Re: The Rashi Descent

Robert E. Heyman <robeh@...>

Jacob,

No, that is not how the argument goes. The argument goes, you have 2^30
ancestors, and they are all Ashkenazic Jews (this is where the flaw is -
how would one know this?). But let's assume you know this for a fact.
This is many more than the number of Ashkenazic Jews who lived during that
time. You pick one person, and compute the probability that you are *not*
descended >from him. This probability is 0, so the probability that you are
descended >from him is 1.

You can still reach this conclusion even if you say that there is only a
1% chance that each ancestor is Ashkenazic. The numbers are that strong.
However, even if the world would agree to this percentage (which is highly
doubtful), you have still implicitly assumed that each ancestor is
independent >from each other one, which is ridiculous. To fix this, you
would have to set up the kind of models you allude to here. This is where
I quit figuring the probabilities.:-)

Robert Heyman
JGSGW and Canberra
robeh@...
http://home.iprimus.com.au/robeh

From: "Jacob D. Goldstein" <jake@...>

The argument that has been put forward runs as follows:

X is Ashkenazi. If X's ancestors were all unrelated for (let's say) 30
generations, then X would have had to have many more ancestors 30
generations back than Jews have ever lived. Therefore, we all descend
from a specific individual who lived and had descendants 30 generations
back.

Am I the only one who sees a logical flaw in this argument?

It seems to me that all that can be concluded is that some of X's
ancestors must have been related to each other.

Questions about the likelihood that two Ashkenazi Jews alive today are
related require models and assumptions regarding how individuals come to
be called Ashkenazi, how they move >from one community to another, how
disease/epidemic/antisemitism affects communities, etc. To be realistic,
these models have to be non-homogeneous (migrations are not random,
especially when one knows History!) and time-varying (it hasn't always
been the same, has it?) In the end, computed probabilities only reflect
the models that were used. To have any relevance (relevance and reality
are closely connected) the conclusions must be fairly insensitive to
reasonable changes in the modeling assumptions. As far as I know, nobody
has carried out this research.

Join {main@groups.jewishgen.org to automatically receive all group messages.