A media computation course for non-majors.

Full text

Page 1

A Media Computation Course for Non-Majors
Mark Guzdial
College of Computing
Georgia Institute of Technology
801 Atlantic Drive
Atlanta, GA 30332-0280
guzdial@cc.gatech.edu
ABSTRACT
Computing may well become considered an essential part of
a liberal education, but introductory programming courses
will not look like the way that they do today. Current CS1
course are failing dramatically. We are developing a new
course, to be taught starting in Spring 2003, which uses com-
putation for communication as a guiding principle. Students
learn to program by writing Python programs for manipu-
lating sound, images, and movies. This paper describes the
course development and the tools developed for the course.
The talk will include the first round of assessment results.
Categories and Subject Descriptors
K.4 [Computers and Education]: Computer and Infor-
mation Sciences Education
; H.5.1 [Information Interfaces and Presentation]:
Multimedia Information Systems
General Terms
Experimentation,Design
Keywords
Multimedia, programming, non-majors
1.ISSUES WITH CURRENT INTRODUC-
TORY CS COURSES
Computer science departments are not currently success-
ful at reaching a wide range of students who are taking intro-
ductory computer science. The evidence for this statement
includes international studies of programming performance
[10], declining retention rates [5], and failure rates sometimes
as high as 30% [17]. In particular, participation of women
in computer science is dropping. Studies suggest that com-
puting courses are seen as overly-technical and avoiding re-
lationships to real applications [9], and are frankly boring
and lacking opportunity for creativity [1].
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior specific
permission and/or a fee.
ITiCSE’03, June 30–July 2, 2003, Thessaloniki, Greece.
Copyright 2003 ACM 1-58113-672-2/03/0006 ...$5.00.
At Georgia Institute of Technology (“Georgia Tech”), all
students are required to take an introductory course in com-
puting, including programming skills. The traditional course
is undoubtedly one of the most unpopular courses on cam-
pus, especially among those not in explicitly computing-
related fields.While this is certainly a problem for the
College of Computing at Georgia Tech (where the course
has its academic home), it points towards a larger problem
for the field.
Alan Perlis in April 1961 made perhaps the first argument
that programming should be part of a liberal education for
all students. If Calculus is the study of rates, and that’s
important enough to be part of the liberal education, then
so should computer science. Perlis argued that computer
science is the study of processes, which is certainly relevant
to even more fields than those concerned with rates. The
argument has been echoed and strengthened over the inter-
vening years—by Seymour Papert arguing for a program-
ming as a way of learning about learning [13][14], to Andrea
diSessa’s arguments for “computational literacy” as a criti-
cal component of many fields [3]. As long as non-CS-majors
have such a dislike for computing, the hope is diminished
for computer science as an accepted part of a liberal edu-
cation and for computing generally to meet its potential for
intellectual impact across the range of disciplines, not just
in computational science and engineering.
We are developing a new course Introduction to Media
Computation around a theme of computation for communi-
cations. The premises and core concepts of the proposed
course are:
• All media are being published today in a digital for-
mat.
• Digital formats are amenable to manipulation, cre-
ation, analysis, and transformation by computer. Text
can be interpreted, numbers can be transformed into
graphs, video images can be merged, and sounds can
be created. We call these activities media computation.
• Software is the tool for manipulating digital media.
Knowing how to program thus becomes a communica-
tions skill. If someone wants to say something that her
tools do not support, knowing how to program affords
the creation of the desired statement. If she under-
stands what her tools are doing, she may become a
more adept practitioner, and more capable of trans-
ferring knowledge between tools.
• Core computer science concepts can be introduced through

