John F Sowa wrote:
Bruce Schuman wrote:
>> In science and
engineering, identity is *never* observable.
>> Similarity is
observable, and identity is *always* an inference.
>
> But in what way, and how
constructed?
If you think about perception,
ask what it would mean to observe identity. We observe qualities and
patterns. We recognize that some new experience is similar to some
previous experience. But there are no unique ids on them. We just
have to rely on our memories and try to decide (i.e., make inferences) whether
two things are "the same".
Or, you could say that the ID is the concatenated value
of all properties in a specific order, such as primary keys using all columns
in RDBs. If two objects have the same ID with that representation, then
they are the same object, i.e., identical, by construction of the primary
key. So similarity is identity with that sort.
Then, step by step, reducing the range of values of each
column in the primary key leads eventually to the minimal range of each primary
key column, which leads to the discovery of empirical knowledge about the
objects in that category.
But in human perception terms, a la Peter Gardenfors' conceptual
spaces based on perception taxes, we distinguish among objects and patterns based
on the way experience wires our heads. Messily to say the least.
> what is the relationship
between a "real object" and an "abstract
> object" (a symbolic construction
that represents the real object, in
> selected dimensions/aspects
that are thought to be significant in the
> present context).
Fundamental distinction (with
credit to C. S. Peirce):
1. Reality is what we
live in. It's what we perceive and act upon.
Things
and processes are chunks of reality that we experience,
analyze, use, and act upon in ways we find desirable.
2. All abstractions are
signs. All thoughts are signs. Some signs
refer
to parts, aspects, or chunks of reality. Some signs refer
to
other signs. And some signs relate signs to signs. The mind
is the
totality of all the signs that anyone has experienced and
accumulated over a lifetime (no matter how short or long).
When indirect addressing was invented, it was hailed as
an abstraction of the machine which would produce great advances. Since
then, I haven't seen it described except in yellowing old manuals in dusty
archives. So I can write this silly function:
sign->sign->sign->sign->sign->sign->sign->sign->sign->sign->sign
and be eleven times as abstract as just plain old "sign".
I suppose that is my elliptical way of wondering if there
is any real value in a long chain of signs that couldn't be constructed with
less obfuscation.
...