Re: In an RDBMS, what does "Data" mean?

In message <FbadnWa-iYfPXTHdRVn-vw_at_comcast.com>, Laconic2 <laconic2_at_comcast.net> writes
>
>"Anthony W. Youngman" <wol_at_thewolery.demon.co.uk> wrote in message
>news:sFcs6eBT2+qAFwPo_at_thewolery.demon.co.uk...
>
>> If we can't set up an experiment (even a Gedanken thought experiment),
>> then relational theory is not provable, therefor it is not scientific,
>> therefor it is irrelevant to the real world, therefor why the hell are
>> we using it :-)
>>
>> As a scientist/engineer type, not a mathematician, I want some
>> experimental proof at least. Unfortunately, all the (anecdotal) evidence
>> I have says that other models work better ...
>
>You make an interesting point here. I would add that the same arguments
>that would render relational theory
>not provable would equally well render the theory non falsifiable. In that
>case, the question for the engineer becomes moot.
>The question "why the hell are we using it" can be countered by "why the
>hell not".

Actually, you and me both have just said exactly the same thing. "provable" and "falsifiable" both mean exactly the same thing as far as science goes - to take that widely misquoted saying "the exception proves the rule (is wrong)". The bit in parentheses is ignored or unknown to most people who quote the saying ... Look in a dictionary. "to prove" can mean "to test".

As for "why the hell not" - well we should be looking for theories that ARE provable/falsifiable. Because if relational theory is not falsifiable, then equally we can't show that it works in practice (for any suitable value of "works"). Would you trust an engineer using Newtonian Mechanics if you had no way of knowing whether relativistic effects were likely in that particular application?
>
>I would suggest that the disciplined practices of engineers are based on
>several sources. One is prior experience, either the personal experience
>of an individual engineer, or the distilled experience of other engineers.
>Another is the accumulated results of science, and of other specialties
>within engineering. A third is the results of mathematics. A fourth is the
>study of how people carry out certain data management and data manipulation
>tasks in the absence of automation. A fifth is the study of the strengths
>and defects of "legacy systems".
>
>Sorry the list got so long.

Well, when I was talking about Newtonian Mechanics metaphysics my list was about that long :-) but if things have to be complete and comprehensive, then sometimes they do get long ...
>
>My personal experience tells me that the relational data model can be, in
>certain circumstances, an enormous aid in managing the complexity of
>defining the data itself, and in clarifying certain issues in the
>development of application software.

I would very much agree ... indeed I would say it almost invariably is a great help, if used as your TOOL and not as your MASTER!
>
>This is a far cry from saying that "all data should be in 1NF".
>
And the same here. As far as I am concerned, the job of the "database analyst designer" is to take real-world information, and convert it to a data schema for the database. If that includes conversion to 1NF, this involves a massive loss of metadata, meaning that the conversion is one-way and cannot be reversed, and therefore the act of conversion renders the whole thing unprovable / unfalsifiable / unscientific.

Cheers,
Wol

-- 
Anthony W. Youngman - wol at thewolery dot demon dot co dot uk
HEX wondered how much he should tell the Wizards. He felt it would not be a
good idea to burden them with too much input. Hex always thought of his reports
as Lies-to-People.
The Science of Discworld : (c) Terry Pratchett 1999