RE: Kerberosaurus manakini

[Mike Milbocker <mmilbocker@psdllc.com>]

Milbocker:

The application of systematics goes astray, in my
> opinion, when people use it to "prove" other competing theories,
hypotheses
> false. Bottom line, systematics is merely a data ordering tool made
> quasi-scientific by the often misunderstood principle of parsimony.

david.marjanovic@gmx.at:

"Data ordering tool"? The goal is not at all to order data. The goal is to
produce a phylogenetic _hypothesis_. Hypothesis emphasized because a
traditional genealogical tree is a speculation that is often not
falsifiable -- cladograms are always falsifiable.

Milbocker:

The goal is to order data based on a principle of parsimony. The hypothesis
is that the principle of parsimony results in an association of traits that
accurately reproduces what happened in evolution. The hypothesis can be
refuted with stratigraphic evidence, weighting certain features over others,
additional fossil evidence, etc. A hypothesis based on say a convergence
model would "subjectively" be excluded by cladistics as being less fit, even
though convergent evolution is demonstrated abundantly in nature. There are
other "forces", other than parsimony that could be used to form hypotheses
about evolution that are valid ways of looking at empiricle evidence. For
example, an alternative hypothesis may be a cellular automaton model, ala
Steve Wolfram or John Conway, where parsimony is placed on the governing
rule set and not on the associations between evolved features. In fact, from
the cellular automata point of view evolutionary convergence is to be
expected for most rules. Universal CA that create endlessly varied entities,
as evolution without convergence would presuppose, are very rare. One could
also apply a geometrical hypothesis by starting with a basic form
(outgroup)and a set of 3-D transformation equations which are applied to the
characters in succession to create a transformation on the animal as a
whole. In this case, two Taxa are to be associated through a convolution of
mathematical transformations which may well negate current coding based on
part traits. Since 3 applications of one transform may yield the same state
as two applications of another, the parsimony rule would be applied to the
number of mathematical transformations on the whole animal. Although,
Occam's Razor is being applied, it's entirely different from current state
coding. So when we talk about phylogenetics, I assume we're talking about
current forms of it - and they are not unique. These alternative forms, such
as the geometric one proposed above, may very well be encapsulated in
someone's "old-fashioned comparison analysis". To say this analysis is
inherently flawed, or less able to arrive at a truth is underestimating the
capacity of the human mind. To say the conclusions arrived at are
coincidental is equally unfair. After all, I would be willing to wager that
crude forms of cladistics were operational in many "old-fashioned" analyses
before computers refined and popularized the approach.