Re: Genealogical evidence and data model

Ian Goddard <goddai01@[EMAIL PROTECTED]
> wrote:
>Tim Powys-Lybbe wrote:
>> In message of 20 Jan, Ian Goddard <goddai01@[EMAIL PROTECTED]
> wrote:
>> 
>>> singhals wrote:
>>>> But, I'd venture to suggest that out of any 100 
>>>> genealogists at least 51% _still_ want a program to record their 
>>>> conclusions so they can print it out.  

I suspect that number is more like 95%, with the emphasis on "print it
out" or otherwise displaying the results.

>>> If that wish were properly implemented things wouldn't be so bad.  But

>>> when I get back to the mid C18th the only conclusion for quite a
number 
>>> of lines is "A or B" or even "A or B or C".
>>>
>>> S/W which assumes that pedigrees can always be resolved unambiguously 
>>> won't handle this situation.  I can't believe that the situation is 
>>> uncommon so how do users of such S/W cope?  Do they simply stop
entering 
>>> their results when they get back to an ambiguity?  Do they allow their

>>> S/W to dictate to them that they should choose what might at the time 
>>> appear to be the most likely ancestor and discard the rest?  Do they 
>>> generate parallel universes of databases to handle all the alternative

>>> combinations?
>> 
>> I detect here, though I could be wrong, a thesis that it is possible to
>> prove facts by logic.  Computers are products of logic.  So computers
>> can easily prove genealogical facts.
>
>No. It's simply remarking on the fact that if the designer of a program 
>assumes that the data will fit an unambiguous structure then they will 
>fail to design a program which can handle alternatives.  And asking what 
>users do when confronted with a situation which the designer has not 
>envisaged.

The software I use (Legacy Family Tree) allows me to enter multiple
parents for an individual, which I can then comment.  But the real
problem is that there is no effective way to display or re****t this.
I can enter all the ambiguous data I want, but the bottom line result
is that when it draws a tree, or creates a re****t, it wants to pick
one of the alternative parents.

Most people doing genealogy like trees or other graphic renditions of
their data.  Some like text-based approaches - but ahnentafl numbering
does not allow for alternatives/ambiguity as well, and I don't know of
any other basic genealogy re****t that can re****t on the ambiguities in
any standard fa****on.

The best I've been able to do is put a special character in the
"title" field that is part of the name for those individuals that are
particularly hypothetical.  But that doesn't really solve the problem.
I may know that my ancestor's name is John Smith, but the ambiguity is
in which of the many people by that name is the correct one.  Thus one
needs more flexibility than software typically allows to enter
alternate dates.

Legacy allows this - I can use "events" as well as notes to enter
alternative values, and I can the source evidence and my certainty
about the interpretation I am putting on that evidence.

But the trees and re****ts don't display all of the "if"s, "and"s, or
"but"s.  And I don't think that any improvement of the data model
would really help if the charting and re****ting don't present results
that meet their needs.

System design, as I see it, needs to take into account the likely
outputs as well as the likely inputs.  Most of the discussion of the
data model seems to be about representing the inputs accurately.  But
most people use software to produce outputs, or to allow effective and
efficient searching of the data that they have entered (the ability to
search is itself a kind of output).

Forgetting the output side in order to maximally represent the
vagaries of the inputs will not get people to enter data properly.
They will enter data properly when there are well designed re****ts and
charts that present that data most clearly and usefully when entered
properly.

>> Regrettably this is not logic.  Logic is not about truth claims, it is
>> about the rules of truth claim and thus its subject is all statements
>> whether true or false.
>
>Ah, I see how I misled you.  I wasn't using "or" as a logical operator 
>but simply saying that, for example, "Mary Collier who was the bride in 
>this marriage was either Mary daughter of George Collier whose baptism 
>is recorded in record A or Mary daughter of Abel Collier whose baptism 
>is recorded in record B".

How would you show that in a tree? (or some other form).  Until people
can see the information they have entered (or its indirect effects) on
output, they likely won't spend a lot of effort on input.

lojbab