Re: Who first (publicly) asserted 3NF is "good enough"?

Roy Hann wrote:

> It has always seemed to me that nothing looks more obviously and intuitively
> wrong than a table that is in third normal form (3NF) but not also 4NF.
> Furthermore I rarely see violations of 4NF "in the wild" in databases
> intended to support core business transaction processing (as opposed to say
> DBs for decision support or data warehousing). This makes me think even
> ill-informed database designers can immediately spot the problem and avoid
> it even if they don't know what it's called.
>
> Assuming I'm right about the above it is baffling that one so frequently
> sees books and articles in which it is asserted that 3NF is "good enough".
> I assume the authors don't actually know what 4NF (and 5NF) is and they are
> quoting and re-quoting some original source that gave them permission to
> ignore 4NF etc.
>
> It's probably far too late to get a confident answer now, but does anyone
> know if there was a single authoritative writer (long ago) who expressed
> this foolish idea?

It's not so much that 3nf is good enough as it is that almost all 3nf designs are also in 5nf. Violating the higher normal forms requires compound keys and complex dependencies, and designers tend to avoid both.

Having a thorough understanding of 3nf may be good enough as long as one also knows to open a reference text at the first hint of a complex dependency or a compound key. Received on Mon Sep 18 2006 - 15:15:22 CEST