software engineering : An Idea Whose Time Has Come and Gone?

Sujets

• Informatique
• Marketing et communication

Informations

• Publié par: kyann
• Publié le: 11 janvier 2013
• Nombre de lectures: 64
• Langue: English

Extrait

viewpoints
Contact Editor: Dennis Taylor ■ [e-mail]
software engineering:
An Idea Whose Time Has Come and Gone?
t om DeMarco
e’re now just past the 40th anniver- Compelled to Control
sary of the NATO Conference on The book’s most quoted line is its frst sentence:
Software Engineering in Garmisch, “You can’t control what you can’t measure.” This
Germany, where the discipline of sof-t line contains a real truth, but I’ve become increas-
ware engineering was frst proposed. ingly uncomfortable with my use of it. Implicit in W Because some of my early work be- the quote (and indeed in the book’s title) is that
came part of that new discipline, this seems like control is an important aspect, maybe the most im-
an appropriate moment for reassessment. portant, of any software project. But it isn’t. Many
My early metrics book, Con- projects have proceeded without much control but
trolling Software Projects: Man- managed to produce wonderful products such as
agement, Measurement, and GoogleEarth or Wikipedia.
Estimation (Prentice Hall/Your- To understand control’s real role, you need to
don Press, 1982), played a role distinguish between two drastically different kinds
in the way many budding soft- of projects:
ware engineers quantifed work
and planned their projects. In ■ Project A will eventually cost about a mil-
my refective mood, I’m wonder- lion dollars and produce value of around $1.1
ing, was its advice correct at the million.
time, is it still relevant, and do I ■ Project B will eventually cost about a million
still believe that metrics are a must for any suc- dollars and produce value of more than $50
cessful software development effort? My answers million.
are no, no, and no.
The book for me is a curious combination of What’s immediately apparent is that control is re-
generally true things written on every page but ally important for Project A but almost not at all
We welcome combined into an overall message that’s wrong. important for Project B. This leads us to the odd
your letters. It’s as though the book’s young author had never conclusion that strict control is something that
Send them to met a metric he didn’t like. The book’s deep mes- matters a lot on relatively useless projects and
software@ sage seems to be, metrics are good, more would much less on useful projects. It suggests that the
computer.org. be better, and most would be best. Today we all more you focus on control, the more likely you’re
Include your full understand that software metrics cost money and working on a project that’s striving to deliver
name, title, time and must be used with careful moderation. something of relatively minor value.
affliation, and In addition, software development is inherently To my mind, the question that’s much more im-
email address. different from a natural science such as physics, portant than how to control a software project is,
Letters are edited and its metrics are accordingly much less precise why on earth are we doing so many projects that
for clarity in capturing the things they set out to describe. deliver such marginal value?
and space. They must be taken with a grain of salt, rather
than trusted without reservation. Continued on p. 95
96 IEEE S E Published by the IEEE Computer Society 0740-7459/09/$25.00 © 2009 IEEE
otwrafViewpoints
Continued from p. 96 and documentation and acceptance test- tion standards. All these strive for consis-
ing incrementally as you go.” tency of practice and predictability.
Can I really be saying that it’s OK to This might sound like an agile- Consistency and predictability are still
run projects without control or with rela- methods prescription, but I’m too far desirable, but they haven’t ever been the
tively little control? Almost. I’m suggest- away today from the actual building of most important things. For the past 40
ing that frst we need to select projects software to recommend at the methods years, for example, we’ve tortured our-
where precise control won’t matter so level. Rather, I’m advocating a manage- selves over our inability to fnish a soft-
much. Then we need to reduce our ex- ment approach, one that might well steer ware project on time and on budget. But
pectations for exactly how much we’re the team toward agile methods, at least as I hinted earlier, this never should have
going to be able to control them, no mat- toward the incremental aspects of the been the supreme goal. The more impor-
