated, but still closely related, and the shortest path analysis
> will demonstrate that relationship.
>
No it won't, unless by "closely related" you just mean "similar". But
what the term means in phylogenetics is "sharing a more recent common
ancestry than with some third taxon", i.e. the sort of thing
phylogenetic trees say. There is no particular justification for making
trees that just measure similarity. (There are methods for making such
trees, e.g. UPGMA, but why would you use them?)

++++++++++++++++ end of post ++++++++++++++++++++

Your very first sentence is of the very essence of phenetics.
And the only alternative you seriously propose is pure cladistics.
What kept you from seeing that?

Puzzling, indeed!

Peter Nyikos
Professor, Dept. of Mathematics -- standard disclaimer--
Univ. of South Carolina at Columbia
http://people.math.sc.edu/nyikos

PS I'll give you about an hour to mull over the above before I post the alternative
that I think Ruben had in mind. Use the time wisely instead of shooting from
the hip in a reply within 5 minutes, like you so often do.

On Tuesday, September 6, 2022 at 8:39:56 PM UTC-4, Peter Nyikos wrote:

++++++++++++++++++++++++ begin included post+++++++++++++++++
> On 4/14/16 1:28 PM, ruben safir wrote:
> > On 04/14/2016 12:09 AM, John Harshman wrote:
> >> I'm sorry, but this is incorrect in regard to species. It would be true
> >> if evolution were perfectly clocklike, b
> >
> > no that is not true. time is not a measured characteristic in closest
> > path analysis. It might be that two species are a billion years
> > separated, but still closely related, and the shortest path analysis
> > will demonstrate that relationship.
> >
> No it won't, unless by "closely related" you just mean "similar". But
> what the term means in phylogenetics is "sharing a more recent common
> ancestry than with some third taxon", i.e. the sort of thing
> phylogenetic trees say. There is no particular justification for making
> trees that just measure similarity. (There are methods for making such
> trees, e.g. UPGMA, but why would you use them?)
>
> ++++++++++++++++ end of post ++++++++++++++++++++
>
> Your very first sentence is of the very essence of phenetics.
> And the only alternative you seriously propose is pure cladistics.

So, what did Ruben have in mind? I believe it was a measure of
disparity between elements (nodes and branch tips) of an evolutionary
tree where the branch tips (and perhaps some nodes) are taxa.

The line segment joining successive elements is given a number estimating degree of disparity,
and the numbers are added together to compute the length of the path between them.

The concept of "more closely related" then could be given a whole new meaning --
or, rather, an old meaning but now quantified. We would be able to say that vertebrate A
in the following example is MUCH more closely related to B than it is to C even though
the LCA of A and B is strictly ancestral to the LCA of B and C.

This is the example we had earlier: A = Tiktaalik, B= Ichthyostega, C = Bos.
A cladist would say B and C are equally distant from A because they are in a clade that excludes A.

A pheneticist would agree with the disparity measurer in this example,
but I think it is not hard to cook up an example where evolutionary convergence
could make the pheneticist disagree with the disparity measurer,
because the latter takes a plunge down the tree and back up to do the measuring,
but the pheneticist hops across the tree in a beeline from one species to another.

I submit that this concept does justice to phylogeny besides being
exactly the sort of thing one would expect from our ordinary
human concept of "more closely related."

Peter Nyikos
Professor, Dept. of Mathematics -- standard disclaimer--
Univ. of South Carolina at Columbia
http://people.math.sc.edu/nyikos

PS Remember those little children who thought it was "way cool"
that birds are dinosaurs? Do you think you get them to believe
that your concept is better than the one with which they
are familiar (aunts, uncles, cousins, etc.) and which the disparity-measurer's
concept so faithfully mirrors?

On 9/6/22 5:39 PM, Peter Nyikos wrote:
> On Monday, September 5, 2022 at 8:14:48 PM UTC-4, John Harshman wrote:
>> On 9/5/22 3:51 PM, Peter Nyikos wrote:
>
>>> But I'll help you. Here is the url for the entire thread.
>>> https://groups.google.com/g/sci.bio.paleontology/c/sFY6QxipSb4/m/T0QLNBF0AQAJ
>>>
>>> Happy surfing up and down the 148 posts. Here is a note I made on one of them:
>>>
>>> "Harshman is stuck in a false dichotomy between cladistics and phenetics"
>>> Apr 14, 2016, 5:03:15 PM
>>>
>>> You may want to start there.
>
>> There is no post with that time stamp. Is there perhaps a typo? No. Or does
>> google try to match time zones?
>
> Well, it obviously gave it to you in Pacific Daylight time, to me in Eastern Daylight time.
>
>
>> There is a post at 2:03:15 PM that might
>> be the one you're thinking of, but it doesn't say anything about
>> cladistics vs. phenetics, so seems to have nothing to do with your note.
>> Puzzling.
>
>
> Take another look at the post:
>
> On 4/14/16 1:28 PM, ruben safir wrote:
>> On 04/14/2016 12:09 AM, John Harshman wrote:
>>> I'm sorry, but this is incorrect in regard to species. It would be true
>>> if evolution were perfectly clocklike, b
>>
>> no that is not true. time is not a measured characteristic in closest
>> path analysis. It might be that two species are a billion years
>> separated, but still closely related, and the shortest path analysis
>> will demonstrate that relationship.
>>
> No it won't, unless by "closely related" you just mean "similar". But
> what the term means in phylogenetics is "sharing a more recent common
> ancestry than with some third taxon", i.e. the sort of thing
> phylogenetic trees say. There is no particular justification for making
> trees that just measure similarity. (There are methods for making such
> trees, e.g. UPGMA, but why would you use them?)
>
> ++++++++++++++++ end of post ++++++++++++++++++++
>
> Your very first sentence is of the very essence of phenetics.
> And the only alternative you seriously propose is pure cladistics.
> What kept you from seeing that?

Because we're talking about trees. Combining phenetic and cladistic
criteria doesn't give you a tree. It might give you a classification,
but that's not the same thing. Even Ernst Mayr proposed that you find
relationships using cladistic methods and then add phenetic closeness in
classification.

