The Challenges of Software Engineering Education

Carlo Ghezzi

Dino Mandrioli

Dipartimento di Elettronica e Informazione

Politecnico di Milano
Piazza L. da Vinci, 32, Milano, 1-20133, Italy

Dipartimento di Elettronica e Informazione

Politecnico di Milano
Piazza L. da Vinci, 32, Milano, 1-20133, Italy

[email protected]

[email protected]

ABSTRACT

* the design methods ofthe discipline;

* the technology and tools of the discipline.
In addition, he/she has to be able to
* keep his/her knowledge current with respect to the new
approaches and technologies;
* interact with other people (often not from the same culture);
* understand, model, formalize, analyze a new problem;
* recognize a recurring problem, and reuse or adapt known
solutions;
* manage a process and to coordinate the work ofdifferent
people;

We discuss the technical skills that a software engineer should

possess. We take the viewpoint of a school of engineering and put
the software engineers education in the wider context of engineering
education. We stress both the common aspects that crosscut all
engineering fields and the specific issues that pertain to software
engineering. We believe that even in a continuously evolving field
like software, education should emphasize principles and recognize
what are the stable and long-lasting design concepts. Even though
the more mundane technological solutions cannot be ignored, the
students should be equipped with skills that allow them to dominate
the evolution of technology.

Categories and Subject Desriptors

K3.2 [Computer and Information Science Education]: Computer

Science Education.

...

Some of these skills are better suited to be taught and leamed at

school, others are fully understood only after some practice in the
real world.
Software engineering differs from the other more traditional fields of
engineering in several specific points, among which:
* Theoretical foundations are less mature, often "far" from
being directly applicable, and hence less immediately useful.
* Well-established models and notations are still lacking,
although much progress has been made. Some of "our"
models are even being absorbed by other engineering fields;
e.g. UML is being used in such fields as logistics or
manufacturing systems.
* It is essential to be able to distinguish between the essential
and the accidental complexity of software [2] and master
different approaches to deal with them.
* More emphasis is needed on interdisciplinary culture and
communication skills due to the very nature of SW

Keywords: Engineering, software engineering, education, model.

1. INTRODUCTION

In this paper, we discuss the technical skills that a software engineer

should possess. Rather than focusing on the organization of a course
(or set of courses) titled "software engineering", we take the global
viewpoint of a school of engineering and put the software engineer's
education in the wider context of engineering education. We believe
that software engineering (SE) should be viewed as an engineering
discipline, and that a software engineer should be viewed as
primarily an engineer. On these grounds, we analyze the principles
and practices that pertain to the engineering culture and also what is
special in the case of SE.
The main dichotomy that we face in engineering is learning by
studying (at school) vs. learning by doing (at work) [1]. Our strong
belief is that even in a continuously and rapidly evolving field like
software, education should emphasize principles and recognize what
are the stable and long-lasting design concepts. Even though the
more mundane technological solutions cannot be ignored, the
students should be equipped with skills that allow them to dominate
the various technological waves without being overwhelmed by
them and thus avoiding the danger of rapid obsolescence.

applications.

As a result perhaps the gap between learning by studying vs.

learning by doing is bigger than in other cases.
Perhaps a consequence ofthis situation is that books and courseware
that aim at teaching practice are often wordy, purely descriptive, and
overly informal. At the same time, books and courseware that teach
mostly theory are often unrealistic, only deal with toy examples, are
not well understood and not well accepted.

2. KNOWLEDGE AND SKILLS

OF A (SOFTWARE) ENGINEER

Like any other engineer, the software engineer must master

the theoretical foundations of the discipline;

3. CONSEQUENCES ON (SOFTWARE)
ENGINEERING EDUCATION

We articulate some general guidelines on teaching software

engineering as an engineering discipline.

Copyright is held by the authors/owner(s).

ICSE 05, May 15-21, 2005, St. Louis, Missouri, USA.
ACM 1-58113-963-2/05/0005.

