education teaching computing to everyone by ghkgkyyt



    doi:10.1145/1506409.1506420                                                                      Mark Guzdial

teaching computing
to everyone
Studying the lessons learned from creating high-demand
computer science courses for non-computing majors.

               evera l    com pu t i n g      pro-      dents, but couldn’t decide who should        everyone’s interest to ensure the class
                     in the U.S. are de-
               g ra m s                                 teach it. The creation of the College of     is good.
              veloping new kinds of                     Computing in 1990 answered the ques-             The class received significant fac-
              introductory computing                    tion of whose job it was to teach com-       ulty interest and used innovative cur-
              courses for non-comput-                   puter science at Georgia Tech. Faculty       ricula. We started out using Shack-
ing majors, some with support from                      in the Ivan Allen College of Liberal Arts    elford’s pseudocode approach to
the NSF CPATH program. At Georgia                       (and in other colleges) embraced the         learning.6 Faculty in the other majors
Institute of Technology (Georgia Tech),                 new requirement. Computing was in-           complained about students not gain-
we are entering our 10th year of teach-                 creasingly relevant for their disciplines,   ing experience debugging programs.
ing computing to every undergradu-                      and was a value-added requirement for        We later moved to Felleisen et al.’s
ate on campus. Our experience gained                    their graduates. The campus adminis-         How to Design Programs text using
during the last decade may be useful to                 tration was kept abreast and involved        Scheme.4 These were, and are, ap-
others working to understand how to                     throughout to maintain support. The          proaches for teaching computing that
satisfy the growing interest in comput-                 new general education requirement            have been successfully used at many
ing education across the academy.                       was defined as an outcome—students           institutions.
                                                        would be able “to make algorithmic               By 2002, however, CS1321 may have
computing in General education                          and data structures choices” when writ-      been the most hated course on cam-
In fall 1999, the faculty at Georgia Tech               ing programs. That simple phrase de-         pus. From 1999 to 2002, the overall
adopted a requirement that all stu-                     scribes a serious introductory course.       success rate (leaving the course with
dents must take a course in computing.                                                               an A, B, or C—not counting those stu-
We modified the academic year from                      teaching everyone in one class               dents who received a D, a failing grade,
quarters to semesters, which gave the                   For the first four years of the require-     or withdrew from the course) was 78%.
campus the opportunity to rethink the                   ment, only a single class met the re-        That’s not too bad for an introduc-
curriculum and our general education                    quirement: CS1321. There were sev-           tory computing course.1 However,
requirements. Russ Shackelford, Rich                    eral reasons for having only a single        this was a course with everyone in it.
Leblanc, Kurt Eiselt, and the College                   course. While we were already teaching       When we examine those majors where
of Computing’s then-dean, Peter Free-                   approximately two-thirds of the stu-         a computing requirement is atypical,
man, convinced the rest of Georgia Tech                 dents at Georgia Tech (because several       we see 46.7% of architecture students
that all students who graduated from an                 of the largest degree programs already       succeeding each semester, 48.5% in
Institute of Technology should know                     required computing), teaching every-         management, and 47.9% in public
computing. We started before publica-                   one on campus meant well over 1,200          policy. We failed more than half of the
tion of the National Research Council                   students a semester. The immensity of        students in those majors each semes-
report Being Fluent with Information                    the task was daunting—splitting our          ter; females failed at nearly twice the
Technology,3 though that report signifi-                resources over several courses seemed        rate of males. Statistics like these are a
cantly influenced implementation.                       a bad start-up strategy. We were also        concern for both the Georgia Tech and
   The new requirement wasn’t a hard                    explicitly concerned about creating a        the College of Computing—it hinders
sell. Faculty in the College of Engineer-               “service ghetto.” Courses just offered       our relations with the rest of campus
ing had wanted to implement a pro-                      as a “service” get less attention. By put-   when computing is the gatekeeper
gramming requirement for their stu-                     ting all students in one class, it is in     holding back their students.