> Puzzling, indeed!
>
>
> Peter Nyikos
> Professor, Dept. of Mathematics -- standard disclaimer--
> Univ. of South Carolina at Columbia
> http://people.math.sc.edu/nyikos
>
> PS I'll give you about an hour to mull over the above before I post the alternative
> that I think Ruben had in mind. Use the time wisely instead of shooting from
> the hip in a reply within 5 minutes, like you so often do.

I assure you that I give your posts all the attention and thought they
deserve. No fear on that score.

On 9/6/22 6:53 PM, Peter Nyikos wrote:
> On Tuesday, September 6, 2022 at 8:39:56 PM UTC-4, Peter Nyikos wrote:
>
> ++++++++++++++++++++++++ begin included post+++++++++++++++++
>> On 4/14/16 1:28 PM, ruben safir wrote:
>>> On 04/14/2016 12:09 AM, John Harshman wrote:
>>>> I'm sorry, but this is incorrect in regard to species. It would be true
>>>> if evolution were perfectly clocklike, b
>>>
>>> no that is not true. time is not a measured characteristic in closest
>>> path analysis. It might be that two species are a billion years
>>> separated, but still closely related, and the shortest path analysis
>>> will demonstrate that relationship.
>>>
>> No it won't, unless by "closely related" you just mean "similar". But
>> what the term means in phylogenetics is "sharing a more recent common
>> ancestry than with some third taxon", i.e. the sort of thing
>> phylogenetic trees say. There is no particular justification for making
>> trees that just measure similarity. (There are methods for making such
>> trees, e.g. UPGMA, but why would you use them?)
>>
>> ++++++++++++++++ end of post ++++++++++++++++++++
>>
>> Your very first sentence is of the very essence of phenetics.
>> And the only alternative you seriously propose is pure cladistics.
>
> So, what did Ruben have in mind? I believe it was a measure of
> disparity between elements (nodes and branch tips) of an evolutionary
> tree where the branch tips (and perhaps some nodes) are taxa.

I suspect that you have misunderstood what he had in mind, but perhaps
he will explain himself.

> The line segment joining successive elements is given a number estimating degree of disparity,
> and the numbers are added together to compute the length of the path between them.
>
> The concept of "more closely related" then could be given a whole new meaning --
> or, rather, an old meaning but now quantified. We would be able to say that vertebrate A
> in the following example is MUCH more closely related to B than it is to C even though
> the LCA of A and B is strictly ancestral to the LCA of B and C.
>
> This is the example we had earlier: A = Tiktaalik, B= Ichthyostega, C = Bos.
> A cladist would say B and C are equally distant from A because they are in a clade that excludes A.
>
> A pheneticist would agree with the disparity measurer in this example,
> but I think it is not hard to cook up an example where evolutionary convergence
> could make the pheneticist disagree with the disparity measurer,
> because the latter takes a plunge down the tree and back up to do the measuring,
> but the pheneticist hops across the tree in a beeline from one species to another.
>
>
> I submit that this concept does justice to phylogeny besides being
> exactly the sort of thing one would expect from our ordinary
> human concept of "more closely related."

You may think so, but I don't. If I understand your point, you are using
patristic distance as a measure of relatedness, smaller distances being
more close relationships. But these distances would depend on what
characters you used to make the measured tree, and there is no real
objective way to choose these measures of disparity. I don't see
patristic distance as much better for this purpose than simple pairwise
distance. Further, I think it would lead to overlapping groups if used
in classification. Velociraptor closer to Archaeopteryx than
Archaeopteryx to Confuciusornis. Archaeopteryx closer to Confuciusornis
than Confusiusornis to Icthyornis. Icthyornis closer to Passer than
Ichthyornis to Confuciusornis. But where do we put the line here? And
what's closer to Hesperornis? It would seem to be out on its own.

> Peter Nyikos
> Professor, Dept. of Mathematics -- standard disclaimer--
> Univ. of South Carolina at Columbia
> http://people.math.sc.edu/nyikos
>
> PS Remember those little children who thought it was "way cool"
> that birds are dinosaurs? Do you think you get them to believe
> that your concept is better than the one with which they
> are familiar (aunts, uncles, cousins, etc.) and which the disparity-measurer's
> concept so faithfully mirrors?

I doubt children of that age would have a clue about this controversy.
But they would probably believe my concept was better if they liked me
or if I gave them candy.

On Wednesday, September 7, 2022 at 12:18:55 AM UTC-4, John Harshman wrote:
> On 9/6/22 6:53 PM, Peter Nyikos wrote:
> > On Tuesday, September 6, 2022 at 8:39:56 PM UTC-4, Peter Nyikos wrote:
> >
> > ++++++++++++++++++++++++ begin included post+++++++++++++++++
> >> On 4/14/16 1:28 PM, ruben safir wrote:
> >>> On 04/14/2016 12:09 AM, John Harshman wrote:
> >>>> I'm sorry, but this is incorrect in regard to species. It would be true
> >>>> if evolution were perfectly clocklike, b
> >>>
> >>> no that is not true. time is not a measured characteristic in closest
> >>> path analysis. It might be that two species are a billion years
> >>> separated, but still closely related, and the shortest path analysis
> >>> will demonstrate that relationship.
> >>>
> >> No it won't, unless by "closely related" you just mean "similar". But
> >> what the term means in phylogenetics is "sharing a more recent common
> >> ancestry than with some third taxon", i.e. the sort of thing
> >> phylogenetic trees say. There is no particular justification for making
> >> trees that just measure similarity. (There are methods for making such
> >> trees, e.g. UPGMA, but why would you use them?)
> >>
> >> ++++++++++++++++ end of post ++++++++++++++++++++
> >>
> >> Your very first sentence is of the very essence of phenetics.
> >> And the only alternative you seriously propose is pure cladistics.
> >
> > So, what did Ruben have in mind? I believe it was a measure of
> > disparity between elements (nodes and branch tips) of an evolutionary
> > tree where the branch tips (and perhaps some nodes) are taxa.

> I suspect that you have misunderstood what he had in mind, but perhaps
> he will explain himself.

