[sc34wg3] SAM-issue psi-generics (was: SAM-issue term-scope-def)

[Bernard Vatant]

> * Bernard Vatant
> |
> | Let M be the topic representing in a topic map the subject
> | "Marc-de-Graauw-the-guy-outhere" Let M1 be the topic representing
> | the subject "M-element-of-this-serialized-topic-map"
> | [...]
> | Now you can rewrite without problem nor ambiguity your assertions,
> | replacing your original topic 'Marc' by M in 1,2, and 3, and by M1
> | in 5 and 6.

*Lars Marius Garshol
> This is correct. We all agree on this.

Great. It seems that we don't agree on the consequences of it, although.

> | [4. 'Marc is an instance of class topic' ]
> |
> | And of course 4 is valid for both M and M1 ... a topic is a topic :)

> This is where we don't agree at all. Either M represents Marc, or it
> represents the topic "Marc".

No! M *is* a topic, and M *represents* the subject Marc-out-there.
M1 is also a topic, representing the subject M, (which happens to be a serialized topic,
but we don't care).

The subject Marc-out-there, the topic M representing the subject Marc, and the topic M1
representing the subject M, are three different animals. Only the two formal ones, M and
M1, we have to care of in the model. All the point of making a model is to be able to set
formal assertions about formal objects, and forget about whatever subjects they represent
*are* or *mean*.

> If it represents Marc then it is false to
> say that Marc is a topic. In other words, 4. is false when applied to
> Marc.

Of course Marc is not a topic, we all know that, but M and M1 are, and they are the ones
who matter really for the model. Marc himself cannot be trapped in the model, and he's
very happy of that :))

> To see why this is so, go back to the first three statements:
>
> * Marc de Graauw
> |
> | 1.  'Marc has a name "Marc de Graauw" '
> | 2.  'Marc has one wife'
> | 3.  'Marc has two sons'
>
> Now, the topic "Marc" doesn't have a wife, so obviously 2. is a
> statement about the subject, and not about the topic.

But 2 is a natural language statement about the subject, it's not in the model !

>   1. The topic map level: The topic representing Marc has an
>   association with the topic representing Marc's wife.

Agreed. That is the only level the model has to care about

 >   2. The subject level: Marc has a relationship (marriage) to his
>   wife.

Yes, but this is a redundant and confusing interpretation/translation of the formal
assertion 1. We don't need that at all.

>   marriage(marc : husband, marcs-wife : wife)
> is an instance of 1. representing 2. And just as clearly, at the
> subject level it is a statement about Marc, and not about the topic "Marc".

Of course, but again the model has nothing to say about the subject level ...

*BV
> | No. All assertions are about topics, but about two different topics
> | (representing different subjects)

*LMG
> Eh, no. All the assertions are about subjects, but using two different
> topics to represent the two different subjects. We don't create topic
> maps to talk about topics, we create them to talk about subjects.
> Otherwise nobody would care the slightest about them.

Of course topic maps are interesting for humans because they make sense when being
interpreted in terms of subjects in the real world, but that is true of *any* model of the
world, and it does not mean we have to make the subjects interfer with the model. Maths
are interesting for non-mathematicians because formal mathematical objects like e.g.
vectors can be interpreted as representing real world objects like e.g. electromagnetic
fields. But when you work with the model, you set and compute formal relationships between
formal objects using formal rules, and you don't let interpretation in terms of subjects
interfer with it. Maths, Physics and other sciences have taken centuries to set that
efficient methodology. My point is that we have to learn from that legacy and get out of
old pre-scientific-world-language-model-confusion pitfall.

Bernard