GEN-MEDIEVAL-L Archives

Archiver > GEN-MEDIEVAL > 2000-01 > 0948647264


From: Rafal Prinke <>
Subject: Re: Ratings issues
Date: Sun, 23 Jan 2000 18:07:44 +0100


Chris Bennett <> wrote:

> The dimensions to be considered include at least the following:

> a) The nature of the documentation: is it original, primary material
> directly attesting to the link? Is it contemporary secondary
> material? Is it tertiary material claiming to be based on primary
> material? Is it an inference made from such material, in isolation of
> in combination?
>
> b) The reliability of the documentation: Is it authentic? Is it
> propaganda? Is there motivation for fraud, either then, now or in
> between?
>
> c) The interpretation of the documentation: Is it unambiguous? What
> are the dimensions of the ambiguities? To what extent does our
> solution depend on contextual assumptions in addition to (or even in
> place of) actual documentation? Do we understand the social
> assumptions of society contemporary with the events? Of society
> contemporary with the documentation? Have we overlooked possible
> solutions because we are imposing our own social filters on the
> material?
>
> No doubt there are others. Devising an appropriate scale is not a
> trivial endeavour.

While the above is obviously a very good "model solution", its
practical application may prove difficult. For example, those who
easily jump to conclusions will tend to answer the questions above
in the affirmative, while the cautious researchers will often
feel uneasy to state that they _really_ "understand the social
assumptions of society contemporary with the events".

So - the actual rating assigned to a given filiation will depend on
who assigns it ("the human factor"). Even on this list we can often
observe how the best specialists in a given field differ in their
opinions about particular links/filiations/facts/events/dates/etc.

Stewart Baldwin suggested "a consensus among scholars" as the minimum
requirement for accepting a given filiation as reasonably certain.
The problem with this criterion (as well as others) is that consensus
is not something stable - it changes with time and itself has
different degrees of "weight". An article published in a peer reviewed
journal represents a much stronger consensus than one in an informal
bulletin of a genealogical society - but much weaker than inclusion of
the same fact in a scholarly synthesis or encyclopedia.

But such consensus is always outdated by currect research. So if we
accept ES as representing the consensus, there is already a whole
body of articles and monographs questioning particular pieces of
information included there. These may - or may not - be incorporated
in a future update.

On the other hand, it seems that a rating scheme like that would be very
useful as a kind of "abstract" which would show the author's own
evaluation/rating of the filiation etc. and the basis for accepting it.
So - a rating like 3.2.0 would indicate by numbers in given positions
what sources were used, what is the researcher's opinion about those
sources, whether any original argumentation is involved or is the information
simply copied, etc.

> 2) Traditionally, a genealogist would make a personal evaluation based
> on the evidence presented, per John's book on Catherine Baillon, but in
> this age of mass genealogy, easy copying and digital databases, this
> mechanism is visibly breaking down. Hence we are looking for a rating
> scale as a mechanism appropriate to the new age.

Such mechanism could be easily extended to translate the information
imported into a database or exported from it. For example, if
a fact is labelled "3.2.4" (this is just an arbitrary example, with
no proposal for actual codes) indicating "I used primary sources
translated into my mother tongue. I think they are quite reliable.
I know the area/period fairly well and thus can evaluate this evidence",
on export/import the label would be translated to something like:
"0.0.0:3.2.4" which would mean that the data were simply copied from
a source which labelled it like that. If I use some other information
and/or have additional arguments, I might modify the first part of the code.

> To be successful, the resultant scale must then be incorporated into
> standard database structures, and made easy to use and understand, e.g.
> by use of drop-menus and help balloons to access and explain the scale,
> and by careful choice of default values (not a good idea to make the
> default "certain" for example). But that is all standard stuff these
> days. Stewart is correct that people can still deliberately remove or
> reset the values, or choose to abuse them, but I disagree that makes
> the effort useless -- its one thing to plagiarise databases intact, but
> it takes positive action to modify them, so the scale should act as
> real brake to the widespread distribution of unevaluated data.

To put it in different words - because there are no "surety codes",
on-line database information is accepted as "true". Even question marks
help novices to think twice before copying the data into their own files.

> A complicating factor is that apparently the industry has already taken
> some steps to address the issue (Roots IV), and it appears they have
> only considered part of the problem (source quality).

This is because some 99% of genealogical activity (industry) is
concerned with modern and early modern data ("vital records oriented").

> What all this leads to is a conclusion that creating a successful
> system requires a great deal more than a discussion in this newsgroup
> (or even in soc.genealogy.computing) -- its a classic standards
> activity, requiring the backing of professional genealogical
> organisations and of genealogy software companies. I have no idea of
> who the real players and movers and shakers are in that world. But if
> we have a discussion here that leads to a reasonable proposal, then the
> next step is a submission to such people, with all the associated
> lobbying that such submissions require. Presumably it can be done,
> seeing the ubiquity of things like GEDCOM. Does anyone know who the
> players are and what is the current state of the industry on this issue?

There are just two players who might be able to enforce such schemes:

1) LDS - who are rather reluctant to cooperate on GEDCOM development
(and I am saying this with over 5 years' activity on GEDCOM-L).
Actually, both GEDCOM 5.5 and GEDCOM-FD (Future Directions - unofficial)
allow to include surety codes but do not enforce any particular
scheme.

2) GENTECH - which has produced Genealogical Data Model (RFC published over
two years ago) and is now looking into its implementation in XML.
The group doing that (Lexicon Working Group) operates under the aegis
of the largest US genealogical organisations and is lead by Robert C.
Anderson
(of the "Great Migration" fame) but the progress is rather slow.

The latter is more promising - and if the results are published, the industry
will have to follow. So they are the best lobby to approach concerning
this project.

Best regards,

Rafal

This thread: