Some thoughts on encodings

[Jonathan Rees]

Suppose that four lab technicians huddled around a plastic tub observe
a litter of mice at the same time, and observe that the litter has six
pups. The technicians independently write in their separate lab
notebooks (a), (b), (c), (d) the following scrawls:

(a) 6 (b) VI

(c) 6 (d) VI

How many specifically dependent continuants (SDCs) are we talking
about here, 4 or 8? That is, do we speak of a "sixness measurement"
SDC inhering in each notebook page, or both a "sixness measurement"
and either a "sixness-encoded-as-roman" or "sixness-encoded-as-arabic"
SDC? Do we need to choose which kind of SDC (encoding or not-encoding)
each one is, or can one SDC be both?

Similarly, we would like to talk about what is "communicated" and how,
using GDCs. We need to talk about encodings (I'm told), so it would
make sense for there to be GDCs for encodings (think choice of mime-
type and character encodings). But we also need to abstract away from
encoding when necessary. In this scenario there are three GDCs of
interest: the one that is concretized by all four SDCs (six-pupness;
the information without regard to encoding), the one that is
concretized by (a) and (c) (six-pupness encoded as arabic), and the
one concretized by (b) and (d).

Now we can ask to what classes do all these individuals belong, and
which classes are asserted and which inferred.

If you don't like this example, try a lab report or data set (the
information without commitment to encoding) and its encodings (pdf,
text, html). The principle is the same.

[Bjoern Peters]

Jonathan Rees wrote:
> Suppose that four lab technicians huddled around a plastic tub observe
> a litter of mice at the same time, and observe that the litter has six
> pups. The technicians independently write in their separate lab
> notebooks (a), (b), (c), (d) the following scrawls:
>
> (a) 6 (b) VI
>
> (c) 6 (d) VI
>
>
> How many specifically dependent continuants (SDCs) are we talking
> about here, 4 or 8? That is, do we speak of a "sixness measurement"
> SDC inhering in each notebook page, or both a "sixness measurement"
> and either a "sixness-encoded-as-roman" or "sixness-encoded-as-arabic"
> SDC? Do we need to choose which kind of SDC (encoding or not-encoding)
> each one is, or can one SDC be both?
>

I thought that there would be 4 SDCs, 2
"sixness-measurement-encoded-as-roman, and 2
sixness-measurement-encoded-as-arabic. Further, I thought that a
sixness-measurement-encoded-as-whatever is_a sixness-measurement.

> Similarly, we would like to talk about what is "communicated" and how,
> using GDCs. We need to talk about encodings (I'm told), so it would
> make sense for there to be GDCs for encodings (think choice of mime-
> type and character encodings). But we also need to abstract away from
> encoding when necessary. In this scenario there are three GDCs of
> interest: the one that is concretized by all four SDCs (six-pupness;
> the information without regard to encoding), the one that is
> concretized by (a) and (c) (six-pupness encoded as arabic), and the
> one concretized by (b) and (d).
>
>

I thought there would only be one GDC, namely what is communicated
(six-pupness), and that encodings are only applied to instances of SDCs.
This is not something I have thought enough about to strongly argue that
this is the right way, but it seems wrong to me to add to the number of
GDCs every time an SDC is created in a different encoding.