Information Technology Reference
In-Depth Information
As a simple example diagram, Fig.
1.3
b shows the progression from poetry gen-
erator version P1 to P2. In the first version, there are two process stacks, hence the
system works in two stages. In the first, the software produces some example poems,
and in the second the user chooses one of the poems (to print out, say). The first
stack represents
two
timesteps in development, namely that (a) the programmer had
a creative act
whereby he/she came up with a concept in the form of some
code to generate poems, and (b) the software was run to produce poems in creative
acts of the form
<
C
g
>
E
g
>
∗
. The second stack represents the user coming u
p w
ith
an idea for an aesthetic, e.g., pref
er
ring lots of rhyming, in creative act
<
,
and then applying that aesthetic
a
g
him/hers
e
lf to the examples produced by the
so
ftware, in the
selection
administrative act
<
A
g
>
[
S
(
a
g
(
e
g
))
]
, which maps the aesthetic
a
g
over the generated examples, and picks the best one
.
In the
P2 version of the software, the programmer undertakes the
translation
act
:{
e
g
}ₒ[
0
,
1
]
,
writing code that allows the program to apply the rhyming aesthetic itself, which it
does at the bottom of the second stack in box P2.
Figure
1.3
c shows a progression in the HR automated theory formation system
[
48
] which took the software to a meta-level, as described in [
49
]. HR operates by
applying production rules which invent concepts that categorise and describe input
data. Each
pro
duction rule was invented by the programmer during creative acts of
the type
[
T
(
a
g
)
]
, then at run-time, HR uses the production rules to invent concepts
an
d ex
amples of them in
<
C
p
>
>
∗
acts. In the meta-HR version, during the
<
C
g
,
E
g
<
creative act, the programmer had the idea of getting HR to form theories
about theories, and in doing so, generate concept-invention processes (production
rules) in acts of t
he
form
C
m
>
<
C
p
>
. The programmer took meta-HR's output and
translated it via
into an implemented production rule that HR could use,
which it does at the bottom of the stack in box H2.
[
T
(
C
p
)
]
1.5.3 Comparing Diagrams and Output
Examining the transition from one epoch-level diagram to another should provide
some shortcuts to estimate audience reactions, especially when these are linked to
strong objectives. As with the original FACE model, the diagrams make it obvious
where creative or administrative responsibility has been handed over to software,
namely where an act which used to be barred has become unbarred, i.e., the same
type of generative act still occurs, but it is now per
fo
rmed by software rather than
program
mer.
For instance, this happened when the
S
became an
S
in Fig.
1.3
b and
when the
C
p
became a
C
p
in Fig.
1.3
c. At the very least, in these cases, an unbiased
observer would be expected to project more autonomy onto the software, and so
progress in the strong sense has likely happened. In addition, the diagrams make it
obvious when software is doing more processing in the sense of having more stacks,
bigger stacks or larger tuples of acts in the stack entries. Moreover, the diagrams
make it clear when more varied or higher-level creative acts are being performed by
Search WWH ::
Custom Search