Weights

Tom DiBenedetto tdib at UMICH.EDU
Tue Mar 4 12:58:19 CST 1997


James, I think we will have to make some efforts to keep the
discussion from exploding out into many different issues,,,,I pick
out a few points here which are directly relevant to the previous
exchanges,,,the rest I will try to deal with later.

 re: "implicit"
My objection to your use of "implicit" is based on the connotation
that probabilism is an inherent, if unspoken factor in the approach
that I use. It is not. You simply continue to fail to appreciate that
we have different conceptual frameworks. My understanding of this
dispute is as follows:
1) Probabilism is not synonymous with uncertainty; probabilism is an
apporach to dealing with uncertainty. It is not the only such
approach.
2) We agree that we have, in any study, a collection of evidence that
is either true or is false, and we need a method for making
discriminatory judgements.
3) One approach is to test the evidence in light of the expectations
which our theories make on such evidence, and to accept or reject the
evidence on the basis of such tests. This is my approach. There is a
series of such tests; congruence being the last.
4) Uncertainty, as well as truth and falsity, can be considered as a
0 or a 1. I advance hypotheses (from wherever) hence provisionally
according them a 1. The tests either allow the hypothesis to survive,
or indicate that they are false (a zero).
5) Probabilism is a different approach which seeks to quantify some
measure of certainty or uncertainty, to fill in the space between 0
or 1 with the hope that lower probabilites will end up being assigned
to those phenomena which really deserve a zero, and higher
probabilities to those which really deserve a 1, such that in the end
the truth will be arrived at. The fact that one can enter into such a
procedure does not mean that it is an inherent part of a different
procedure. IF the probabilities had any real-world meaning relative
to the phenomena (ontologically), then you miight have a point. They
do not. The probabilities are an artificial guide which you construct
to help you make judgements. If I decline the offer to make my
judgements in such a way, I am not ignoring any essential reality, I
am simply rejecting your method for dealing with uncertainty. I think
my methods deal with it better.

>In case I've not been clear enough, what I mean to say is that
>probability is involved in the nature and essence of "unweighted"
>parsimony, and although it has been revealed, it in not often expressed,
>and not fully developed.

You have been clear enough. I still think you wrong. Probabilities do
not inhere in reality, they do not address the question of whether
the transformation actually occurred; they simply measure the extent
to which it would be "usual" were it actually to have occurred. I do
not find this to be a consideration which should influence our
methods of discovering what actually did and did not happen in
particular cases.

>       Given our shared, enlightened understanding of what I mean
>       by "implicit", I don't expect ANYONE to "accept" the
>       assumptions.  They are not mine, by the way.  They simply
>       are.  If an assumption plays a role, you can't change
>       that by claiming that it doesn't.  Each of the assumptions
>       I listed poses a very real threat to accuracy if they
>       are assumed away and nevertheless obtain (in specific
>       instances).

My my, sounds like a downright religous insistance on your
assumptions mixed with an inability to concieve that others may
approach a problem differently than you do (and know what they are
doing). Given that it is your approach which is the non-traditional
one in systemtics, I think you should take a bit more seriously and
open mindedly the boat which you are jumping off of.

>       What probabilities are you referring to?  I;m certain that
>       the probabilities that you calculate (implicitly (ie. 1 and 0)
>       may be useful to increase phylogenetic accuracy, but the problem
>       is that you can't tell when you're on the mark with equal
>       probabilities.

Living in a 1 or 0 world is not probabilism. The 1 and 0 are
reflective of truth and falsity, which impinges upon us as
uncertainty. We all share 1 and 0. Some of us devise tests to
discriminate between the two. Some of you try to quantify the middle
ground. That is probabilism. It is not a necessary strategy, for the
*ontology of the phenomena are 1 or 0*. Probabilism is a heuristic to
help you go one way or the other. It is not implicit in any other
strategy to choose between the two.

>> The real transformations are inherently equal.
>
>This is simply an assumption.  You claim that you CONSIDER them to be
>equal, but you also don't KNOW that they are equal.  This contradiction
>hobbles phylogenetic inference via parsimony alone.

Yes I do know they are equal. An evolutionary event, the rise of an
apomorphy, which has left its trace in the distribution of that
feature in a group of taxa, indicates a particular set of
phylogenetic relationships. If it is a true homology, it is true,,a
fact, plain and simple. There is no sense in which it can possibly do
more or less than what it does; indicate the relationship between
certain taxa.

> Wouldn't it be nice if one could assess by examining character covariation whether >or not there was a local erosion of the fidelty of the information content
>contained in differences among character states (i.e., pattern in the
>distribution of character states)?

I dont see the relevance of that to the question of whether a
particular transformation occurred or not or to the issue of
weighting.

I'll stop here before we get too deeply into the particualrs of your
personal approach, for they are a bit off the main mark. Since they
are however, crucial to what you have been talking about, perhaps you
could grant us all a very brief summary of the concepts behind your
RASA approach. I am sure that most here might have heard of it, but
might not be sure of exactly the conceptual context. An explanation
which doesnt delve into the statistics might be broadly
understandable here, and may help us (I guess especially me) to avoid
talking past you.
I will return to the concerns you raise in the rest of this message
as soon as I can.




More information about the Taxacom mailing list