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

From: mAsterdam <mAsterdam_at_vrijdag.org>
Date: Wed, 19 May 2004 01:20:12 +0200
Message-ID: <40aa9a2e$0$33919$e4fe514c_at_news.xs4all.nl>


Anthony W. Youngman wrote:

> mAsterdam writes:

<major chomp>

>> How does the type of DBMS affect what we consider data?
>>

> Okay. That's actually a very good insight ...
>
> Now let's go back to "The Philosophy of Science" :-) and Newton :-) For
> my first attempt at a Masters, practically the first thing we did was
> "The philosophy of Science". And, helped by both students and a lecturer
> who didn't have a clue (the student extrapolated a line from the origin,
> through an asymptote, to a random position in number-space, and then
> used this to ridicule the theory he didn't like. And the lecturer said
> "good argument" !?!?!? )
>
> I'm going to start saying "metaphysics" instead of philosophy here - I
> think it's a subset of philosophy, and a better word to use, but as you
> can see, I'm really getting into territory I don't understand ...
>
> Anyway. Newtonian Mechanics is a self-contained, consistent,
> mathematical theory. It relies on the concepts (call them "axioms") of
> mass, energy, space, and time (and maybe more). We can define mass in
> mathematical terms as "F=ma, where mass m is the constant property
> describing the resistance of an object to a change in its velocity".
> Likewise, space "is a co-ordinate system with distance measured in
> metres along three mutually perpendicular axes". I won't attempt to
> define energy or time ...
>
> But just as those four concepts have neat, clean, mathematical
> definitions they also have messy real world definitions. Mass can be
> defined as "my god it's heavy", or "come on! PUSH!". Space is "where are
> you?" or "I'm here, you're there".
>
> Metaphysics is, I believe, the attempt to clarify both the real-world
> definitions and the mathematical definitions, and to try to make sure
> that they are describing the same thing. This is why, despite knowing
> that Newtonian Mechanics is wrong, we find it so useful. We know the two
> definitions don't match, we know WHERE they don't match, and we can
> predict with certainty that where the discrepancy is minimal, Newtonian
> Mechanics will give us a suitably accurate answer.
>
> Now I'm going to get into the difference between "relational theory" and
> "relational database theory" :-) Another analogy coming up - Linux and
> microkernels :-) Linus realised that all this research into "Microkernel
> Operating Systems" was actually just as applicable to "Operating
> Systems". I'm putting peoples' noses out of joint because, whether they
> realise it or not, they believe in "relational database theory" (think
> Tanenbaum saying he'd give Linus an F :-) And yet, I keep on saying Pick
> data should be normalised! So I'm actually very pro relational theory
> (just leave relational databases out of it! :-)
>
> Now here comes the crunch. As I see it, in relational *database* theory,
> the concept of "data" lies on this metaphysical boundary. And this is
> why I view every relational database I've ever seen as a tangled mess of
> spaghetti. What the hell is "data"! What's the real world equivalent?
> Like any true mathematician :-) the relational database theorists seem
> to be saying "metaphysics? that's not our problem. That's just an
> implementation detail!". Except that, going back to Newton, the fact
> that energy and mass are interchangeable and, as such, the equation
> "F=ma where m is a constant" isn't true, isn't an "implementation
> detail". Well, to God it may be, but it certainly isn't to us!
>
> Going to another thread, where Lauri asked what were the advantages of
> Pick, I'd say that one of them is a very clear metaphysical interface.
> To compare Pick and Relational Database Theory ...
>
> A Pick FILE is a real-world collective noun. What's a relational table?
  1. A contradictio in terminis.
  2. A collection of similarly shaped utterances.

> A Pick RECORD is a real-world object. What's a relational row? A noun?
> An adjective? A gerund? (relation, for those who don't know their grammar)

  1. A contradictio in terminis.
  2. One utterance.

> A Pick FIELD is a real-world adjective. What's a relational column? An
> adjective? A gerund?

Mu.

You compare

P.FILE to S.TABLE,
P.RECORD to S.ROW and
P.FIELD to S.COLUMN.

What do we learn from this comparison? Nothing. These terms are all taken out of the context where they have meaning. One may just as well choose to compare
P.FILE to S.SCHEMA,
P.RECORD to S.VIEW and
P.FIELD to S.TABLE

- it doesn't mean anything.
It is out of context.

> Because Pick's metaphysical layer is at a higher level

That depends on which terms you compare from one realm to which other terms from the other. It's your pick. (sorry :-)

> than Relational
> Database Theory, we can then implement relational theory WITHIN our
> model without having the nasty spaghetti of a vague and undefined
> real-world interface. And I can righteously and reasonably throw my
> hands up in horror and tear my hair out when presented with a Pick
> database that hasn't been normalised :-)

Yup. That even goes for very old fixed record batch processing.

> So. Can anyone come up with a clear, simple, and NON-VAGUE definition of
> what "data" means when specified in a real-world, not a mathematical,
> context. Or come up with a perfectly good reason of why you don't have
> to! (Basically, because you've done it somewhere else, because you've
> got to do it somewhere!)

Yup. It seems most people prefer to have that done implicitely or at least by someone else. Received on Wed May 19 2004 - 01:20:12 CEST

Original text of this message