Re: [Scid-users] Relying on a Tournament File

[Benoit St-P. <ben...@gm...>]

Hello,
This morning, I dreamt exactly about that :

You start out with what you have at hand: a game.
> Then you identify the tournament and set up an authority
> record for that one containing a name, probably the date it
> took place and so on, enough information ot make it a real
> individualized recored. And you assign it an identifier.


Here is what I dreamt about : suppose you have a game G from a tournament T
in your database D.  To maintain G, you invoke a future version of the
maintainance window.  It checks the validity of the information in its
tournament file.  Then it prompts :

<< We know that this game exists : [...].  Do you want to make the
corrections ? >>

proceeding then in the ellipsis to mention the similarities and divergences.
 So it is a factchecker, really : what it does is that it tries to see if it
already knows there is a G in T already.

This is not ideal, since only working with the data about the game can be
misleading : suppose Fischer plays a certain Szabo in a tournament in 1951,
and that he plays his brother too.  To know which games Fischer played
against whom, one must rely on the scoresheet, in a way.


> Then you identify the persons involved. You idividualize
> them using person authority records you have at hand already
> or your create a new one if the person is not known to your
> system. The latter again contains enough information to
> individualize each person. You assign again an identifier.
> (For persons using VIAF-ID, if it exists, is strongly
> recommended).
>
> Finally, you hook up the game to your tournament.
>

That's where my intuition was not the same as yours : for me, it was easier
to maintain a database if the historical validation was going by tournament.
 But I see now what you mean when you say that working with games is the
bottom-up approach. I disagree, since a game is a complex, but that's not
important : I can follow what you mean.

Second order. Build a database of the above structure, and a
> way will be found. This is "only" about coding. The hard
> thing is to get such a DB done.


My point is that the only way to have the DB done is to have the tool am
asking about.  If I have to spellchecks every Szabo by hand, I am afraid
that it will never happen.  If the tool I am asking about exists, the job
gets way more easier.  And it gets way more modular : by cleaning up
tournament by tournament, the job could easily rely on a very big team of
validators.  In fact, every chess organizers could be held responsible,
then, in a way.

This kind of file could also be of interest for an historical database of
chessgames.  It would look like www.chess-results.com, but for tournaments.

I don't have much time left to discuss the idea, but I like it more and more
!