Page 23 - FCW, May 30, 2016

M
Fre
A
quent
BY STEVE KE
I
NAGIN
CONTRACT
nok and Joiwind
perform
outcomes without burying
indefinite-quantity (IDIQ) contract
without specifying req
the beginning. Sp
upfront, of course, whole idea of agile.
very
The government
general description
but be specific
will use to develop ments during agile s
view
results in
is
ing agile sprints.
Recently, I had a
L
n the preceding pages, Dan
for agile contracting — most notably, the fear that some principles
of
e
agile
cannot
about t
MAN
some management challenges
be reconciled with
xisting procurement regulations.
As I’ve noted on FCW.com and elsewhere, however, good practice
suggests — and pro
tions allow — issuing a solicitation
for an agile contract or task order
under an umbrella indefinite-delivery,
curement
wo
that less specificity upfront
greater
ance
Ronen discuss
uirements at
ecifying every detail
uld violate the
should give only a
he
of
the work
process it
and refine require- prints. My own
attention and rigor
in the post-award evaluation of deliv- erables that contractors produce dur-
conversation with
Che-
regula-
G
tor
A
Mark Schwartz,
the
ty’s
alike.
GILE
AWARD
evaluations can dramatic
an
agency in FAR
Department
U.S.
the dynamic CIO
of
Services (USCIS), to explore such trade-offs in more detail. He had great
insights on
two im
cal issues: post-award manageme
agile contracts and task orders, and past-performance evaluations in an
agile environme
In my
read carefully by everyone in th
nt.
government
Too much of a good thing?
Schwartz strongly
agile contracting enables better
post-award monitoring
tor
performance for two reason
One is that the government gets the results of a contractor’s efforts very
frequently,
of performance
portant
view, his advice should be
federal community who is working
on agile
—
AFT
Homeland Securi-
Citizenship and Immigration
bu
and practi-
and contrac-
believes that
of
not only after long periods
against
requirement. The second is that agile development often entails the use of
automated
tools that
tinct parts of the work on
contrac-
a complex
signal when
e
s.
a sprint are
at
nt
dis-
of
ER
ally improv
rea
ucracy
done and
part
require
of
TH
e
that to the governm
orders of what constitutes “done,” and they usually specify that the soft-
ware has
been
have been uncovered.
I asked w
tors couldn’t use
mated tools
require
hy
ments and sign
the work,
sidered done until it has
And
limited
in
full requirement has been c many moons later.
In
waterfall
tors submit reports
amount of the requirement been completed, but those
subjective and
enough.
tested.
transparently communicate other
information of the proce
E
transparently communicate
There are definitions in
en
if not the
ment, was done. Schwartz
said no piece of software — even in a waterfall environment — can be con-
waterfall contracting, only
t
esting
ca
n be do
t.
The t
about various aspects ss, such as what defects
agencies and contrac-
the same auto-
for traditional waterfall
agile task
ools
al when at least
development, contrac-
of
the e
not nearly transparent
Agile sprints are either done
May
30, 20
whole
been
ne
also
tested.
before
ompleted,
stimated
that has reports
16
FCW.COM
the
are
19