[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: DISSENT GOOD, CONFORMITY BAD



--Orig. Message-- From: chris brochu <cbrochu@fmppr.fmnh.org>
Date: Sun, 28 Mar 1999 12:06:57 -0600


>John V Jackson wrote:
>
>>>[Matt T:] "It is hard to deny the power of the theropod hypothesis, ..."
>
>> People often cast the issue in the form of "theropod" vs "non-theropod",
>> but there is more than one theropod theory, and most are mutually
>> exclusive.  In some ways, all the wrong theories have more in common
>>with each other than with the truth.
>
>But ARE these "mutually exclusive?"  There are quantitative ways of
>addressing that question.  The days when we could simply step back and
>say "Wow - those trees are, like, really different and stuff" are long
over.

The quantitative procedures for comparing them will depend entirely on the
processes that created them.  The statistical algorithms used will have
confidences associable with each feature of a tree or clad, but what if the
trees were generated in some other way?

In my way for example, the TRANSPECTION method (I haven't come up with the
basis for that acronym yet!) I think about various aspects and significances
of the known features, plump for the most reliable ones whose significances
we feel we can understand, and then build a perfect tree without any
homoplasies.

The tree may be low on detailed resolution, but then again, once we have
segmented the overall structure we can deal with the smaller chunks more
easily.  Features which may be homoplastic across the whole range may not be
locally.  It may be hard to associate confidences with heuristic processes,
and they may not be considered very respectable by some, but they are
fashionable in some areas - natural vision for example!


>It is generally thought that highly congruent results from disparate
>sources suggest an approximation of truth.  But, we don't necessarily
>expect different data sets to yield precisely identical signals -

Rough clusters around a target are to be expected, but though precision is
not expected from every shot, the average of all the shots might be expected
to have some significance if the weapon is "true enough".  No-one ever gets
near my bullseye though!


[snip]
>So we are forced to ask ourselves "how different must two
>trees be to really be different?," and an extensive literature has
>emerged about it.

They will be applicable to their own methods of tree generation - but not
mine.


>This started out from the perceived "molecules versus
>morphology" debate, and it's become clear from these approaches that
>molecules and morphology actually produce highly congruent results most
>of the time.

For them to produce highly congruent results they must be working on extant
lineages, and probably with numerous individuals densely distributed in time
and phylogeny.  This doesn't apply at all to the dinobird topic; also
molecules and morphology did not agree on humans-chimp-gorillas - my other
general area of interest.  The tree's I tend to look at don't enjoy the
blessing of convergent mol. & morph.

[snip]
>And all [the trees] agree that birds are derived
>coelurosaurs - topology within Coelurosauria is labile, but given the
>different taxon samplings and strategies for dealing with polymorphism
>among different analyses, this isn't surprising.

Though none I think do 2F embrace - but hey - what the heck - yes - it's all
much the same thing really... There will come a morning for each of us
when we'll wake up and say "How much of a problem would it be if
Archaeopteryx were just a smidgeon more basal...?" and it will be no big
deal - after all the important thing is that it's still the theropod theory!

Thanks for the refs Chris - I can't wait to get amongst them - though I'll
take in your recent "Nature" one first.


John V Jackson    (en "krul")

"For Harry, Euroland, St. George and the theropod theory (though not
necessarily in that order)"

Dare you visit...
http://www.geocities.com/CapeCanaveral/Hall/2099/DinoKabin.html
Dare you not?