Re: the two questions

On Nov 27, 5:13 pm, "David Cressey" <cresse..._at_verizon.net> wrote:
> "JOG" <j..._at_cs.nott.ac.uk> wrote in message
>
> news:0c832d02-88f5-495c-ab2b-8098afcd8818_at_d21g2000prf.googlegroups.com...
>
>
>
> > On Nov 27, 3:49 pm, "Brian Selzer" <br..._at_selzer-software.com> wrote:
> > > "JOG" <j..._at_cs.nott.ac.uk> wrote in message
> > > Each individual that existed, exists, or can exist has a property that
> > > distinguishes it from all other individuals that existed, exist or can
> > > exist; so, yes, there is a property that the caterpillar and butterfly
> > > share.
>
> > Great, we have agreement :)
>
> > > The problem is: I don't think haecceity can be observed directly.
>
> > This time I agree with you (although I did have to look up what
> > 'haeccity' meant) - it is often the case that the identifier we need
> > isn't available to us (I mean we can't often check a butterflies dna
> > right...).
>
> > But we have to find a solution to this in the real world right - If I
> > have a butterfly, how do I know it came from the caterpillar from
> > earlier? Would you agree there are two options?
>
> > 1) Check an identifier that we can manage to observe (dna if we're
> > lucky, more likely the jar number we've kept it in, etc.)
> > 2) If we couldn't access that identifier (or it was just too much of a
> > pain to do so), we'd have needed to invent a new identifier as a
> > replacement, that was trackable (a representative identifer for the
> > insect's 'haeccity' - similar to what biologists do when they 'tag'
> > birds).
>
> > Again, all in the real world, before we get to a database.
>
> > > If one were able to examine the history of the butterfly, one should be
> able
> > > to determine that it coincides with the history of the caterpillar--up
> to
> > > the point of the initial snapshot. The problem is: I don't think
> history
> > > can appear in a snapshot.
>
> > I get your gist here but hope we can come back to it after you've
> > looked at the above question. Regards, J.
>
> As a practical matter, what we do when we want to track individual living
> creatures is to tag them. Let's take birds, rather than butterflies. Near
> where I live, they are banding birds. When a dead bird is found, they ask
> for the number from the band, if there is one. Among other things, this
> allows better tracking of bird flu.
>
> A few days ago some people, in jest, suggested that people get an ID tattoed
> to their butt. For purposes of this discussion, banding birds is precisely
> the same concept, minus the humor. The number on the band identifies the
> band. Since the band is an artifact, we can build it so as to carry its
> identity around in visible form.
>
> Using the band number as if it were a bird number depends on attaching the
> band to the bird in a relatively permanent way. I think it's fair to call a
> band number an "artificial key". I'm less sure about the term "synthetic
> key". I think it's misleading to call it a "surrogate key". Surrogate for
> what?

Well, it's a surrogate for the attribute (or set of attributes) that makes that bird unique. That'd be its DNA code in the case of an individual bird I guess.

But then as a caveat I'd also say talking about surrogate "keys" is a nonsense, period. Keys apply to propositions, and we're still at the conceptual layer talking about entities. The band is a surrogate 'identifier' - an attribute of the bird entity. That it happens to be referred to later on in a proposition about the bird is just a knock on effect. (Is this what Bob means about them just being unfamiliar keys? I may have finally twigged what he's on about.)

The key vs identifier may seem a pedantic distinction to make, but I honestly think its a bloody important one given how much confusion between logical and conceptual layers there is. The fact that E/R modelling talks about keys as if they were analagous to RM keys also me feel decidedly squiffy.

>
> We do the same thing when we use Social Security Number to identify people.
> In the narrowest sense, a SSN identifies an account in the Social Security
> system. The attachment between the person and the account is less tangible
> than the band on the bird, but is intended to be quasi permanent.

Yup, pretty much I'd say. But its worth noting that the SSN probably identifies a US citizen, rather than a person. Different entity types.

>
> Again, all of this is "real world" before anything goes into a database.

In the real world I use a lot more bad words. Received on Wed Nov 28 2007 - 02:04:30 CET