ter how assiduously we apply ourselves to agile school. tant goal is transformation, creating sof-t
control. ware that changes the world or that trans-
o far, I’ve mostly discussed software forms a company or how it does business.
An Unsettling Analogy engineering’s metric component. How We’ve been rather successful at transfor -
Imagine you’re trying to control a teen- S about the rest? I’m gradually coming mation, often while operating outside our
ager’s upbringing. The very idea of con- to the conclusion that software engineer- control envelope. Software development is
trolling your child ought to make you ing is an idea whose time has come and and always will be somewhat experimen-
at least a little bit queasy. Yet the stakes gone. I still believe it makes excellent sense tal. The actual software construction isn’t
for control couldn’t be higher. If you to engineer software. But that isn’t exactly necessarily experimental, but its concep-
fail in your task, fail utterly, lives can be what software engineering has come to tion is. And this is where our focus ought
ruined. So, it’s absolutely essential that mean. The term encompasses a specifc set to be. It’s where our focus always ought
you not lose your grip entirely. You’re like of disciplines including defned process, in- to have been.
a fencer who’s learning to hold his sword spections and walkthroughs, requirements
as though it were a bird: too tight and the engineering, traceability matrices, metrics,
bird will be injured; too loose and it will precise quality control, rigorous planning Tom DeMarco is a principal of the Atlantic Systems
fy away. and tracking, and coding and documenta- Guild. Contact him at [e-mail].
Now apply “You can’t control what
you can’t measure” to the teenager. Most
things that really matter—honor, dignity,
IEEE Software (ISSN 0740-7459) is published bimonthly by the IEEE Computer So-discipline, personality, grace under pres-
ciety. IEEE headquarters: Three Park Ave., 17th Floor, New York, NY 10016-5997. sure, values, ethics, resourcefulness, loy-
IEEE Computer Society Publications Offce: 10662 Los Vaqueros Cir., PO Box 3014,
alty, humor, kindness—aren’t measurable. Los Alamitos, CA 90720-1314; [phone]; fax [phone]. IEEE Com-
You must steer your child as best you can puter Society headquarters: 2001 L St., Ste. 700, Washington, DC 20036. Subscription
without much metric feedback. It’s hard, rates: IEEE Computer Society members get the lowest rate of US$51 per year, which
includes printed issues plus online access to all issues published since 1988. Go to www.but then parenting is hard. You get a little
computer.org/subscribe to order and for more information on other subscription prices.
bit of measurement in the form of school
Back issues: $20 for members, $163 for nonmembers (plus shipping and handling).
grades, and you’re grateful for it. But you
Postmaster: Send undelivered copies and address changes to IEEE Software, Member-
also know that your child’s math grade is ship Processing Dept., IEEE Service Center, 445 Hoes Lane, Piscataway, NJ 08854-4141.
a better indicator of achievement than his Periodicals Postage Paid at New York, NY, and at additional mailing offces. Canadian
Spanish grade, because math understand- GST #125634188. Canada Post Publications Mail Agreement Number 40013885. Re-
turn undeliverable Canadian addresses to PO Box 122, Niagara Falls, ON L2E 6S8, ing is easier to measure. And his “grade”
Canada. Printed in the USA.
in comportment is much more likely to
Reuse Rights and Reprint Permissions: Educational or personal use of this material is tell you something about the teacher than
permitted without fee, provided such use: 1) is not made for proft; 2) includes this no-
about the child.
tice and a full citation to the original work on the frst page of the copy; and 3) does not
So, how do you manage a project with- imply IEEE endorsement of any third-party products or services. Authors and their com-
out controlling it? Well, you manage the panies are permitted to post their IEEE-copyrighted material on their own Web servers
without permission, provided that the IEEE copyright notice and a full citation to the people and control the time and money.
original work appear on the frst screen of the posted copy.
You say to your team leads, for example,
Permission to reprint/republish this material for commercial, advertising, or promo-“I have a fnish date in mind, and I’m not
tional purposes or for creating new collective works for resale or redistribution must
even going to share it with you. When I