RE: Time, Causality and Demand-Pull
Chris,
It seems I have not explained myself properly. So, I will try again, at
least on some key points.
Chris Menzel wrote:
"> ...in the practical real 'interpretation' of time, we do not work
> directly with timepoints.
I don't believe you. If I say that the visioning meeting started at noon
(today) and the planning meeting started at one, I am "working directly"
with the timepoints denoted by "noon" and "one" in this context.
> In information systems we may represent timepoints - but the issue is
> how we interpret this."
This seems to me to make my point exactly. One of the problems is to
differentiate between (re-)representing the way we talk about the world and
representing the way we actually work with the world. Let's say that you say
that the visioning meeting will start at noon (today) and the planning
meeting will start at one. We can represent what you are saying using
timepoints (as I said above) - or we can look at what actually happens. How
do you make sure you turn up at the visioning meeting (how does an engineer
make sure that he starts a process on time)? You set an alarm on your
computer or you check your watch. I am assuming that you do not look for the
noon (today) timepoint.
This point is even clearer if we were to say that the planning meeting will
start immediately after the visioning meeting. You might ask at what
timepoint that will be. And the answer could be, we do not know because we
are not sure how long the visioning meeting is going to be, but when that
finishes the planning meeting will start. If you know that the planning
meeting is immediately after the visioning meeting and in the same room, you
do not even need to worry about timepoints - you just stay in the same room.
Yet the PSL axiomatization implies that all 'before' relationships between
occurrences must be via timepoints - that they cannot be directly related.
Many practical processes are managed without any mention of timepoints. My
question is why does PSL force timepoints to be used for all processes.
There may well be a good answer, I just have not been given it.
Chris Menzel wrote:
"> It is true that you can have different axiomatisations of Euclidian
> geometry, and that no one is right - but it is also true you can have
> different interpretations. For example, I understand, that you can
> have non-standard interpretations of these axioms which give you
> non-Euclidian geometries.
Only if you leave out the parallel postulate. Non-Euclidean geometries
-- the familiar ones, anyway -- arise by including axioms that are
inconsistent with the parallel postulate in different ways. (But this
niggling point doesn't affect your line of argument.)"
I think your parallel postulate point only holds for standard
interpretations. I seem to remember a seminar where it was shown that if, in
the geometry of a surface of a sphere, you interpret the lines of latitude
as parallel (a non-standard interpretation) then you have a model of the
axioms of Euclidiean geometry.
Chris Menzel wrote:
"> Two systems that use, say, PSL but with widely differing
> interpretations are not going to be as easily interoperable as two
> systems that use the same interpretation.
I'm not sure how that could be, as interoperability is accomplished
through the axioms (at least the way I, and I think many others,
conceive of interoperability). Interpretations are typically infinite
objects that are not representable in a computer. What you say *could*
be a problem only if a given axiomatization was inadequate in some way
that did not permit adequate interoperability, and users had to fill in
the missing details for themselves. They might then fill in the gaps
differently from someone with a different interpretation. But the
solution to *that* problem is a better axiomatization at the start."
Chris Menzel wrote:
"> A natural interpretation of PSL seems to suggest that there are
> individuals which it calls timepoints. Now it is not clear to me what
> these are.
Take them to be whatever you want, so long as they form a structure that
satisfies the axioms. You seem to be asking a philosophical question
about which PSL Does Not Care."
It seems to me odd that you cannot give at least one example of an
interpretation of these timepoints. It seems to me an eminently practical
matter to know how to interpret a primitive like this.
Chris Menzel wrote:
"> Like PSL let's start "with activities and activity occurrences. One
> way in which these differ is that activity occurrences are temporally
> extended entities, while activities are intuitively at the "type"
> level." Lets also try and formalise the notion that one temporally
> extended activity occurrence can be before another - or more
> accurately that the end-of one can be before the begin-of the other -
> and call this the 'before' relation.
I'm not sure I understand what relation "this" is picking out here. It
sounds like you are talking about a relation between timepoints --
the end-of occurrence e is *before* the begin-of occurrence e'. And
begin-of and end-of map occurrences to timepoints. But that doesn't
seem what you have in mind below...
No, as you have guessed I am not talking about timepoints - or time
intervals. The issue is Liebnitz's - about whether it is practically better
to 'construct' timepoints and time-intervals out of more 'fundamental'
relations. Another way of thinking about my 'before' relation would be to
say activity occurrences have a temporal end-boundary - what could be
thought of as an end event (if they have temporal extents they have temporal
boundaries). This cannot be a timepoint in a PSL sense as two disjoint
activity occurrences can have temporal end-boundaries that are neither
before nor after each other, but are not identical. This, I think, would
capture the intuition behind the way we talk about many processes without
recourse to timepoints.
Hopefully this is enough to give a clearer view of my position.
Chris
Chris Partridge
LADSEB-CNR partridge@ladseb.pd.cnr.it
Corso Stati Uniti, 4 Voice: +39 049 829 5715
I-35127 Padova Fax: +39 049 829 5763
Italy