Thanks to Stefan for the forwarded eamil.
Von: "SD Magazine"<sd@newsletters.sdmediagroup.com>
Betreff: SD People & Projects: Mo' Developers, Mo' Problems?
Datum: Mon, 03 Oct 2005 19:09:26 -0400 (EDT)
SD PEOPLE & PROJECTS
October 2005: Bigger Teams Not Always Better
By Amit Asaravala
>>>> MO' DEVELOPERS, MO' PROBLEMS?
Thinking about assigning more developers to a project
to accelerate your schedule? Be careful. Putting a large
team on the job could cause you more trouble than it's
worth, according to a new study by software estimation
and analytics vendor QSM.
The study, based on data that QSM collected from 564
information systems projects completed since 2002,
revealed that large teams don't complete projects much
faster than small teams, though they cost much more. In
particular, teams with 34 people on average completed a
100,000-line project in 5.6 months at a cost of $2.1
million, while teams of four people on average took
about two weeks longer but cost just $294,000. Thus,
shaving two weeks off the schedule cost some companies
as much as $1.84 million.
Why such disproportionate production rates? Blame it
on the bugs. The larger teams produced more than five
times as many bugs as the smaller teams, which required
the teams to reexamine their code more often, according
to QSM. In the end, this ate into a large portion of
the time saved by having more developers turn out more
code per day.
But before you decide to cut your team back to just
four people, consider this: The size of the small team
in the study was just an average, and QSM readily admits
that it's saving the question of "optimum" team size for
a future study.
Rather, the real lessons here are that you'd better be
sure that accelerating your schedule by adding more
developers is worth the extra cost, and that you should
have realistic expectations about how many days you'll
actually save by doing so.
--Amit Asaravala