Re: Pizza Example

"Anthony W. Youngman" <wol_at_thewolery.demon.co.uk> wrote in message news:<CbzHNpBGVwgAFwaa_at_thewolery.demon.co.uk>...
> In message <pan.2004.04.18.01.32.20.45299_at_erols.com>, Jerry Gitomer
> <jgitomer_at_erols.com> writes
> >>>You still persist in believing that nonsense about "losing information"
> >>>by decomposing data? This just reflects your ignorance of how an RDBMS
> >>>works, and it is time you rectified that.
> >>
> >> It happens in the scientific world all the time. Why should the data world
> >> be any different?
> >>
> >Because data integrity standards are much more rigorous in the commercial
> >world than in the world of science. For example, it is not at all unusual
> >for a scientist to be pleased that 90% of the sample data collected on a
> >project is good data and only 10% is bad data that will be discarded
> >and/or ignored. When it comes to dealing with payroll data, payables and
> >receivables, and inventories anything less than 100% is unacceptable and
> >checks and balances are built into commercial systems to insure that all
> >of the data is collected and that, before it goes into the database,
> >all is valid.
>
> Who gives a sh*t about "standards" when there's a scientific law that
> says you can't do it?
>
> I do appreciate what you're saying, about how database integrity
> standards are rigorous, but the scientific equivalent of what you
> propose is not separating the results into "90% good, 10% bad"; it is
> claiming that you have just invented a perpetual motion machine. In the
> real world, IT CANNOT BE DONE. PHYSICS FORBIDS IT.
>
> The standard version of the law is basically "in any closed system, any
> change will increase entropy". A physicist would also recognise and
> accept the following paraphrase - "in any closed system, any change will
> destroy information" - the terms "entropy" and "information" are
> recognised as a sort of opposite (I can't remember the detail, it's too
> long ago).

So am I right in thinking that you do not believe (as it appeared earlier you did) that normalisation by non-loss decomposition loses information always? Rather you are saying that mistakes could be made while storing data such that some small fraction of the data would lose its correct linkage? Is that your position? If so it is a lot less ridiculous than I previously thought, while being typically irrelevant to the topic. By extracting the data from the real world and storing it in ANY database (even Pick) you are changing it in some sense, and the chance of errors exists. So what? Received on Tue Apr 20 2004 - 01:36:57 CEST