Re: dbdebunk 'Quote of Week' comment

> I have found I can use surrogate keys and
> suffer none of the grief you attribute to them.
> [...]
> I guess I should declare 27 years of involvement

In my travels, I have been to all corners of North America. I've been atop mountans. I've peered out of skyscraper. I've even flown in a plane. With all my traveling experience, I have yet to find any evidence whatsoever that the earth is round; as far as I can tell, it's flat as a board (moutants et al withstanding).

Despite my personal observations, I have not joined the Flat Earth Society. I rely on the expertise of others to mathematically prove a round earth. I trust the experience of those who have personally observed that the earth is in fact round.

When you teach yourself something, you end up relying on your own observations and assumptions. When you rely on what you know, you lose out on volumes of collective knowledge and experience. Had you not read it in a science book, do you really think you'd come to the conclusion that the earth is round?

The same holds true for databases. When a programmer self-teaches himself databases, he applies his own experience: procedural programming. This is what happened in the fifties when there were no databases and, therefore, no experience or expertise to go by. Programmers back then modeled data exactly like you do now: like a programmer.

Fifty years, the data field has made a lot of progress. The best model we have now is the relational model. One of the fundamentals of this model is that data do not need to be accessed by pointers (meaningless keys); they can be accessed (and related) by their value.

When you ignore this fundamental, your model becomes closer and closer to the ineffective ways of the past. Worse still, you lose the tools they had back then (IDMS pointer functions, for example) to keep their models running. As Santayana said, "Those who cannot remember the past are condemned to repeat it." Received on Fri Aug 19 2005 - 17:15:08 CEST