03 October 2001: Algorithms and Computational Logic
Leaders: Steve and Tom
JU: Demo of a MOO.
TH: How much to you need to know to "play" a MOO? Do you have to know
anything about programming?
JU: Nothing. In fact, most MOO's control who can actually program, who
can actually build and program. Most MOO's require that you propose a
project or demonstrate some facility before being allowed to program or
build.
SR: Good idea, good pedagogical tool. Create illusion of moving through
space--kind of like wordprocessors, where it seems like the document
extends below the bottom of the window but it really doesn't.
WM: Create illusion on two levels--in code and in English that describes
the "space" that you built.
GR: Can you connect MOO's? Can you make a tunnel?
JU: MOO's can use all common protocols--FTP, HTTP, gopher, etc. But can't
actually "go north" from one MOO to another. Must log into other MOO.
Make all students programmers and builders. Not wizards.
WM: Can you resrtict access to different areas or functions?
JU: Yes. Players could be children of some class that has restricted
access to different functions and rooms.
----SR: I would recommend Chaitin and Knuth for the actual KR course. Chaitin: computer is side effect of a moment of crisis and uncertainty near the beginning of the century. Colburn (Models of Reasoning): object lesson in what computers can (or can't) do.
WM: Difficult to get computers to infer _interesting_ things.
JU: What uninteresting things can computers generate?
WM: Number theory. Create an indefinite number of statements that are symbolically "true" but are the really interesting. Like music. A computer can generate infinite combinations of notes, but how many of them count as "music."
JU: So maybe the challenge is generating inferences with meaning.
SR: First question: Could a computer do literary criticism? Could a computer philosophize? What can a computer do? And what can they not do?
WM: Chaitin talk makes no reference to Babbage and that "other" history of computers. What do you think about that?
SR: Chaitin is revisionist history.
JU: Is the problem that we're lacking the mechanical history?
WM: How much does control structure matter? To change something in Babbage's machine involved
GR: Where does the mechanical and Turing's UC come together?
JD: ENIAC and EDVAC. Stored program. Also need to reinsert the social history of this as well. Need to understand two things: philosophical and social/cultural history?
RG: _The Control Revolution_, James Benniger. Also a history of representation? History of media?
WM: Where these come together. In thinking about what is possible. How do the conceptual tools match the mechanical tools? Maybe we need to think about the distinction between the conceptual and the mechanical. If we want students to think about what is (or is not) possible, then maybe students need to be able to think about computers as conceptual beasts as well.
GR: How is a program a representation of human knowledge? Which articles might help us think about this? Knuth. When did someone decide that we need a meta-language to describe the algorithms?
JD: Boole and DeMorgan? 1840's and 1850's. Point at which abstract, logical language. Meta-language for logical process. Defines finite set of operations for computation. "and" and "not" in Boole but not "if" and "then" until Turing.
Boole: Determine a set of relations between a set of terms. Classificatory sense of logic, of logical relations. Allows manipulation of symbols in systems.
SR: Something about logic at this stage besides the actual sources. Something from history of logic.
WM: Switch from declarative to imperative. Propositional logic statement --> no expressions of process.
JU: It seems like most of these questions are definitional. If we define "programming language" then someone can probably tell you what the first one is. If you define "algorithm" then we can tell you what the first one was. Maybe the important notion here is "what is a computer?"
WM: Rule-based systems and declarative systems--reduce imperative or inferencing part--historical, practical choice. Solution to the semantics problem.
GR: What is the nature of computation and what are its limits? What can you describe formally? Social history question: how did we end up with things like Python?
JD: Still need descriptive historical account of intersections of mathematical formalisms, linguistics, analytic philosophy, programming languages.
JU: Social history of operating systems.
GR: Social history of python from a linguistics perspective?
SR: Sociology of languages. Many programming languages created by committees.
TH: More recently languages have been created by individuals.
JD: Should we tie in Chomsky somewhere too?
GR: In the 80's a movement toward visual programming. Attempts to create languages for kids to use. Didn't work, but there was a literature produced to situate this.
TH: ADA, language to end all languagegs, sponsored by the dept. of defense. Project has now been dropped.
JU: Social history of unix is very interesting. W3C as a social/political history of Web standards. SGML has similar lineage of ADA.
What's the relationship between these mechanisms and the social institutions that fund these projects.
DS: Could different programming languages be understood as different epistemologies?
JM: Knuth article. Each of the terms in the definition of algorithm should be investigated. For instance, what's "definiteness." Wittgenstein writes about what constitutes definiteness. Is language algorithmic? It is. Rhetorics are algorithmic.
JD: Language may be algorithmic but its rules may not be recoverable.
WM: Does a particular programming language allow you to do everything that this machine can do?
JD: Critical epistemology for these things?
JU: Purpose of KR seminar: students think that computer programs are _a_ type of representation of human knowledge.
BR: Teach student important paradigms; let students investigate these.
Teaching students what "religion" is. Read philosophy, anthropology, psychology, history, etc. These are the paradigms. Then they work on a definition in using these paradigms.
GR: Students confronted with defining "computer." They have to define this. Discover possible definitions and possible moments in history.
What is a "program"?
WM: When do we deal with this?
JU: This isn't the only type of knowledge representation. We're talking only about computers here. What about a library card catalog on notecards.
JD: We want to teach students to find problems and ask questions. Certain tangible skills that we want them to have. But we should remember that we aren't trying to be exhaustive.
WM: Are we producing practitioners? What basic skills do these people need to have to be practitioners? Can't let this go.
JU: We want people to understand the problems involved in setting up systems across various knowledge domains.
JU: What would be the exercise for this unit of the class? How do we join the knowing and the doing?
JD: Ambiguity between data and operations. Naively ask the question: now you have to think about x and how you would model it as data and operations.
JU: Take some object of inquiry and model it as data and operations.
GR: Program a Turing machine?
One thing that has to come from this unit is that they need to learn some of the discouse of programming. A vocabulary exercise? Define constant, list, etc.
JU: Back to Steve's initial question: What is computable? Do they need to start thinking about this?
GR: Reading from last week--book on Java--would introduce some basic concepts of programming.
WM: MIX, Knuth.
GR: Modelled the Altair?
JD: To what extent is something programmable. Get back to Knuth's definition of algorithms. How are these process definite, finite, etc.
JM: Early kit computers. Have them put these together?
JD: Maybe this should be a part of the design course.
MOO's and LEGO
___________
Possible first questions:
What is a knowledge representation?
What is knowledge?
What is a computer? ___________________
Things to continue thinking about:
WM: In thinking about what is possible. How do the conceptual tools match the mechanical tools? Maybe we need to think about the distinction between the conceptual and the mechanical. If we want students to think about what is (or is not) possible, then maybe students need to be able to think about computers as conceptual beasts as well.
--
AL: Program a Turing Machine. This could be done on paper. Other students could follow the other students' programs.
WM: Turing too complicated. Try Knuth's MIX.
This archive was generated by hypermail 2b30 : Thu Oct 04 2001 - 08:30:39 EDT