On Feb 11, 2009, at 1:23 PM, Chris Partridge wrote: Hi Pat, I think someone (you?) may be working under different premises here. That isn't what 'extensional' means. Everything would be a lot easier
if y'all used the terminology properly.
Properly? Is there an agreed definition????
Yes.
will do, and the links from there. Notice this defines "extension" differently from you, by the way. What is PatH's definition?
See above. Relations or properties or sets or concepts are extensional when they are 'defined by their members', ie they are identical when they have, or apply to, exactly the same instances. They are intensional when they aren't extensional. Your example of equilateral and equiangular is a fine example. But my point is that this says nothing at all about things being spatiotemporal. A set of numbers or (choose your own aetherial abstraction) is just as extensional as one of pieces of apple pie.
Thats about the word "extension", not "extensional". And in any case, see
BTW the list has just had a long discussion of this point, where the various senses were pointed out. But why do you think that
strict extensionality (in your sense, here) is so desirable? What is
the advantage to all this extensional rigor?
As Ian has said, it is the elusive grail of a criterion of identity - two things are the same if ... etc.
Yes, but its just one way to provide that. What makes this way better than any other way? (Actually you already answered this: people can be got to agree on this one more easily. OK, good answer.) Seems on the face of
things that there can be more than one way to describe a given piece
of space/time, eg the famous example of a vase being also a chunk of
plastic, yet having properties (like, being a vase) that the chunk of
plastic doesn't have.
Plastic or clay - well, I suppose clay is a bit passé. And maybe the issue is that the type vase and the type chunk of plastic/clay have different properties - rather than a specific token.
I used plastic because the clay, arguably, exists before the vase is a vase, whereas with injection-molded plastic manufacture, the separate piece of plastic stuff comes into existence exactly at the same time the vase does.
Or the famous example of the Ship of Theseus, or the practical example of a pump's tag and serial number or a military person and their rank or posting. All these are used as standard arguments for a 4D *extensional* approach. The reason 4D is important is that in all the cases above, being co-extensive merely at a point in time (3D) is too strong and gives false positives.
Oh sure, 3-d alone isnt enough, I agree. I didn't even consider that possibility :-)
So why does data integration require extensionality (in your sense,
here)?
If system A talks about a person (tag) and system B talks about a posting (serial number), then one can use the criterion of identity to determine whether they are talking about the same thing.
But thats true of any criterion of identity. Matthew can give you examples of commercial systems which assumed an identity between tag and serial number of things such as pumps which had the consequence you can imagine. I have had engineers tell me that it was extremely useful to have a simple explanation - one they can easily grasp - as to why these are different. There is a paper I wrote when at LOA (or LADSEB) (i.e. long ago) that includes this point. www.loa-cnr.it/Papers/ladseb_tr04-02.pdf . (In case Ed asks, there was no need for talk about characteristics etc.) If you want to say these are very odd engineers, I'll let you take that point up with them.
No, this is very helpful, thanks.
Pat H.
------------------------------------------------------------ IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola (850)202 4440 fax FL 32502 (850)291 0667 mobile
|