Re: RM formalism supporting partial information

Marshall wrote:
..
> It seems to me that anything that we can say about partial
> information can be said with total information. In other words,
> efforts at making the *system* understand partial information
> are merely pushing systemward calculations that could be done
> in a system without any understanding of partial information.
> ...

Not that I'd call my own experience conclusive, but the big systems I've seen had way too much partial information. That sounds like an oxymoron - a better way to put that might be that they had way too much optional information. The people who signed the cheques were usually at some distance from the apps or mostly ignorant of them or mostly ignorant, period, full-stop. Nearly every time I pressed requirements with the authoritative people I would find that partial mean optional and was the result of wishful thinking, slipshod biz practices or suchlike.

I'll grant that most dbms products don't help the situation since the ones I've heard of generally offer only a logical interface for storage.   Eg., a small business may well have customers who have addresses but not telephone numbers. While I hope I would choose to starve before working on a "data-mining" app, mostly because I don't agree with that as a way to do business, I don't see any wrong with nulls in an app that is concerned with trends and statistics rather than individuals. At one time I did a lot of consumer survey programming and the "DK's"/"don't know" responses were sometimes just as useful to clients as specific answers.

It might have been you, Marshall who said here that we humans usually have no choice but to make decisions based on partial information (which certainly seems true when it comes to dbms implementers!) but I think that is different from a dbms who's purpose is merely to give a mechanical way to save us effort. Received on Sun Nov 25 2007 - 01:22:55 CET