1   comm unicatio ns o f the ac m   | m ay 2009 | vo l . 5 2 | no. 5

Developing contextualized                                                                  concatenation by splicing sounds. We
computing education                                                                        were able to cover all the introductory
Around this time, several studies         We chose to teach                                computing concepts using media ex-
were published critiquing computing       computing in terms                               amples. In their homework, students
courses, including the AAUW’s Tech-                                                        created pictures, sounds, HTML pages,
Savvy report2 and Unlocking the Club-     of practical domains                             and animations. We created an inte-
house by Margolis and Fisher.5 These      (a “context”) that                               grated development environment that
reports describe students’ experiences                                                     provided the media functions as well
in computing as “tedious,” “asocial,”     students recognize as                            as tools for inspecting pictures and
and surprisingly, “irrelevant.” A 2002    important.                                       sounds.
task force, chaired by Jim Foley, found
similar issues at Georgia Tech. How                                                        impact of contextualized
could computing be “irrelevant” when                                                       computing education
it pervades so much of our world? Per-                                                     Faculty and students are happier with
haps the problem was that our course                                                       the new courses. The success rates rose
had little connection to the computing    students in the colleges of liberal arts,        above 80% in both the engineering and
in our students’ world. While students    architecture, and management using a             media courses. When comparing suc-
are amazed at the Web, handheld video     context of manipulating digital media.           cess rates to those same majors men-
games, and smartphones, most intro-          The engineering course was de-                tioned previously, we found the average
ductory courses introduce students to     veloped jointly with faculty from the            success rate in the first two years for ar-
the computing concepts behind these       schools of aerospace, civil, mechani-            chitecture students rose to 85.7%, man-
wonders with Fibonacci numbers and        cal, and chemical engineering. Several           agement to 87.8%, and public policy to
the Tower of Hanoi. What students saw     faculty members in these schools had             85.4% per semester. The media com-
as computing was disconnected from        already started developing an alterna-           putation course has been majority fe-
what we showed them in our comput-        tive to CS1321, using MATLAB, a com-             male, and women succeed at the same
ing class.                                mon programming language in engi-                or better rates than the male students.
    We adopted an approach that we call   neering. Their model involved small              Similar improvements in success rates
contextualized computing education.       classes in a closed lab working on real          in media computation courses have
We chose to teach computing in terms      engineering problems. That course                been seen among underrepresented
of practical domains (a “context”) that   was prohibitively expensive to ramp              groups at other campuses.7
students recognize as important. The      up to over 1,000 engineering students               New opportunities appear on cam-
context permeates the course, from        each semester. The engineering faculty           pus when all students succeed at com-
examples in lecture, to homework as-      worked with David Smith of the College           puting. We have introduced a minor
signments, and even to the textbooks      of Computing to create a course that             in computer science. We had enough
specially written for the courses. We     used their examples and MATLAB, but              students interested in computing after
decided to teach multiple courses, to     taught the same computing concepts               the media course that we now offer a
match majors to relevant contexts.        as CS1321.                                       second course, on data structures with-
    In spring 2003, the College of Com-      The course around “media compu-               in a media context. A second course
puting began offering three different     tation” was built with an advisory board         was also developed for engineering
introductory computing courses. The       of faculty from the colleges of liberal          students, so we now teach three second
first was a continuation of CS1321,       arts, architecture, and management.              computing courses, as well as three in-
aimed at computing and sciences ma-       The board’s awareness and support                troductory courses.
jors. The second was a new course for     for the course was important in getting             Faculty in the School of Interactive
students in the College of Engineer-      the course approved as fulfilling the            Computing and the School of Litera-
ing, with much the same content, but      computing requirement in programs                ture, Culture, and Communication (in
in MATLAB and using an engineering        of those colleges. The advisory board            the College of Liberal Arts) now offer
context. The third was a new course for   favored a programming language that              a new joint undergraduate degree, a
                                          was perceived as being easy to learn             bachelor of science degree in com-
                                          but was not associated with “serious”            putational media. The course was de-
By putting all                            computer science. We chose the Py-               veloped because of growing common
                                          thon implementation, over concerns               interest in areas like video games,
students in one class,                    about both Scheme and Java.                      augmented reality, and computer ani-
it is in everyone’s                          Media computation is the context of           mations. While the common research
                                          how digital media tools like Photoshop           interests were clearly the motivating
interest to ensure the                    and GIMP work. We created cross-plat-            factor in deciding to create the new de-
class is good.                            form libraries to manipulate pixels in           gree program, having a media compu-
                                          a picture and samples in a sound. We             tation course that could draw students
                                          taught, for example, iterating across            into the new program from liberal arts,
                                          an array by generating grayscale and             as well as from computing, facilitated
                                          negative versions of an image and array          the joint effort.

