Re: Has E/R had a negative impact on db?

From: Bob Badour <>
Date: Sun, 23 Apr 2006 15:48:01 GMT
Message-ID: <RkN2g.64642$>

David Cressey wrote:
> "Bob Badour" <> wrote in message
> news:V1L2g.64571$

>>David Cressey wrote:

> [snip]
>>>I honestly believe that this long discussion has lost sight of the idea

> that
>>>the E/R model is an analysis model, and not a design model.  Arbitrary
>>>choices made during the analysis phase can be useful, in order to come

> up
>>>with a conceptual model that is easily communicated from one person to
>>>another.  That's the main usefulness of an E/R model, IMO.  If the E/R

> model
>>>were a design model,  then arbitrary choices would be far more costly.
>>Perhaps, our experiences differ. I have found the NIAM/ORM methods of
>>constructing english sentences to describe the concepts more useful for
>>communicating with domain experts.

> Perhaps. I don't have enough experience with NIAM/ORM to compare with E/R.

That's a pity. You aren't doing yourself any favours.

>>Programmers like pretty pictures, but users' and business experts' eyes
>>glaze over. This usually precedes: "That's good. When will the software
>>be ready?" ;)

> Our experiences differ here, as well. The users and business experts I've
> dealt with like pretty pictures, provided they are few in number!

But that's the problem. "Oooh, aaah! That's good. When will the software be ready?" is not the same as: "No, this over here is wrong. It should be..." or "I don't understand. What other possible alternative could there be?!?"

If the goal is to get users to sign off on a design as soon as possible regardless of the cost of inevitable failure, then I suppose pretty pictures are the ideal solution. Are you a consultant by any chance? Does "Big 5" feature predominantly on your resume?

Once again, I direct your attention to a gem of great wisdom from the mind of EWD and others before him:

In that vein, NIAM surpasses ORM. However, I believe the availability of tools for ORM surpasses the availability of tools for NIAM, and ORM has the desirable feature that one can still disregard the pictures and communicate in english.

  On the
> other hand, a large number of English sentences, bound together in something
> like a "functional spec" make their eyes glaze over.

I don't recall where I suggested such an absurdity. NIAM or ORM allows one to present the domain expert a list of fairly simple sentences with which the user can agree, disagree, or express confusion.

Since the list provides an exhaustive reflection of the past communications between the analyst and the domain expert plus the analyst's assumptions about same, the process quickly reveals errors in perception and assumption, and alerts the analyst to areas requiring further probing.

It can also trigger memories of special cases in the domain expert's mind. Contradictions among domain experts generally lead to very fruitful discussions. Received on Sun Apr 23 2006 - 17:48:01 CEST

Original text of this message