637

* We should focus on lasting principles rather than last-minute

fashionable technologies/buzzwords. This can be harder for
SE than for other engineering fields because:
o Principles and theory that may not be directly applicable,
and yet students should be motivated to learning them
because they shape their mentality and improve their
approach to solving practical problems. Because the
recognition of the value of principles may come only later,
even after years of experience, some "faith and trust" is
needed from the students.
o Technology is evolving faster and often it is hard to
distinguish between hype and real new good approaches.
* We should integrate class teaching with projects. This is often
recognized as very critical issue in software engineering
education. Replaying the complexity of real-life projects in an
educational environment can be impossible. Thus we need to
find innovative ways of integrating project work in curricula
(see, for example, [1] and [3]). We argue that projects should
be realistic, but students should be aware ofthe differences
with the real life, in terms of team size, requirements for
compatibility with legacy systems, unavailability of "real"
stakeholders. At the same time, project work should exploit
the opportunities (research methods, prototyping, ...) that
often are unavailable in the industrial world. Students may
turn out to be carriers of innovation when they enter the
business world.
* As a generalization of the previous point, we should try to
make things as easy and understandable as possible without
hiding the intrinsic difficulties. Often this requires addressing
toy problems. Such problems are useful and relevant if well
chosen, but can be useless, or even counterproductive, if they
hide the real problem.
* We should teach how to select/evaluate different methods and
approaches rather than follow them like recipes. Often
methods force a normative behavior, but their application
must be preceded by a careful and scholarly analysis of
competitive approaches, a rigorous evaluation of the trade-offs
and costs and benefits. As opposed to more traditional fields
of engineering, software engineering has only a limited
availability of ready-made normative methods that are
scientifically supported. The ability to perform costlbenefit
analysis, however, is a general skill that is typical and
common to all areas of engineering.

The context of SE can be categorized as follows:

* Mathematical background. This is a controversial issue.
Mathematics is fundamental for every engineer but less
mature in SE. We claim that any engineer, not only a software
engineer, must have a solid background in all fundamental
areas of mathematics:
o traditional continuous mathematics, which constitutes the
general cultural background and educates to rigorous

thinking;

o discrete mathematics (logic, combinatorics, algebra; more

generally, "non-continuous math"), which provide the
essential prerequisites for the theory of computing;
o statistics and probability theory.
Ofcourse, the balance between these can vary among fields.
* The core of computer science: programming, computer
architecture, algorithms and data structures, theoretical
computer science, programming languages, data bases,
operating systems, networking, and all the topics that are
usually taught under "software engineering" (from
programming methodologies to software architectures,
requirements engineering, software specifications and
modeling, testing and analysis, software processes, cost
estimation, project management, ...).
* The essentials of the "physical world", with which software
intensive systems interact, such as
o mechanics, thermodynamics, electricity, etc.
o technology: electronics, telecommunications, ...
o the "social physical world": economics, business
organizations, communication, management, etc.
Note that some principles-such as modularization, abstraction,
modeling, verification, etc.-are quite general, and crosscut the
various engineering fields. They should be used systematically not
only in teaching computer science topics, but also teaching "the
essentials ofthe physical world".
We would like to conclude by saying that it is not only true that SE
has to learn from traditional engineering, but also the converse is
true. Software engineers cannot ignore continuous mathematics and
the basics of physics, but civil engineers, industrial engineers etc.
cannot ignore discrete mathematics and logic, and the basics of
computer science (by this we do not mean how to use the Intemet,
productivity tools or, in general, computer tools, but the principles
of computer science). Cross-fertilization with other engineering
fields is needed, in both directions; no matter how focused the
curricula in the various fields are.

4. REQUIREMENTS ON CURRICULA

Engineering is seldom highly specialized and narrowly focused.

Much more often it has a system aspect; often involves
heterogeneous and interdisciplinary aspects; this is even more true
in the case of SE for its very nature. The ultimate purpose of
software, in fact, is to allow computer-based systems to interact with
their extemal environment, to control it, automate functionality, or
provide service. The extemal environment may be the physical
world of a controlled chemical plant or an intensive-care unit in a
hospital. Or it may be the organizational world of a business unit to
be supported in their sale operations.
Although a software engineer cannot be an expert in every physical
domain, he or she must be able to interact with experts from those
domains. Thus, SE cannot be taught in isolation. IT must be put in
context -how broad is an open issue.

5. REFERENCES

[1] M. Jazayeri, Education of a Software Engineer, Keynote

presentation at ASE 2004, Linz, Austria, 2004.
[2] F. P. Brooks, "No Silver Bullet: Essence and Accidents of
Software Engineering," Computer, Vol. 20, No. 4, 1987).
A.
[3]
Baker, E.O. Navarro, A. van der Hoek, "An Experimental
Card Game for Teaching Software Engineering Processes",
Journal of Systems and Software, to appear.

638