I've told him about this new thread in direct reply to him, but he hasn't replied yet.
I certainly hope he does join, because he might learn a lot about all three methods of
defining "more closely related." Back in 2016, he wrote that he was working on a
Master's degree and I wish I had known then what I know now.

> > The line segment joining successive elements is given a number estimating degree of disparity,
> > and the numbers are added together to compute the length of the path between them.

I should have ended that sentence with "the path between any two taxa on the tree,
assuming it is rooted."

> >
> > The concept of "more closely related" then could be given a whole new meaning --
> > or, rather, an old meaning but now quantified. We would be able to say that vertebrate A
> > in the following example is MUCH more closely related to B than it is to C even though
> > the LCA of A and B is strictly ancestral to the LCA of B and C.
> >
> > This is the example we had earlier: A = Tiktaalik, B= Ichthyostega, C = Bos.
> > A cladist would say B and C are equally distant from A because they are in a clade that excludes A.
> >
> > A pheneticist would agree with the disparity measurer in this example,
> > but I think it is not hard to cook up an example where evolutionary convergence
> > could make the pheneticist disagree with the disparity measurer,
> > because the latter takes a plunge down the tree and back up to do the measuring,
> > but the pheneticist hops across the tree in a beeline from one species to another.
> >
> >
> > I submit that this concept does justice to phylogeny besides being
> > exactly the sort of thing one would expect from our ordinary
> > human concept of "more closely related."

> You may think so, but I don't. If I understand your point, you are using
> patristic distance as a measure of relatedness, smaller distances being
> more close relationships.

I am unfamiliar with the concept of patristic distance. Are there any
published articles about it?

>But these distances would depend on what
> characters you used to make the measured tree, and there is no real
> objective way to choose these measures of disparity.

That is an extraordinarily difficult problem, but I think interested researchers [and they still exist]
can make a start with the tree of Equioidea in Kathleen Hunt's excellent FAQ:

http://www.talkorigins.org/faqs/horses/horse_evol.html

You can see a lineage in the tree going directly from Hyracotherium all the way to Equus,
with a genus at each node in an old-fashioned evolutionary tree.
The whole tree has a good number of side branches, and each node is a great source of characters.

Even if you believe that there isn't a single ancestor-descendant relationship
depicted there, the fossils in most cases are so nearly complete
that the named genera can be a source for a nearly complete
set of skeletal characters for the actual species that REALLY belong at the nodes.

I think a good set of disparity numbers can be worked out,
given enough time and brainstorming. It could be far advanced
by now, had the victory of the cladists in "the cladist wars" not
been a complete rout of those disagreeing with them,
and had the victors not been so intolerant of the Linnean system.

> I don't see
> patristic distance as much better for this purpose than simple pairwise distance.

That's because of your dislike of paraphyletic taxa, in solidarity with the victors.

> Further, I think it would lead to overlapping groups if used
> in classification.

Are you stuck in the 1990's, when I tried to interest people in such a system?
By the time I returned to s.b.p. in late 2010, I had come to realize that
the Linnean system was like an endangered species [though not as critically
endangered as s.b.p. itself was at the time], and I put all my efforts into
promoting a dual system of classification.

I made the analogy of the Dewey Decimal and Library of Congress systems
of classifying books, each promoted by its own libraries. But such
peaceful coexistence is not good enough for the victors: it's their way or the highway.
And you've made yourself a propagandist for them in s.b.p. and t.o.

> Velociraptor closer to Archaeopteryx than
> Archaeopteryx to Confuciusornis. Archaeopteryx closer to Confuciusornis
> than Confusiusornis to Icthyornis.

Where do you get these comparisons? You sure haven't worked out a disparity-based
distance between them.

> Icthyornis closer to Passer than
> Ichthyornis to Confuciusornis.

Really? Aren't you using the cladistic "closer to" here?

> But where do we put the line here? And
> what's closer to Hesperornis? It would seem to be out on its own.

Wrong. The beauty of the "straight down to the LCA and straight up to the other genus"
is that each pair is separated by a given distance, and there is no conflict between
pairwise distances. Want to know who is closest to Hesperornis? the one who
ends up with the smallest distance to it. Duh.

Peter Nyikos
Professor, Dept. of Mathematics -- standard disclaimer--
Univ. of South Carolina at Columbia
http://people.math.sc.edu/nyikos

On 9/7/22 2:08 PM, Peter Nyikos wrote:
> On Wednesday, September 7, 2022 at 12:18:55 AM UTC-4, John Harshman
wrote:
>> On 9/6/22 6:53 PM, Peter Nyikos wrote:
>>> On Tuesday, September 6, 2022 at 8:39:56 PM UTC-4, Peter Nyikos wrote:
>>>
>>> ++++++++++++++++++++++++ begin included post+++++++++++++++++
>>>> On 4/14/16 1:28 PM, ruben safir wrote:
>>>>> On 04/14/2016 12:09 AM, John Harshman wrote:
>>>>>> I'm sorry, but this is incorrect in regard to species. It would
be true
>>>>>> if evolution were perfectly clocklike, b
>>>>>
>>>>> no that is not true. time is not a measured characteristic in closest
>>>>> path analysis. It might be that two species are a billion years
>>>>> separated, but still closely related, and the shortest path analysis
>>>>> will demonstrate that relationship.
>>>>>
>>>> No it won't, unless by "closely related" you just mean "similar". But
>>>> what the term means in phylogenetics is "sharing a more recent common
>>>> ancestry than with some third taxon", i.e. the sort of thing
>>>> phylogenetic trees say. There is no particular justification for
making
>>>> trees that just measure similarity. (There are methods for making such
>>>> trees, e.g. UPGMA, but why would you use them?)
>>>>
>>>> ++++++++++++++++ end of post ++++++++++++++++++++
>>>>
>>>> Your very first sentence is of the very essence of phenetics.
>>>> And the only alternative you seriously propose is pure cladistics.
>>>
>>> So, what did Ruben have in mind? I believe it was a measure of
>>> disparity between elements (nodes and branch tips) of an evolutionary
>>> tree where the branch tips (and perhaps some nodes) are taxa.
>
>> I suspect that you have misunderstood what he had in mind, but perhaps
>> he will explain himself.
>
> I've told him about this new thread in direct reply to him, but he
hasn't replied yet.
> I certainly hope he does join, because he might learn a lot about all
three methods of
> defining "more closely related." Back in 2016, he wrote that he was
working on a
> Master's degree and I wish I had known then what I know now.
>
>
>>> The line segment joining successive elements is given a number
estimating degree of disparity,
>>> and the numbers are added together to compute the length of the
path between them.
>
> I should have ended that sentence with "the path between any two taxa
on the tree,
> assuming it is rooted."
I don't see the need for a rooted tree here. The path is the same length
regardless of where the tree is rooted or whether it's rooted at all.

