[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: <eyebeam><blast> Art and the attention economy



I would like to answer some posting by Simon Biggs about a few concepts
borrowed from computer science, which are IMHO misunderstood and just
follow the current hype. We are all concerned with improper uses of
Derrida/Barthes/D&G, so why not of Turing ?

Those of you not interested in computers, should please skip this
message...

1) Parsing.
Basically, the grammatical analyses of sentences written in some
language, as K-12 kids are doing. That is, syntactic analysis of the
sentence, which is of course of prime importance in computer languages.
Computer programming languages (or "artificial" languages as opposed to
what computational linguists call "natural" language, which is what I am
currently struggling with) are very simple and can be represented  by
generative grammars, following and best exemplifying the Chomskian
revolution. The basic idea of parsing is to start from a linear sequence
of lexical units (a sentence) and to produce a hierarchical structure (a
tree) that represents its grammatical structure and determines what
grammar rules have been used to contruct the sentence. This is not
really  related to Turing machines and what Simon's call "Turings
concept of a self-modifying system". Encoding/decoding schemes are
indeed  used in various ways in the theory of computational complexity
and Turing machine (see for instance "The Universal Turing machine" by
M. Davis), but this would not be called parsing because there is no
underlying grammar. Of course, every code, language, sign system can be
seen as a semiotic system and considered in a unified way. But we are
here interested in the details and in the algorithms at work, which are
very different. Classical parsing algorithms can be found in any
standard CS texbook, eg Aho/Ullman, but this is now a rather dead
research topic. Simon is right in saying that parsing is the first thing
your Web browser load your HTML file, VRML file, javascript code, etc,
in order to extract the structure of the document and perform the
appropriate actions. (Note: in fact the first phase is lexical analysis,
ie aggregating letters into words, and recognizing specific symbols and
keywords).

2) Java
So much hype around Java that it should come out in the Eyebeam forum 
The idea of  what Simon calls "Virtual Computer" did not come with Java.
It was popularized in the 70's with the Pascal Language and the idea of
the P-code, an abstract machine language, portable across computers.
Some people said that the idea of bytecode interpreters was invented at
the Academy of Science in Kiev in the late 50's ...
This idea, followed by Java, is to compile your program (written in the
Java language) into a machine-independent low-level ("byte-code")
language, the Java Virtual Machine language. This code can be
communicated over the net and executed on any other computer that has a
JVM interpreter, as for instance our dear Web browsers (Netscape
Navigator or MS Internet Explorer). We can nowdays consider that every
computer is equipped with a Web browser, *this* is the only reason why
Java programs can be executed everywhere ! Indeed, what I just said is
not so true : if I send you a Java program, your web browser will not be
able to execute it, your browser can only execute JVM code. Therefore, I
first have to compile my Java program into JVM code, with for instance
Sun's JDK, and bla bla bla.

3) Object-Oriented Programming
OOP was *not* primarly used for developing large-scale software such as
W95 ... OOP ideas were first proposed in a language called Simula (late
60's), and generalized in the Artificial Intelligence research community
in the late 70's and 80's, for instance with the Smalltalk language.
(And it has been taken by big compagnies for producing large software
only after, in the late 80's) Now this paradigm evolved in what is
called agent programming, software agents, intelligent agents, etc.
Trendy paradigms in AI for the modelisation of complex systems are now
rather evolutionary computing, genetic programming, Alife and the like. 

Philippe Codognet
SONY Computer Science Laboratory




-
-------------------------------------------------------------
a critical forum for artistic practice in the network
texts are the property of individual authors
to unsubscribe, send email to eyebeam@list.thing.net
with the following single line in the message body:
unsubscribe eyebeam-list
information and archive at http://www.eyebeam.org
Eyebeam Atelier/X Art Foundation http://www.blast.org
-------------------------------------------------------------