Re: Do Data Models Need to built on a Mathematical Concept?

From: André Nęss <>
Date: Wed, 07 May 2003 22:41:14 +0200
Message-ID: <b9br28$2ae$>


>> > You don't see the parallel between a db representing things
>> > and a brain representing things?
>> What you propose (as it seems to me, anyway) is to reveal the
>> incomprehensible mess of neurons and axons directly to us.
>> This is supposed to help us how?

> Yes, behind what looks like an "incomprehensible mess of neurons and
> axons" to you, looks to me to be the implemenation of a more flexible
> data representation system.

The ultimate data representation then is binary code, which can represent everything. But how useful is it?

>> If you find that the design of the brain provides you with ideas for how
>> you would like to *implement* a DBMS, fine, but using this implementation
>> as the data model can only lead to a very complex system with little or
>> no benefits.

> You haven't yet realized that the reason the brain is more flexible is
> because it uses a simpler, not a more complex, data model!

Atoms are simpler than cars. That doesn't mean thinking about cars in terms of atoms give you a better understanding (if any!) of cars. I frankly don't know what the data model of the brain is, and I don't think anyone else know either.  

But I guess what this boils down to is that you have this system where everything can be related to everything, and you say that this is how the brain works to give it credibility, because the brain is after all an incredibly advanced thing.

But the real issue here then is whether one can really benefit from being able to relate everything to everything else, regardless of whether this is how the brain works or not. And it is my opinion that one cannot. In order to bring structure and order to data one must relate parts of the data to other parts of the data in a defined way, for if the relationsships are arbitrary, then how are we to understand what the data means? We will have to interpret them manually, but interpreting stuff manually is boring and precisely what we want to avoid, we want computers to do all that tedious stuff.

So let me put it this way: until you can show some real world examples where your data model is better than the relational (which I believe is the best), I'm not going to be convinced.

André Nęss Received on Wed May 07 2003 - 22:41:14 CEST

Original text of this message