>>> The concept of "more closely related" then could be given a whole
new meaning --
>>> or, rather, an old meaning but now quantified. We would be able to
say that vertebrate A
>>> in the following example is MUCH more closely related to B than it
is to C even though
>>> the LCA of A and B is strictly ancestral to the LCA of B and C.
>>>
>>> This is the example we had earlier: A = Tiktaalik, B= Ichthyostega,
C = Bos.
>>> A cladist would say B and C are equally distant from A because they
are in a clade that excludes A.
>>>
>>> A pheneticist would agree with the disparity measurer in this example,
>>> but I think it is not hard to cook up an example where evolutionary
convergence
>>> could make the pheneticist disagree with the disparity measurer,
>>> because the latter takes a plunge down the tree and back up to do
the measuring,
>>> but the pheneticist hops across the tree in a beeline from one
species to another.
>>>
>>>
>>> I submit that this concept does justice to phylogeny besides being
>>> exactly the sort of thing one would expect from our ordinary
>>> human concept of "more closely related."
>
>> You may think so, but I don't. If I understand your point, you are using
>> patristic distance as a measure of relatedness, smaller distances being
>> more close relationships.
>
> I am unfamiliar with the concept of patristic distance. Are there any
> published articles about it?
No entire articles that I'm aware of. But what you describe is a
patristic distance, i.e. a count of branch lengths between A and B along
the tree.

> >But these distances would depend on what
>> characters you used to make the measured tree, and there is no real
>> objective way to choose these measures of disparity.
>
> That is an extraordinarily difficult problem, but I think interested
researchers [and they still exist]
> can make a start with the tree of Equioidea in Kathleen Hunt's
excellent FAQ:
>
> http://www.talkorigins.org/faqs/horses/horse_evol.html
>
> You can see a lineage in the tree going directly from Hyracotherium
all the way to Equus,
> with a genus at each node in an old-fashioned evolutionary tree.
> The whole tree has a good number of side branches, and each node is a
great source of characters.
Good luck with that. You would have to decide which characters to
include, make that consistent across all taxa, decide how to score the
amount of difference among states within a character, consider how to
handle missing data (consider: how would you score, for example, the
shape of the 4th toe in Equus?). There are more problems, but that's
enough for now.

> Even if you believe that there isn't a single ancestor-descendant
relationship
> depicted there, the fossils in most cases are so nearly complete
> that the named genera can be a source for a nearly complete
> set of skeletal characters for the actual species that REALLY belong
at the nodes.
But are skeletal characters the only measure of disparity? You have
already biased the measure.

> I think a good set of disparity numbers can be worked out,
> given enough time and brainstorming. It could be far advanced
> by now, had the victory of the cladists in "the cladist wars" not
> been a complete rout of those disagreeing with them,
> and had the victors not been so intolerant of the Linnean system.
You take "wars" much too literally here. Nobody was harmed or even
shouted down. Systematists just became convinced of proper methodology.
And the Linnean system, ranked groups within ranked groups, was not
harmed. The decline of ranked groups was a separate matter entirely.

Now, disparity numbers can in fact be useful for many purposes. But all
they are would be distances measured over trees, and they would vary
depennding on the data. They could as well be used (and are used) with
molecular data as with morphological data. They can tell us mamy things.
But using them to define closeness of relationship is not one of those
things.

>> I don't see
>> patristic distance as much better for this purpose than simple
pairwise distance.
>
> That's because of your dislike of paraphyletic taxa, in solidarity
with the victors.
Again, it's not a matter of solidarity or of victors. It's because
paraphyletic taxa are not useful.

>> Further, I think it would lead to overlapping groups if used
>> in classification.
>
> Are you stuck in the 1990's, when I tried to interest people in such
a system?
No, I am not. But that's the implication of defining relationships that way.

> By the time I returned to s.b.p. in late 2010, I had come to realize that
> the Linnean system was like an endangered species [though not as
critically
> endangered as s.b.p. itself was at the time], and I put all my
efforts into
> promoting a dual system of classification.
You are misusing the term "Linnean system". It is not part of the
meaning of the term that it allows, requires, or forbids paraphyly. It's
silent on the question.

> I made the analogy of the Dewey Decimal and Library of Congress systems
> of classifying books, each promoted by its own libraries. But such
> peaceful coexistence is not good enough for the victors: it's their
way or the highway.
> And you've made yourself a propagandist for them in s.b.p. and t.o.
That's a pejorative characterization of a reasoned and reasonable position.

>> Velociraptor closer to Archaeopteryx than
>> Archaeopteryx to Confuciusornis. Archaeopteryx closer to Confuciusornis
>> than Confusiusornis to Icthyornis.
>
> Where do you get these comparisons? You sure haven't worked out a
disparity-based
> distance between them.
Nope, just made them up based on a general impression. Consider it a
hypothetical example if you like.

On Wednesday, September 7, 2022 at 8:52:06 PM UTC-4, John Harshman wrote:
> On 9/7/22 2:08 PM, Peter Nyikos wrote:
> > On Wednesday, September 7, 2022 at 12:18:55 AM UTC-4, John Harshman
> wrote:
> >> On 9/6/22 6:53 PM, Peter Nyikos wrote:

I've snipped the text that appears above, where you added nothing new.

Below, I'm making marked snips to get to what looks like some fundamental
miscommunications. We can talk about what I've snipped another day.

