Re: Nulls, integrity, the closed world assumption and events

On Jan 14, 12:46 pm, "dawn" <dawnwolth..._at_gmail.com> wrote:
> Marshall wrote:
> > On Jan 14, 10:32 am, "dawn" <dawnwolth..._at_gmail.com> wrote:
> > > JOG wrote:
> > > > However, I am a firm believer that an elegant
> > > > theory will lead to good practice.
>
> > > Any elegant theory? That has surely steered us wrong in the past.
>
> > That's a clear an anti-theory statement as I expect to see.
> > Since you are anti-theory, why are you in a theory newsgroup?
> Come on, Marshall, you know that there can be an elegant mathematical
> theory that should not be applied to this or that.

I know that we don't apply the theory of evolution to arithmetic, or game theory to naval architecture, but otherwise I don't know what you mean. Theories are intimately connected to their area of applicability.

> When we are talking
> about applying theory, rather than advancing the theory in and of
> itself, there is an additional factor that it needs to be useful. On
> top of that, there might be another elegant theory that is more useful,
> so just because a theory is elegant and useful does not make it the
> best choice.

What do you think a theory is, that it can fail to be useful? A theory with no utility is immediately discarded. Utility isn't an "additional" factor; it's how we evaluate theories. Elegance can be seen as an aesthetic response to a theory with a high ratio of utility to complexity.

> > Doesn't that make you a creationist in an evolutionary biology
> > newsgroup?
> if this group were comp.theory.relational or some such, then it would
> not be related to applying theory, but simply advancing a specific
> theory, whether an application can be found or not in one area or
> another. But it is comp.databases.theory which to me implies not just
> talking about mathematical theory, but its application to databases.

Sure. And since you are of the opinion that theory can steer us down the wrong path, and since you regularly deprecate relational theory but do not provide an alternative *theory*, I wonder at your presence in a theory newsgroup.

> In that case, there are surely many mathematical theories that have
> been applied, and there are choices to make on when to employ what
> theory. Both two and three-valued logic are theories that can be
> applied to databases, for example. Discussing which one to use, how to
> choose one, etc would seem to me to be appropriate in a database theory
> forum. So, I would think this would be a place to discuss what makes
> the most sense for databases. Do I have this wrong?

Your attribution of 3VL to relational theory is wrong, for example; the two are independent. (Actually some here would probably argue that they are antithetical, even.) Your attribution of of a proscription against
nesting to relational theory is also wrong. These are attributes of SQL,
not the relational model; I observe that you often omit the distinction.

Marshall Received on Mon Jan 15 2007 - 03:01:52 CET