Page 2

media computation. We introduce these concepts as
answers to questions that the students naturally de-
velop when working with the media programs. For ex-
ample, programs can get large and cumbersome. Ab-
straction is our tool for managing program complexity
and allowing programs to become even larger yet more
flexible. However, computing has limitations. There
are some programs that cannot complete in our life-
time, and knowing that these limitations exist is im-
portant for technological professionals.
This paper describes the course, its approach, and the
technological materials being developed for it. The presen-
tation will also include results of the pilot course evaluations.
2.CURRENT STATE IN DEVELOPMENT
OF“INTRODUCTIONTOMEDIACOM-
PUTATION”
CS1315 Introduction to Media Computation was offered
for the first time in Spring 2003 to a pilot class of 120 stu-
dents. We will iterate on the course during Summer 2003
and implement at a full-scale in Fall 2003, with two sections
of 120 students and three sections in Spring 2004.
Our learning objectives are:
• Students will be able to read, understand, and make
functional alterations to small programs (less than 50
lines) that achieve useful communications tasks. Note
that this is a very different goal than being able to
write 50 lines from scratch. We see these students as
developing tool building skill, not software development
skill.
• Students will appreciate what computer scientists do
and the key concerns of that field that relate to stu-
dents’ professional lives.
– Students will recognize that all digital data is an
encoding or representation, and that the encoding
is itself a choice.
– Students will understand that all algorithms con-
sist of manipulating data, iteration (looping), and
making choices — at the lowest level, these are
choices about numbers, but we can encode more
meaningful data in terms of those numbers.
– Students will recognize that some algorithms can-
not complete in reasonable time or at all.
– Students will appreciate some differences between
imperative, functional, and object-oriented ap-
proaches to programming.
– Students will appreciate the value of a program-
ming vs. direct-manipulation interface approach
to computer use and will be able to describe situ-
ations where the former is preferable to the latter.
• Students will be able to identify the key components
of computer hardware and how that relates to software
speed (e.g., interpretation vs. compilation)
• Students will develop a set of usable computing skills,
including the ability to write small scripts, build graphs,
and manipulate databases – not necessarily using the
common tools, but in a manner that exposes concepts
and enables future learning.
We have developed (and are continuing to develop) a set
of course notes and lecture slides that support the course.
Overall, the course is designed to meet the “Imperative
First” CS1 general structure and requirements in the new
ACM/IEEE Computing Curriculum 2001 [2]. The order of
media covered in the course is arranged to correspond to an
increasing level of complexity in data structures.
• A sound is an array of samples.
• A picture is a matrix (two-dimensional array) of pixels.
• A directory structure (of media files, to process many
files with a single recipe) is a tree of files.
• A movie is an array of matrices (frames, as pictures).
The media thus serve as a way of visualizing and making
concrete (and interesting, we believe) the programs that the
students are writing. Once the students are writing pro-
grams of increasing complexity, we introduce the ideas of
algorithm complexity, object-oriented programming, and re-
cursion as techniques for managing that complexity.
The syllabus for the course walks through each media
type, with some repetition of concepts so that condition-
als and loops can be re-visited in different contexts. Here’s
a rough description of the syllabus.
• Week 1: Introduction to the course and the argument
for why media computation. Introduction to variables
and functions, in the context of playing sounds and
showing pictures.
• Weeks 2–3: Pictures as a media type, including psy-
chophysics (why don’t we see 1024x768 dots on the
screen?), looping to change colors with a simplified for
loop (Figure 1), conditionals to replace specific colors,
then indexing by index numbers to implement mirror-
ing, rotating, cropping, and scaling.
• Weeks 4–6: Sound as a media type, including psy-
chophysics (how human hearing limitations make MP3
compression possible), looping to manipulate volume
(Figure 2), then indexing by index numbers to do splic-
ing and reversing of sounds (Figure 3). Include discus-
sion of how to debug and how to design a program,
as those issues arise. One lecture on additive and FM
sound synthesis.
• Week 7: Text as a media type: Searching for text, com-
posing text, reading text from a file and writing it to a
file. An example program parses out the temperature
from a downloaded weather page.
• Week 8: Manipulating directories. Manipulating net-
works, including making the temperature-finding pro-
gram work from the “live” Web page. Introduction to
HTML.
• Week 9: Discuss media transitions. Moving from sound
to text and back to sound again. Using Excel to ma-
nipulate media after converting it to text.
• Week 10: Introduction to databases: Storing media in
databases, using databases in generating HTML.
• Week 11: Movies: How persistence of vision makes an-
imations and movies possible, generating frames using
the various techniques described earlier in the semester.

Page 3

def greyScale(picture):
for p in getPixels(picture):
intensity = (getRed(p)+getGreen(p)+getBlue(p))/3
setColor(p,makeColor(intensity,intensity,intensity))
Figure 1: An example Jython program using our API to convert a picture to greyscale
def normalize(sound):
largest = 0
for s in getSamples(sound):
largest = max(largest,getSample(s) )
multiplier = 32767.0 / largest
print "Largest sample value in original sound was",
print "Multiplier is", multiplier
largest
for s in getSamples(sound):
louder =
setSample(s,louder)
multiplier * getSample(s)
Figure 2: An example Jython program using our API to normalize sounds to a maximum volume
• Week 12: “Can’t we do this any faster? Why is Pho-
toshop faster than Python?” Introduction to how a
computer works (e.g., machine languge), and the dif-
ference between an interpreter and a compiler. Algo-
rithmic complexity and the limits of computation.
• Week 13: “Can we do this any easier?” Decomposing
functions, modularity, and functional programming (map,
reduce, filter, and simple recursion).
• Week 14: “Can’t we do this any easier?” Introduction
to objects and classes.
• Week 15: “What do other programming languages
look like?” Brief overview of JavaScript and Squeak.
We developed the course in a collaborative process with a
board of faculty advisors from across campus and a team of
undergraduate and graduate students developing materials.
We have been using both on-line and face-to-face forums to
gather input on the course (http://coweb.cc.gatech.edu/
mediaComp-plan). Both student and faculty feedback has
been very positive.
3. TECHNOLOGICALSUPPORTFORTHE
COURSE
The language for the course is Python (http://www.python.
org). Python is a popular programming language used to-
day by companies including Google and Industrial Light &
Magic. It’s most often used for Web (e.g., CGI script) pro-
gramming and for media manipulation. Python was specifi-
cally developed to be easy-to-use, especially for non-traditional
programmers.
The specific version of Python that we are using is Jython
(http://www.jython.org). Jython is an implementation of
Python in the popular programming language Java. Any-
thing that one can do in Java (e.g., servlets, database pro-
gramming via JDBC, GUI programming via Swing) can be
done in Jython. Jython is Python—learning one is the same
as learning the other.
We chose Jython in order to enable cross-platform multi-
media manipulation. We have written a set of Java classes
that encapsulate the kind of multimedia functionality that
our examples require, as well as a set of Jython classes
that provide a simple and useful API to those functional-
ities. The API was designed based on existing literature
on challenges that students find in learning to program,
e.g., we allow set-based manipulation of samples and pix-
els before more complex and general iteration structures are
learned [11][12].
Our API allows for access to the samples that make up
sounds and the pixels that make up pictures.
• Figure 1 is an example program using our API that
converts a picture object to greyscale. It computes the
intensity of a given pixel by averaging the red, green,
and blue components, and then replaces the color of
that pixel with a gray pixel (red, green, and blue com-
ponents the same) with the same intensity. Notice that
the loop in this example is phrased as a set operation—
essentially, “for every pixel p in the pixels of the given
picture, do....” Some research on novice programming
suggests that this is a simpler concept to begin with,
as a way of easing into iteration [11].
• Figure 2 is a program that normalizes sounds to a
maximum volume, by searching for the largest sam-
ple, computing a multiplier so that that sample would
reach the maximum amplitude, and then multiplies
all samples in the sound to raise the amplitude of the
overall sound. We continue to use the simpler form of
iteration here, but using multiple loops—an increase
in complexity.
• Figure 3 takes a filename, then returns the sound in
that file in reverse. Here, we use a more conventional
for loop, with explicit indices in order to copy the
array elements correctly.
We have also created a set of tools to support the stu-
dents’ tasks in this course. Our first and immediate need

Page 4

def backwards(filename):
source = makeSound(filename)
target = makeSound(filename)
sourceIndex = getLength(source)
for targetIndex in range(1,getLength(target)+1):
sourceValue = getSampleValueAt(source,sourceIndex)
setSampleValueAt(target,targetIndex,sourceValue)
sourceIndex = sourceIndex - 1
return target
Figure 3: Return the sound in the file backwards
was for some kind of development environment. Jython is
a new language [15], so most developers simply use plain
text editors, or make do with Python or Java development
environments. We believe that non-CS major freshmen re-
quire more support. A team of undergraduate senior design
students created our tool for students, JES (Jython Envi-
ronment for Students) as a simple editor and program ex-
ecution IDE (Figure 4). JES runs identically on Windows,
Macintosh OS X, and Linux systems.
Figure 4: JES: Jython Environment for Students,
with a graphics example running
We also realized that our students would need to visualize
and explore media and to prepare media for use in their pro-
grams. For example, students want to look at sounds using
a variety of visualizations, record their own sounds, investi-
gate the RGB values in pictures of their choosing, and burst
MPEG movies into folders of JPEG frames for ease in ma-
nipulation. By using their own media, we make the student
programming assignments into a creative activity, and thus,
make it more attractive to women and others dissuaded by
the stereotype of computer science as non-creative [9][16].
Another team of undergraduate students have modified the
media tools in Squeak [4][6] to create cross-platform media
exploration and manipulation tools (Figure 5).
Figure 5: MediaTools: Movie tools (ul), image tools
(ur), sound editing (ll), and sound views such as a
sonogram (lr)
We used our CoWeb collaboration tool to support a col-
laborative experience for students in the Media Computa-
tion course. The CoWeb has been used successfully in a
variety of classes, including computer science. By encourag-
ing students to share their creative artifacts via the CoWeb,
we further erode the perspective of computer science as a
loner, non-creative activity. Further, we used the CoWeb to
support student help-seeking, e.g. asking questions about
the multimedia assignments. Such support may be critical
to the success of the project. A factor analysis considering a
range of variables influencing CS1 success, with completion
as an outcome variable, suggests that comfort asking ques-
tions is the most critical factor for succeeding in CS1 [18].
Findings suggest that Web-based collaboration tools encour-
age much greater participation and comfort than in-class
questions [7][8].
4.ASSESSMENT RESULTS SO-FAR
Our retention results have been quite positive. 120 stu-
dents enrolled for the course, 2/3 female. Only two students
withdrew from the course.
We have been using initial, midterm, and final surveys to
get impressions from across the course. We have used the