> >>> The line segment joining successive elements is given a number
> estimating degree of disparity,
> >>> and the numbers are added together to compute the length of the
> path between them.
> >
> > I should have ended that sentence with "the path between any two taxa
> on the tree,
> > assuming it is rooted."
> I don't see the need for a rooted tree here. The path is the same length
> regardless of where the tree is rooted or whether it's rooted at all.

I was using "rooted" the way graph theorists and set theorists use it,
whereby a tree lacking a root (i.e. a unique minimum element)
consists of two or more components which look like trees in the usual sense.

I take it that you mean that the tree was determined by selecting a
specimen well outside the clade to "root" the tree, as a means of
locating a unique minimum element. Correct?

> >>> The concept of "more closely related" then could be given a whole
> new meaning --
> >>> or, rather, an old meaning but now quantified. We would be able to
> say that vertebrate A
> >>> in the following example is MUCH more closely related to B than it
> is to C even though
> >>> the LCA of A and B is strictly ancestral to the LCA of B and C.
> >>>
> >>> This is the example we had earlier: A = Tiktaalik, B= Ichthyostega,
> C = Bos.
> >>> A cladist would say B and C are equally distant from A because they
> are in a clade that excludes A.
> >>>
> >>> A pheneticist would agree with the disparity measurer in this example,
> >>> but I think it is not hard to cook up an example where evolutionary
> convergence
> >>> could make the pheneticist disagree with the disparity measurer,
> >>> because the latter takes a plunge down the tree and back up to do
> the measuring,
> >>> but the pheneticist hops across the tree in a beeline from one
> species to another.

[...]

> >> If I understand your point, you are using
> >> patristic distance as a measure of relatedness, smaller distances being
> >> more close relationships.
> >
> > I am unfamiliar with the concept of patristic distance. Are there any
> > published articles about it?

> No entire articles that I'm aware of. But what you describe is a
> patristic distance, i.e. a count of branch lengths between A and B along
> the tree.

What do you mean by branch lengths? The word "count" rather than "addition"
makes me suspect that patristic distance means the number of nodes,
which completely misses the role of disparity in what I wrote.

> > >But these distances would depend on what
> >> characters you used to make the measured tree, and there is no real
> >> objective way to choose these measures of disparity.
> >
> > That is an extraordinarily difficult problem, but I think interested
> researchers [and they still exist]
> > can make a start with the tree of Equioidea in Kathleen Hunt's
> excellent FAQ:
> >
> > http://www.talkorigins.org/faqs/horses/horse_evol.html
> >
> > You can see a lineage in the tree going directly from Hyracotherium
> all the way to Equus,
> > with a genus at each node in an old-fashioned evolutionary tree.
> > The whole tree has a good number of side branches, and each node is a
> great source of characters.

> Good luck with that. You would have to decide which characters to
> include,

All known ones for each skeleton. We are talking about extinct species here,
most with nearly complete skeletons. [Of course, many would come from
distinct specimens from the same genus.]

[...]
> > Even if you believe that there isn't a single ancestor-descendant relationship
> > depicted there, the fossils in most cases are so nearly complete
> > that the named genera can be a source for a nearly complete
> > set of skeletal characters for the actual species that REALLY belong
> at the nodes.

> But are skeletal characters the only measure of disparity? You have
> already biased the measure.

Try to get your mind off extant species. Skeletons are all we have of horse fossils,
and the disparity between Equus and Dinohippus is so small that we might
as well exclude Equus to level the playing field.

[...]

> [disparity distances] can tell us mamy things.
> But using to define closeness of relationship is not one of those
> things.

Three posts up to now have featured a radically different meaning
of "closeness of relationship" than the one that is second nature to you.
Please try to stick to that one when discussing disparity distances,
if only to facilitate communication between us.

[...]
> >> Further, I think it would lead to overlapping groups if used
> >> in classification.
> >
> > Are you stuck in the 1990's, when I tried to interest people in such
> a system?
> No, I am not. But that's the implication of defining relationships that way.

I completely disagree. Please try to reveal your line of reasoning.

> > By the time I returned to s.b.p. in late 2010, I had come to realize that
> > the Linnean system was like an endangered species [though not as critically
> > endangered as s.b.p. itself was at the time], and I put all my efforts into
> > promoting a dual system of classification.

> You are misusing the term "Linnean system". It is not part of the
> meaning of the term that it allows, requires, or forbids paraphyly. It's
> silent on the question.

It's rife with paraphyletic taxa. Reptilia, Amphibia, Osteichthyes, Agnatha,
just to name the classes involved. And then there are what are
derisively termed "garbage taxa": Cotylosauria, Thecodonta, Insectivora, and Condylartha.

If Linnaeus had been into subclasses and infraclasses,
he might have defined Theria as a subclass of Mammalia,
split into the following infraclasses: Marsupilia, Cetacea, Chiroptera
and Terraplacentalia. After all, he was only using extant animals,
and the last named would have been a paraphyletic group.

[...]

> >> Velociraptor closer to Archaeopteryx than
> >> Archaeopteryx to Confuciusornis. Archaeopteryx closer to Confuciusornis
> >> than Confusiusornis to Icthyornis.
> >
> > Where do you get these comparisons? You sure haven't worked out a
> disparity-based
> > distance between them.

> Nope, just made them up based on a general impression. Consider it a
> hypothetical example if you like.

So far, so good. But the following is debatable:

> > > Icthyornis closer to Passer than
> >> Ichthyornis to Confuciusornis.
> >
> > Really? Aren't you using the cladistic "closer to" here?
> No. Again, a general impression of morphology.

Ichthyornis had long jaws with teeth, Confuciusornis
had a "birdlike" beak, and the rest of the skull [1] looks
more like that of a pigeon than that of Ichthyornis.
What features are you looking at to compensate for this?

[1] I'm looking at the skull in the 3rd edition of Benton's _Vertebrate_Paleontology_

[...]

> >> what's closer to Hesperornis? It would seem to be out on its own.
> >
> > Wrong. The beauty of the "straight down to the LCA and straight up to the other genus"
> > is that each pair is separated by a given distance, and there is no conflict between
> > pairwise distances. Want to know who is closest to Hesperornis? the one who
> > ends up with the smallest distance to it. Duh.