                                                                       mo n t h 2 0 0 9 | vo l. 0 0 | n o. 0 0 | c om m u n ic at ion s of the acm   2

     We see an increasing number of
  courses around campus that require
  students to write programs, though                     Building successful,
  not necessarily as an outcome of the                   high-demand courses
  computing requirement. Computing
  is growing in importance in all fields.                for non-computing
  Non-computing faculty request us to                    majors gives us a
  include particular concepts or tools in
  the introductory courses and to pro-                   different perspective
  vide prerequisite knowledge and skills                 on the current
  for advanced courses. In this way, the
  computing requirement has become                       enrollment crisis.
  part of curricula across campus.
     In the first years, the success rates
  for the new courses were sometimes
  higher than the success rate in the con-               in later courses and in their students’
  tinuing CS1321. We realized that even                  future professions. Further, we need
  computer science majors need intro-                    them as context informants as we de-
  ductory courses that connect explicitly                velop courses that teach through ex-
  to a context that students recognize as                amples from their domains.
  computing. In a joint effort with Bryn                    Finally, building successful, high-de-
  Mawr College and with funding by                       mand courses for non-computing ma-
  Microsoft Research, we launched the                    jors gives us a different perspective on
  Institute for Personal Robotics in Edu-                the current enrollment crisis. Students
  cation (IPRE, http://www.roboteduca-                   want these courses. Other schools on to develop a new introductory                campus want to collaborate with us to
  course that uses robotics as the context               build even more contextualized class-
  for teaching introductory computing.                   es. Our challenge is not just in finding
                                                         more majors, but finding enough fac-
  Lessons Learned                                        ulty time to develop and teach these
  We in the College of Computing be-                     courses that bring real computing to a
  lieve the use of contextualized comput-                wide range of students.
  ing education has been a significant
  step in making Georgia Tech’s univer-
  sal computing requirement success-                     1.	 Bennedsen,	J.	and	Caspersen,	M.E.	Failure	rates	in	
  ful. Developing contextualized courses                     introductory	programming.	ACM SIGCSE Bulletin 39,	2	
                                                             (2007),	32–36.
  is challenging and expensive (for ex-                  2.	 Commission	on	Technology,	Gender,	and	Teacher	
                                                             Education.	Tech Savvy: Educating Girls in the New
  ample, writing textbooks, developing                       Computer Age,	American	Association	of	University	
  new integrated development environ-                        Women,	2000.
                                                         3.	 Committee	on	Information	Technology	Literacy,	
  ments), but the results can be shared.                     National	Research	Council.	Being Fluent with
  Other campuses are adopting our con-                       Information Technology.	The	National	Academies	
                                                             Press,	1999.
  textualized approaches, and some are                   4.	 Felleisen,	M.,	Findler,	R.B.,	Flatt,	M.,	and	Krishnamurthi,	
  developing their own.                                      S.	How to Design Programs: An Introduction to
                                                             Programming and Computing.	MIT	Press,	2001.	
     We recommend involving faculty                      5.	 Margolis,	J.	and	Fisher,	A.	Unlocking the Clubhouse:
  from the other departments in build-                       Women in Computing.	MIT	Press,	2001.	
                                                         6.	 Shackelford,	R.L.	Introduction to Computing and
  ing courses for non-major students.                        Algorithms.	Addison	Wesley,	1997.
  They understand their students’ needs                  7.	 Sloan,	R.H.	and	Troy,	P.	CS	0.5:	A	better	approach	
                                                             to	introductory	computer	science	for	majors.	ACM
                                                             SIGCSE Bulletin 40,	1	(2008),	271–275.

                                                         Mark Guzdial (	is	a	professor	
  Developing                                             in	the	College	of	Computing	at	Georgia	Institute	of	
                                                         Technology	in	Atlanta,	GA.	
  contextualized                                         Copyright	held	by	author.

  courses is
  challenging and
  expensive, but
  the results can be
  3   communicatio ns o f the ac m   | m ay 2009 | vo l . 5 2 | no. 5

To top