On Nov 14, 3:35 am, Ian Goddard <godda...@hotmail.co.uk> wrote: > Steve Hayes wrote: > > On Tue, 13 Nov 2007 20:39:27 +0000, Ian Goddard <godda...@hotmail.co.uk> > > wrote: > > >> The underlying data model of GEDCOM is very simple. IMV far too simple. > > > Or far too complex. > > > GEDCOM tries to group people into families, which is an unnecessary > > complication. > > > The basic relations are father-child and mother-child, and if GEDCOM stuck to > > that it could simplify things. > > Same difference! The *evidence* provides names of people (often > inconsistently spelled even on the same document. A data model should > provide an entity to represent data (names and roles) extracted from the > evidence; this is *analysis*. A simple data model which stops there > would be what you seek and I could go along with that. > > If we then try to group into families we build hypotheses which identify > one analysis record with another. This is *interpretation*. A data > model which makes provision for interpretation should provide a separate > entity for this - and make the link between the analysis and > interpretation representations of individuals a further entity. This is > the sort of model I would prefer. > > GEDCOM does neither. It provides a single entity to represent > analytical and interpretive views of individuals. It tries to achieve > complex ends by simple means. > > -- > Ian > > Hotmail is for spammers. Real mail address is igoddard > at nildram co uk If anyone can come up with a better standard please feel free to do so. However to accomodate the peculiarities of every program is a difficult if not impossible task. First of all perhaps all programs would have to adhere to a standard defining what data is acceptable.