> The smallest patristic (along the tree) distance, one supposes. That
> would presumably be, one supposes, Ichthyornis of those listed. But
> though Hesperornis would be closer to Ichthyornis than to Passer, would
> Ichthyornis be closer to Passer than to Hesperornis?

You are talking about an everyday property of distance. If you live in
California, you are closer to Hawaii than you are to Newfoundland.
But Newfoundland is closer to you than it is to Hawaii.

> Relationships, based on disparity, would appear not to be transitive.

What would transitivity mean in this context? The salient property
of disparity distance is symmetry: taxon A is exactly as close to
taxon B as taxon B is to taxon A.

On 9/8/22 10:11 AM, Peter Nyikos wrote:
> On Wednesday, September 7, 2022 at 8:52:06 PM UTC-4, John Harshman wrote:
>> On 9/7/22 2:08 PM, Peter Nyikos wrote:
>>> On Wednesday, September 7, 2022 at 12:18:55 AM UTC-4, John Harshman
>> wrote:
>>>> On 9/6/22 6:53 PM, Peter Nyikos wrote:
>
> I've snipped the text that appears above, where you added nothing new.
>
> Below, I'm making marked snips to get to what looks like some fundamental
> miscommunications. We can talk about what I've snipped another day.
>
>
>>>>> The line segment joining successive elements is given a number
>> estimating degree of disparity,
>>>>> and the numbers are added together to compute the length of the
>> path between them.
>>>
>>> I should have ended that sentence with "the path between any two taxa
>> on the tree,
>>> assuming it is rooted."
>> I don't see the need for a rooted tree here. The path is the same length
>> regardless of where the tree is rooted or whether it's rooted at all.
>
> I was using "rooted" the way graph theorists and set theorists use it,
> whereby a tree lacking a root (i.e. a unique minimum element)
> consists of two or more components which look like trees in the usual
sense.
>
> I take it that you mean that the tree was determined by selecting a
> specimen well outside the clade to "root" the tree, as a means of
> locating a unique minimum element. Correct?

No. "Rooted" means the same thing in phylogenetics as you describe as
your usage, the existence of a (conventially shown) "lowermost" point
dividing the tree into two subtrees. Sometimes this is done by declaring
an outgroup, but there are other ways, and it doesn't matter to the
present question how rooting is done. What matters is that the position
of the root or its absence has no effect on the lengths of paths through
the tree. Would you not agree?

>>>>> The concept of "more closely related" then could be given a whole
>> new meaning --
>>>>> or, rather, an old meaning but now quantified. We would be able to
>> say that vertebrate A
>>>>> in the following example is MUCH more closely related to B than it
>> is to C even though
>>>>> the LCA of A and B is strictly ancestral to the LCA of B and C.
>>>>>
>>>>> This is the example we had earlier: A = Tiktaalik, B= Ichthyostega,
>> C = Bos.
>>>>> A cladist would say B and C are equally distant from A because they
>> are in a clade that excludes A.
>>>>>
>>>>> A pheneticist would agree with the disparity measurer in this
example,
>>>>> but I think it is not hard to cook up an example where evolutionary
>> convergence
>>>>> could make the pheneticist disagree with the disparity measurer,
>>>>> because the latter takes a plunge down the tree and back up to do
>> the measuring,
>>>>> but the pheneticist hops across the tree in a beeline from one
>> species to another.
>
> [...]
>
>>>> If I understand your point, you are using
>>>> patristic distance as a measure of relatedness, smaller distances
being
>>>> more close relationships.
>>>
>>> I am unfamiliar with the concept of patristic distance. Are there any
>>> published articles about it?
>
>> No entire articles that I'm aware of. But what you describe is a
>> patristic distance, i.e. a count of branch lengths between A and B along
>> the tree.
>
> What do you mean by branch lengths? The word "count" rather than
"addition"
> makes me suspect that patristic distance means the number of nodes,
> which completely misses the role of disparity in what I wrote.

No, that's not what it means. Branch lengths are expressed as the
magnitude of inferred change between nodes, same as what you seem to be
saying. Adding up the total length of branches between taxa gives you a
patristic distance. The number of nodes traversed is irrelevant.

>>>> But these distances would depend on what
>>>> characters you used to make the measured tree, and there is no real
>>>> objective way to choose these measures of disparity.
>>>
>>> That is an extraordinarily difficult problem, but I think interested
>> researchers [and they still exist]
>>> can make a start with the tree of Equioidea in Kathleen Hunt's
>> excellent FAQ:
>>>
>>> http://www.talkorigins.org/faqs/horses/horse_evol.html
>>>
>>> You can see a lineage in the tree going directly from Hyracotherium
>> all the way to Equus,
>>> with a genus at each node in an old-fashioned evolutionary tree.
>>> The whole tree has a good number of side branches, and each node is a
>> great source of characters.
>
>> Good luck with that. You would have to decide which characters to
>> include,
>
> All known ones for each skeleton. We are talking about extinct
species here,
> most with nearly complete skeletons. [Of course, many would come from
> distinct specimens from the same genus.]

Most with nearly complete skeletons? Have you ever actually looked at a
paleontological data set? They're full of missing data. And as for
"distinct specimens from the same genus", you should know that, for
example, around half of all non-avian dinosaur genera are known from
single specimens only.

> [...]
>>> Even if you believe that there isn't a single ancestor-descendant
relationship
>>> depicted there, the fossils in most cases are so nearly complete
>>> that the named genera can be a source for a nearly complete
>>> set of skeletal characters for the actual species that REALLY belong
>> at the nodes.
>
>
>> But are skeletal characters the only measure of disparity? You have
>> already biased the measure.
>
> Try to get your mind off extant species. Skeletons are all we have of
horse fossils,
> and the disparity between Equus and Dinohippus is so small that we might
> as well exclude Equus to level the playing field.

All that does is point to the comparative lack of essential data, making
measures dependent on that data less than ideal.

>> [disparity distances] can tell us mamy things.
>> But using to define closeness of relationship is not one of those
>> things.
>
> Three posts up to now have featured a radically different meaning
> of "closeness of relationship" than the one that is second nature to you.
> Please try to stick to that one when discussing disparity distances,
> if only to facilitate communication between us.

My point is still valid with respect to your favored definition, and in
fact that's what I was thinking of here, but it does contribute another
reason why your definition isn't useful.

>>>> Further, I think it would lead to overlapping groups if used
>>>> in classification.
>>>
>>> Are you stuck in the 1990's, when I tried to interest people in such
>> a system?
>> No, I am not. But that's the implication of defining relationships
that way.
>
> I completely disagree. Please try to reveal your line of reasoning.

I believe I did, below. In short, attempts to use these "relationships"
to define uniquely supported, paraphyletic groups will necessarily fail
and will imply overlapping groups.

>>> By the time I returned to s.b.p. in late 2010, I had come to
realize that
>>> the Linnean system was like an endangered species [though not as
critically
>>> endangered as s.b.p. itself was at the time], and I put all my
efforts into
>>> promoting a dual system of classification.
>
>> You are misusing the term "Linnean system". It is not part of the
>> meaning of the term that it allows, requires, or forbids paraphyly. It's
>> silent on the question.
>
> It's rife with paraphyletic taxa. Reptilia, Amphibia, Osteichthyes,
Agnatha,
> just to name the classes involved. And then there are what are
> derisively termed "garbage taxa": Cotylosauria, Thecodonta,
Insectivora, and Condylartha.

Again, you misuse the term "Linnean system". It's not about any
particular classification, not the one you remember from your youth, not
any other. It's about a style of classification, of ranked groups within
ranked groups.

This is a change of pace: A lighthearted (for me, anyway) treatment of John's reaction to a PS
of mine at the end of a reply he did to me here:

On Wednesday, September 7, 2022 at 12:18:55 AM UTC-4, John Harshman wrote:

> On 9/6/22 6:53 PM, Peter Nyikos wrote:

> > PS Remember those little children who thought it was "way cool"
> > that birds are dinosaurs? Do you think you [can] get them to believe
> > that your concept is better than the one with which they
> > are familiar (aunts, uncles, cousins, etc.) and which the disparity-measurer's
> > concept so faithfully mirrors?

> I doubt children of that age would have a clue about this controversy.
> But they would probably believe my concept was better if they liked me
> or if I gave them candy.

That depends on whether you invite me twice to talk to them.
The first time, I would try to establish rapport with them by talking about
some neat things that occurred on a pair of fossil hunts in Wyoming on which
I participated a little over a year ago.

The second time, I would get around to asking them if any of them
had one or more great-grandparents still alive, after matching any
candy bribe of yours. If it's a class of ten or more, there is an excellent chance
that at least one hand would go up, because nowadays a goodly
fraction of people live to that age.

With twenty, there might arise a longish bidding as to how many were still alive.
Once that had settled down, I would ask them whether any of those great-grandparents
had any brothers or sisters. And if any hands are raised [some might not know the answers]
I would ask the raisers whether they had any brothers or sisters.

If yes, I would turn to one of them and say,

Did you know that Dr. John's [or whatever they call you] way saying how different kinds of
animals are related is like him saying, "Your great-grandmother is more closely related
to your brother than she is to her own brother."

Once this has sunk in and the expressions of astonishment subsided,
I will say, "Now Dr. John will explain to you why this is the best way
to talk about how different kinds of animals are related."

My suggestion for you to save face is to begin by saying that we use the
same words for very different things. The one that comes to mind
is "pipe". You could make a little joke about the absent-minded
professor who tried to smoke a pipe that gas runs through,
and to run water through the pipe that he smokes. But you could
probably do better than that, given all this lead time.

And then, you could explain that the expression "more closely related"
works the same way.

But you'd still have a big problem explaining why it is the RIGHT
meaning for dealing with animal species.

TGIF.

Peter Nyikos
Professor, Dept. of Mathematics -- standard disclaimer--
University of So. Carolina in Columbia
http://people.math.sc.edu/nyikos

On 9/9/22 4:40 PM, Peter Nyikos wrote:
> This is a change of pace: A lighthearted (for me, anyway) treatment of John's reaction to a PS
> of mine at the end of a reply he did to me here:
>
> On Wednesday, September 7, 2022 at 12:18:55 AM UTC-4, John Harshman wrote:
>
>> On 9/6/22 6:53 PM, Peter Nyikos wrote:
>
>>> PS Remember those little children who thought it was "way cool"
>>> that birds are dinosaurs? Do you think you [can] get them to believe
>>> that your concept is better than the one with which they
>>> are familiar (aunts, uncles, cousins, etc.) and which the disparity-measurer's
>>> concept so faithfully mirrors?
>
>> I doubt children of that age would have a clue about this controversy.
>> But they would probably believe my concept was better if they liked me
>> or if I gave them candy.
>
> That depends on whether you invite me twice to talk to them.
> The first time, I would try to establish rapport with them by talking about
> some neat things that occurred on a pair of fossil hunts in Wyoming on which
> I participated a little over a year ago.
>
> The second time, I would get around to asking them if any of them
> had one or more great-grandparents still alive, after matching any
> candy bribe of yours. If it's a class of ten or more, there is an excellent chance
> that at least one hand would go up, because nowadays a goodly
> fraction of people live to that age.
>
> With twenty, there might arise a longish bidding as to how many were still alive.
> Once that had settled down, I would ask them whether any of those great-grandparents
> had any brothers or sisters. And if any hands are raised [some might not know the answers]
> I would ask the raisers whether they had any brothers or sisters.
>
> If yes, I would turn to one of them and say,
>
> Did you know that Dr. John's [or whatever they call you] way saying how different kinds of
> animals are related is like him saying, "Your great-grandmother is more closely related
> to your brother than she is to her own brother."
>
> Once this has sunk in and the expressions of astonishment subsided,
> I will say, "Now Dr. John will explain to you why this is the best way
> to talk about how different kinds of animals are related."
>
> My suggestion for you to save face is to begin by saying that we use the
> same words for very different things. The one that comes to mind
> is "pipe". You could make a little joke about the absent-minded
> professor who tried to smoke a pipe that gas runs through,
> and to run water through the pipe that he smokes. But you could
> probably do better than that, given all this lead time.
>
>
> And then, you could explain that the expression "more closely related"
> works the same way.
>
> But you'd still have a big problem explaining why it is the RIGHT
> meaning for dealing with animal species.

