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

Re: PARSIMONY!!!



Scott A Hartman wrote:
>Lately parsimony has become the horse to beat on the list.
        Lately? Been here long?

>Parsimony is often stated to mean that all things being
>equal, the simplest explanation tends to be right.
        I prefer "the explanation which best fits the data with the least
number of unsubstantiated assumptions and the fewest  unsupportable
hypotheses is most likely to correct" or is closest to the truth, or whatever.

>In that case, the hypothesis that needs the
>fewest leaps of faith is the most parsimonious.
        Agreed.

>This is not really a theory, it's a way of evaluating the evidence.
        Well, a criterion for choosing among competeing hypotheses intended
to explain the evidence.

>And of course, once all of
>the data is in (which, sadly, it never will be), then the correct answer
>HAS to be the simplest one (which, presumably, would also be the _only_
>one).
        True, but perhaps self-evident. The "correct answer" is presumably
one of the elements of the set of *all* data, but we get the idea.

>So really, how can evolution _not_ be parsimonious?
        That depends on what you mean by evolution "being" parsimonious. No
one has ever established, to my satisfaction, how evolution could "be"
parsimonoius, or what process within evolution would be ascribed (by anyone)
with the responsibility of ensuring parsimony.
        Generally, I assume that people who suggest that "evolution is not
parsimonious" are implying that rigorous application of the principle of
parsimony to phylogenetic reconstruction inherently implies the assumption
that homoplaisy is minimal. However, as pointed out by Sereno in _Dinosaur
Systematics_, this is not the case at all. The application of parsimony in
this context simply minimizes assumed homoplaisies.
        There is nothing wrong with this. Sure, the answer you get may turn
out later to not be the best one. But how did you figure this out? You
gathered more data, ran it through, and discovered that the original
hypothesis requires the assumption of more homoplaisies than the new one. It
is a simple, straightforward, and reproducable formulation of phylogenetic
argumentation and not, as you point out, a theory of evolution.
        That said, I have heard (from Chris Brochu) that some schools of
phylogenetic reconstruction use parsimony as a "model" of evolution. I don't
understand this, and thus I cannot comment on it.

>I suspect that
>what the issue is here isn't about parsimony per se, but rather about the
>application of parsimony to a data set as small as ours is in
>paleontology 
        This is science, you do the best with what you have, and you
forumlate *supportable* hypotheses. We should not abandon theoretically
sound methodology simply because we feel our dataset is small. Paleontology
(especially dinosaur paleontology) does suffer from a great paucity of
physical evidence. Under such circumstances, it becomes even *more*
important to cling to methodologies which will strengthen our conclusions
against the possibility of conceptual or methodological flaws.

>(after all, look at how any times in the past the "most
>parsimoniest" explanation has been overturned by new data).
        This is, of course, not only to be expected but eagerly anticipated.
Science is, in some ways, like the (idealized) courtroom aspect of the legal
profession: we seek to approach the truth through rigorous exploration,
examination, and debate. The extensive debate which has in the past
characterized discussions of many aspects of science (plate tectonics,
natural selection and evolution, "granitization", and cladistics to name a
few near and dear to my heart) has only served to strengthen our
understanding of the natural world.

>So what we need is a rule to help filter out some of the arbitrariness from
>the process of selecting from competing hypothesis.
        Well summarized!

>P.S. Could you imagine if, instead, we decided to select the most
>complicated of the competing hypothesis...?
        Yes, I've seen it happen on this list... ;)
------------------------------------------------------
Dwight Stewart wrote:

>       The point about data size is important, but not necessarily a
>defining issue, unless later data deviates significantly from existing data
>and/or the interpretation of extant data stretches [...]
>One big cause of this can [b]e hidden variables.  Another special cause can be
>unexpected interactions.  Simplicity is not always so simple to arrive at.
        "Simplicity" is not necessarily the key, we need *relative*
simplicity. And yes, there may be many things that we are missing in our
view of life, and specifically phylogeny. However, the data we have are
*all* we have, and it simply isn't possible to incorporate all of the
possibilities into our hypotheses, especially when we have no evidence for
them. So, what then is the answer? Hypothesize on what you know, not what
you don't. => Parsimony.

>       The solution is (boringly) more data.  More data & more data.
        In paleontology, new data is seldom "boring". :)

        Wagner
----------------------------------------------------------------------------
    Jonathan R. Wagner, Dept. of Geosciences, TTU, Lubbock, TX 79409-1053
"Only those whose life is short can... believe that love is forever"-Lorien