Page 5

same surveys at the same time with a section of our tradi-
tional CS1. On the midterm survey, 97% of the students in
the Media Computation course agreed with the statement
“Are you learning to program?”
our traditional CS1 course. When asked what students like
about the class, the students affirm that we’re succeeding at
creating a course that students recognize its applicability,
even among non-CS majors: (All of the quotes below are
from female students.)
compared with 88% in
• “I like the feeling when I finally get something to work.”
• “Very applicable to everyday life.”
• “I dreaded CS, but ALL of the topics thus far have
been applicable to my future career (& personal) plans–
there isn’t anything I don’t like about this class!!!”
• “When I finally get a program to work like I want it
to.”
• “The professor answers questions in class and online
and is concerned about our success in the class. He also
seems to understand that most of us are not engineers
and most likely won’t do straight programming in the
future- just the way of thinking is important.”
• “Collaboration! If I can’t figure it out, I can ask for
help.”
When we ask students “What is something interesting,
surprising, or useful that you learned?” we found that stu-
dents were buying into the relevance of the course and even
finding the computer science interesting (again, all female
respondents):
• “The most useful things I have learned are the basics
about computers and pictures/sound. I think when we
learn HTML–that will be interesting and useful to real
life applications.”
• “Just general concepts about programming. It’s pretty
logical, sort of like in math, so it’s understandable.”
• “Programming is fun and ANYONE can do it!”
SUMMARYANDFUTUREDIRECTIONS
We believe that programming and computation will be-
come part of a general, liberal education, but computing
courses will have to change to make this happen. We are
exploring a media computation approach that we feel will
appeal to a liberal arts major, while still retaining a focus
on programming. Our assessment explores how well these
students learn, but also how motivated they are. A key ques-
tion for us is whether these students will have a continuing
interest in learning about computation, a critical goal of any
introductory course.
5.
6.REFERENCES
[1] AAUW. Tech-Savvy: Educating Girls in the New
Computer Age. American Association of University
Women Education Foundation, New York, 2000.
[2] ACM/IEEE. Computing curriculum 2001.
http://www.acm.org/sigcse/cc2001 (2001).
[3] diSessa, A. Changing Minds. MIT Press, Cambridge,
MA, 2001.
[4] Guzdial, M. Squeak: Object-oriented design with
Multimedia Applications. Prentice-Hall, Englewood,
NJ, 2001.
[5] Guzdial, M. Summary: Retention rates in cs vs.
institution. Message posted on ACM SIGCSE
moderated members list, Georgia Tech, April 23 2002.
[6] Guzdial, M., and Rose, K., Eds. Squeak, Open
Personal Computing for Multimedia. Prentice-Hall,
Englewood, NJ, 2001.
[7] Guzdial, M., and Turns, J. Effective discussion
through a computer-mediated anchored forum.
Journal of the Learning Sciences 9, 4 (2000), 437–470.
[8] Hudson, J. M., and Bruckman, A. Irc francais: The
creation of an internet-based sla community.
Computer Assisted Language Learning (CALL) 15, 2
(2002), 109–134.
[9] Margolis, J., and Fisher, A. Unlocking the Clubhouse:
Women in Computing. MIT Press, Cambridge, MA,
2002.
[10] McCracken, M., Almstrum, V., Diaz, D., Guzdial, M.,
Hagan, D., Kolikant, Y. B.-D., Laxer, C., Thomas, L.,
Utting, I., and Wilusz, T. A multi-national,
multi-institutional study of assessment of
programming skills of first-year cs students. ACM
SIGCSE Bulletin 33, 4 (2001), 125–140.
[11] Miller, L. A. Programming by non-programmers.
International Journal of Man-Machine Studies 6
(1974), 237–260.
[12] Miller, L. A. Natural language programming: Styles,
strategies, and contrasts. IBM Systems Journal 20, 2
(1981), 184–215.
[13] Papert, S. Teaching children to be mathematicians
versus teaching about mathematics. AI memo no. 249
and Logo memo no. 4, MIT, 1971.
[14] Papert, S. Mindstorms: Children, computers, and
powerful ideas. Basic Books, New York, NY, 1980.
[15] Pedroni, S., and Rappin, N. Jython Essentials.
O’Reilly and Associates, 2002.
[16] Pfleeger, S. L., Teller, P., Castaneda, S. E., Wilson,
M., and Lindley, R. Increasing the enrollment of
women in computer science. In The Proceedings of the
Thirty-second SIGCSE Technical Symposium on
Computer Science Education, R. McCauley and
J. Gersting, Eds. ACM Press, New York, 2001,
pp. 386–387.
[17] Roumani, H. Design guidelines for the lab component
of objects-first cs1. In The Proceedings of the
Thirty-third SIGCSE Technical Symposium on
Computer Science Education, 2002, D. Knox, Ed.
ACM, New York, 2002, pp. 222–226.
[18] Wilson, B. C., and Shrock, S. Contributing to success
in an introductory computer science course: A study
of twelve factors. In The Proceedings of the
Thirty-second SIGCSE Technical Symposium on
Computer Science Education, R. McCauley and
J. Gersting, Eds. ACM, New York, 2001, pp. 184–188.