All I can say is that you would be ill-advised to quite your day job.

On 9/5/22 14:42, Peter Nyikos wrote:
> The heart of cladistics is found in the phylogenetic trees, which put all
> organisms at the branch tips.

nodes not tips

On 9/10/22 2:28 PM, Popping Mad wrote:
> On 9/5/22 14:42, Peter Nyikos wrote:
>> The heart of cladistics is found in the phylogenetic trees, which put all
>> organisms at the branch tips.
>
>
> nodes not tips

"Branch tips" here refers to terminal nodes. The intended meaning is
that organisms are placed at terminal nodes, not internal nodes. In case
you are still unclear on meanings, a terminal node connects to a single
branch, while an internal node connects to three or more branches,
exactly three on a fully resolved tree.

So the claim is true.

On Saturday, September 10, 2022 at 2:29:09 PM UTC-7, Popping Mad wrote:
> On 9/5/22 14:42, Peter Nyikos wrote:
> > The heart of cladistics is found in the phylogenetic trees, which put all
> > organisms at the branch tips.
> nodes not tips

Yea, who needs em anyway.

On 9/6/22 21:53, Peter Nyikos wrote:
> The line segment joining successive elements is given a number estimating degree of disparity,
> and the numbers are added together to compute the length of the path between them.
>

This was correct because I was trying to write a computer program or
algorithm that would take a CT scan of a fossil while still in the rock,
and attempt to find the placement of that fossil within the evolutionary
tree based on common attributes. In this case, the age of the fossil
would either be ignored or considered one attribute among many.

So the fundamental question is can two species, a billion years
separated, be closely related, or how closely related and how do we
define relatedness. Ideally, we know two individuals are closely
related because they share genetic material. But I don't even know what
that means and admittedly, I know less now than I did 20 years ago as an
undergrad. The older I get, the stupider I become.

Two siblings, in theory, share about 33%-50% +- random variation, of
their genetic material, since they have the same parents. It is
possible, although HIGHLY unlikely, that they share zero DNA. Or they
are some form of a twin and share 100% of their DNA.

Fine. Now Random Chimps and Humans seem to have a closer relationship
than siblings. They share 98.6% of their DNA, or so I see reported? So
in the context of siblings, I am not at all sure what that means.

Aside from that, we don't have genetic material for most species. All
we have is a fraction of skeletal material or a soft tissue imprint.

So for evidence we have
A) incomplete phenotyped characteristics
B) an imperfect time line or age of a fossil

And that is it.

The tools we have to analyze this include a few algorithms and to attach
your ego to any of them, given the limits of the data, is a colossal
error, IMO. Every way we analyze something gives us a different window
to view our specimens and their relationships to each other. These
algorithms themselves are grounded into a few evolving sciences and
mathematics which includes the study of Genetics, Geography, Cosmetology
and Astronomy, Biological computational statistics, and more to name a few.

And then we have the problem with definitions and the understanding of
concepts. Sometimes one has to step back and ask, what exactly am I
trying to learn here?

Are two species closely related despite 300 million years of evolution
if they genetics have been largely conserved other that time? There is
a molecular clock but the functional expression of genes might well be
well preserved in two species separated by by a long period of time.
Hydrothermal vent microbial organism might be more closely related to
their ancient ancestors from a billion years ago than a Moerithriun to
an African Elephant, let alone an Asian Elephant to most species in
Paenungulata. It is suggested that a Hyrax is more basal and an African
elephant, and I would ask, "how so since they are both existing species
and Elephants didn't spring from a modern Hyrax".

So the language itself, here, gets very ambiguous which fuels
disagreements and trolls alike.

And added to this amphibological language a new problem seems to have
emerged recently with core elements of evolutionary theory. It has been
a conceit that a species evolves traits to conform with its environment
through natural selection. Therefor, in theory at least, any individual
could prove to be a node to a new species if they have a mutation that
gives its linage an advantage. There has been shown to be a HUGE
conceit as we have learned from the study of the evolution of Homo and
other species. It has come to be understood that mutations that give
advantages can and do spring up multiple times, even among closely
related species... the same mutation. And then we share DNA material
outside of reproduction. And then mutation will fade if unneeded and
spring up again. Instead of a nice tree, evolution might well look more
like a river delta. Or a suppressed shared gene might become active in
two different lineages, or two species might interbreed?

Everything wants to become a crab...

None of the above is my own original ideas. They all come from reading
scientific literature both in the popular press and in peer reviewed
journals.

So are all Ceratopsidae from the same common ancestor? Maybe, likely,
not definitely.

The last common ancestor of Pachyrhinosaurus... that might be true. But
as we get more evidence of this transition, we might find that more than
once species of closely related dinosaurs evolved similarly, utilizing
similar mutations. Finding a Least common ancestor might be unknowable.
But that might not make this method of defining families and orders and
genus, any less useful. We need to use some definition.

No maybe all these ambiguities have been solved in Paleontology, I
don't know. But I do know that over my lifetime many streams of thought
within the science has come and gone and evolved. I wouldn't take a
theological attachment to any system for the analysis of the
evolutionary procession of species. Nature tends to surprise us and
these are all just tools of logic to help us understand the ancient past
and the nature of biological systems.

> The concept of "more closely related" then could be given a whole new meaning --
> or, rather, an old meaning but now quantified. We would be able to say that vertebrate A
> in the following example is MUCH more closely related to B than it is to C even though
> the LCA of A and B is strictly ancestral to the LCA of B and C.

I have no idea what this means, FWIW.

On 9/7/22 00:18, John Harshman wrote:
> I don't see patristic distance as much better for this purpose than
> simple pairwise distance.

FWIW - I have no idea how Christian fathers relate to distances in a
graph but....

ignoring the cost values between nodes, ie treating them all equal,will
get a wrong result every time.

Distant is a measured of a cost and that has to be defined using the
same group of attributes between two nodes to be of any use at all.
This is a mathematical fact and I am not really up to debate it. To
remove cost from the measurement of the distance between nodes gives a
senseless answer.