Second Order Metaprogramming and 'hasIntrinsic'
- (previously Copyright David Dodds 2008)
Second Order Metaprogramming and 'hasIntrinsic'
(referring to my previous post) --- Bob is a person (actually Bob is
name of Person, so it is more accurate to say Bob is a name, Bob is a
string value instance of name.) Exactly the same thing could be said
about Mary. The only thing which makes them not exactly identical is
that those two strings do not have exactly the same collating order.
Perhaps one might want to have a predicate such as 'hasIntrinsic'.
Bob 'hasIntrinsic' a temperature, a volume/morphology/topography, a
gender, etc. Bob may be a student or a professor depending on whether
he has a Phd, but he always has SOME temperature, even if he is dead.
and until he returns to dust (and even then) Bob has a
volume/morphology/topography (aka size).
Because we are people WE know that Bob and Mary are (typically)
different in particular ways, the content of this knowledge is
metadata and is 'commonsense' or 'background knowledge', just like we
all know / expect that unsupported objects 'fall', and that all
'living things' eventually cease 'living'. Just yet computers do not
take instrument readings, such as TV camera images, and place
'knowledge', obtained / derived from such readings of the world, into
knowledgebases and databases, by (means of) themselves (the computer
selves). Right now the computers require humans to interpret the
instrument outputs and produce / populate the knowledge , which the
humans place into knowledgebases like ontologies. What will the world
be like once we have figured out how to program computers to
consistently and adequately perform these actions (for) themselves?!
Parts of this are not so far away from reality already. For example,
that collection of fancy beach sand in a box (aka computer) can
already take a streamed TV video signal and amidst all the stuff "on"
the screen, pick out a particular face, in a crowd. Granted this is
not a conscious act on the part of the computer, but the point is that
the technology is here which allows us to describe the visual
morphology of things (faces for example) and how to locate /
differentiate them in a crowded natural scene. By associating
ontologies with these programs one can add semantics ("meaning") to
the things in the picture and the meaning of the scene the picture
"shows", such as "the crowd in Grand Central Station milling around.."
If someone mills in a way deviating from the predefined milling
pattern then robo-dobermans can be dropped out of the ceiling to deal
with it. No deviated preversions are permitted in GCS.
Using programs like "flocking" and "schooling" algorithms to model
group traffic dynamics it is possible to use such models as
descriptions of 'standard' or 'typical' (transportation) behaviour.
If there is a reasonably finite set of 'milling around' patterns which
can be decently described via algorithms (or other) then these could
be used to evaluate scenes such as GCS crowds, detecting 'unusual' or
'atypical' 'milling behaviour', such as putting a briefcase or other
package onto the floor and then walking or otherwise moving away from
that location without bringing the package with one. Of course the
computer would also have to have some (perhaps algorithmic)
'understanding' of throwing / dropping / putting garbage into a
garbage recepticle (or even, gurk) onto the floor ('the ground'). It
would also help those ontological systems if something like OASIS' HML
(Human Markup Language) were included for it provides a focus for
interpreted actions to be compared against. Equal vs not equal is
pretty much all the computer can do well, similar(ity) being
notoriously difficult to "explain to a computer" / program. In this
way, using HML terms, such as "deceipt" and associating the term with
transportation patterns (milling) and "object caretaking/stewardship"
(abandonment/retention) the computer / camera can be operated in
'scene serenity failure mode'.
In fact Edward de Bono, in his groundbreaking book "Atlas of
Management Thinking", clearly has notions in mind (similar to that
"human taxonomy") in the "working" of his diagrams. The book reader's
cognitive system must be monitoring / watching for a particular set of
notions depicted in the diagrams of the book. An example of one such
notion is motion (*), and that is suggested / depicted in a cartoon
drawing by having two or three curved lines pointing in the direction
of motion "behind" the object in the cartoon which is "moving". A
thrown baseball in the air has those curved lines "behind" it (and,
yes, sometimes in front too), and we viewers (of the cartoon /
drawing) cognize those lines as (meaning) the ball is moving, and not
by coincidence in the direction the curves are pointing. Often this
cognition on our part is mostly subconscious, but the cues are in the
picture to trigger that. (How often does one examine (the content /
components of) one's own conscious?)
* In de Bono's book, AoMT, he uses a dashed line with an arrowhead to
suggest motion along the depicted path. This form of motion depiction
is of a slightly more abstract nature than the wavy lines behind a
moving object. The wavy lines are reminiscent of air waves behind
moving things, which can been seen on bright hot days as shimmer, for
example. The dashed or dotted line is not evoking of air waves but
rather the higher-level or more abstract visual (plan-view) depiction
of a succession of location (point) or placement. Sometimes blurring
is used to represent motion also, such as showing fan blades in one
position and (complete or partial) outlines of the blades in another
position and or curved lines in between the stationary blades.
That we use spatial metaphors in daily life and aren't even conscious
of the (cognitive) metaphor also suggests that <much of our cognitive
life (estimating / recognizing / interpreting, and especially initial
recollection ("that finding something in memory" which is then passed
up to the conscious for "realization") actually occurs in our
subconscious>, witness walking along a non-empty sidewalk and (you)
not crashing into other people or objects there. You are not aware of
all the distance, time, and occupancy estimates occurring and
re-occurring. You dont even have to know where everyone / everything
actually is, only an analog approximation (an estimate from
subconscious visual interpretation handed to our awareness as a
(cognitive not emotional) 'feeling' or 'sense'. A 'sense' of the
distances.) is known and all the while you do not see numbers and
symbols superimposed on your visual input, like in The Terminator, nor
do you deliberate on moving along a non-clear sidewalk, it just
'happens' 'automatically' 'without thinking (about it)'.
In fact, it would be an intellectually marvelous feat if computers
could "look at" a cartoon / drawing and "understand" it. Perhaps the
computer's understanding of a page in "Mad Magazine" would be even
more entertaining than the page itself.