Re: A Normalization Question

"D Guntermann" <guntermann_at_hotmail.com> wrote in message news:I0JwEE.KnI_at_news.boeing.com...
>
> "Larry Coon" <lcnospam_at_assist.org> wrote in message
> news:40EC69EB.723F_at_assist.org...
> > Tony wrote:
> >
> > > You are of course confusing logical and physical issues - and your
> > > issue is absurd in any case. It is not LOGICALLY redundant to record
> > > that "Car X is brown" and "Dog Y is brown", because these are two
> > > different facts. You are presumably saying that it is PHYSICALLY
> > > redundant to store the 5 characters of "brown" twice on the disk, and
> > > so you want to physically store the word "brown" once and then point
> > > to it many times.
> >
> > You couldn't do that -- if more than one thing POINTS
> > to "brown," then the pointers are stored redundantly.
> > Infinite regress rears its ugly head.
> >
> Yes! Neo read and listen!
>
> The proof of the inconsistency of your logic and its inception as a
fallacy
> is here, stated succinctly by Mr. Coon.
>
> To eliminate redundancy of words (sequences of characters), you create one
> word and subsequently create references or pointers to that that word.
But
> this is not good enough. These characters within sequences are redundant
> themselves, so you create references or pointers to a single symbol.
>
> But then these references or pointer values are redundant, so to ensure no
> redundancy in these values, you must create a new set of pointers or
> references using a new alphabet. But wait! Even this new set of
> pointers/or references have values, both in terms of digits and as a
> sequence, that are needlessly replicated, so a new set is needed to point
to
> the pointers that pointed to the pointers that point to a single value.

I tried that argument already. He apparently ignored it. It's also in an old joke I keep referring to:

"Did you hear about the prgrammer that was found dead in his shower? He was frozen staring at the instructions on a bottle of shampoo: "Lather, Rinse, Repeat."

>
> I leave it as an exercise to see how far you can go repeating this cycle
> before you finally get the "ultimate generalized form of normalization",
> where no duplication of a symbolic value exists.
>
> Otherwise you can skip that and attempt to eliminate redundancy at the bit
> level. Of course the only two values you could use are 0 and 1 which
gives
> you two states to create both values and unique non-redundant references
or
> pointers. Good luck with that one!
>
> Another option would be to create an alphabet that no one uses or is aware
> of that consists of enough symbols to ensure that there will never be a
need
> to replicate the symbol in a database ever again. The tens of thousands
of
> chinese symbols is but a drop in the bucket in comparison to this
endeavor.
> When you get this done, you can get rid of the binary, octal, decimal, and
> hex number systems and create a 1 digit numeric base system where each
> numeric value, integer or real, is represented as a single symbol, don't
> forget up to infinity, and infintessimally small!

The data entry would be done by the infinite number of monkeys I referred to elsewhere.

>
> Good Luck!

He'll need more than that!

>
> > Larry Coon
> > University of California
>
> - Dan
>
>
Received on Fri Jul 09 2004 - 00:16:30 CEST