Positivism vs Realism

[Tom DiBenedetto]

James Francis Lyons-Weiler wrote:

Thomas Pape wrote:
>> How can an estimate which requires MORE AD-HOC explanations
>> ever be superior?

>        When there is evidence that the explanation needed is more
>        complex than parsimony would allow.

More precisely, when there is evidence that the explanation needed is
more complex than simple descent. And of course, the test of
congruence is exactly the procedure by which you get this evidence.
Which means that your conclusion will require more ad hocs than you
began with (you began with none - just the assertion that all of your
matrix was correct). But there is no reason to adopt more ad hocs
than the test indicates.

>        Parsimony clearly assumes model of
>        lots of descent with little modification, whether it
>        is used as a summary, a criterion, or test (the last
>        bit there for completeness).

gee, like maybe we should assume more modification even when we dont
see it, eh?
Hey, anything COULD have happended! Cladistics, like science in
general, is focussed on explaining the data we discover, not any set
of data which could possibly be imagined.
The hypothesis tested is that the distributions coded in the matrix
are the result of descent. Logical violations of that are revealed as
homoplasy.

>        ....our ideas
>        on the expected prevalence of homoplasy, its independence and
>        its distribution have everything to do with whether we can be (I
>        mean as individuals) confident about the use of the term "best".

As if these ideas have any justification themselves! Where do they
come from? What universal priniples allow you to claim that more
homoplasy exists than what we find in a parsimony tree? What gives
you the predictive power to claim that parsimony is underestimating
homoplasy, or that  a particular synapomorphy is really an artifact
of something?

>       There are numerous
>        types of analyses that take precedence over parsimony
>        because they can predict, for a specific case, that parsimony
>        is likely to fail.

But it can be, and has been, demonstrated that a prediction of (e.g.)
an unavoidable long branch problem, can be specious.
How exactly do you propose to derive trees?