Mail Archives: djgpp/2006/12/08/00:00:52
CBFalconer wrote:
> Samuel Thibault wrote:
>> Rod Pemberton, le Thu 07 Dec 2006 05:59:48 -0500, a écrit :
>>
>>> You're questions about Vista on a few NGs brings up an interesting
>>> point that I've been wanting to discuss. Do you think that long
>>> term Linux can compete against Microsoft given the small amount of
>>> capital available to Linux versus MS (7 Billion on Vista alone...)
>>> and the continuing rapid change or advancement in PC hardware?
>>
>> What does money have to do with software development??
>
> The most efficient size of development team is 2, or possibly 3
> [1]. Once the team size exceeds 5 efficiency, accuracy, and
> productivity usually goes down. Now consider how many alleged
> programmers are/have been working on Vista.
The first problem with Vista was that too much stuff was planned to go into
it, very likely there appeared many interdependences because the changes
went into almost every part of the OS. That was way too much. And apparently
the processes didn't work well or just couldn't fix the problem quickly
enough with that much of work to do and crowd to manage. Yes, hundreds and
thousands people have been involved in making Vista. I think it's always bad
if you can never rely upon things you're dependent on if they keep changing.
You have to find and fight problems, find out workarounds to make further
progress. Of course, this scales poorly. And there's a very high possibility
of missing things and cases since there're too many of them and I wouldn't
expect people in such circumstances (under load and pressure, working extra
hours, you name it) to take time and review what they have done and how it
works with other components made by others. It will be very good if Vista
turns out to be more secure from the start. But as for the other things, the
picky users have to wait for service packs and other fixes and
workarounds...
And as for the "alleged programmers", there're many people hired just from
the college, there're summer interns, there're people whose background/major
wasn't related to programming and engineering, there're lots of foreigners,
and as everywhere, especially, where there're lots of people, there's a
certain percentage of bad guys who are there because they have to make up
the statistics. So, it should be no wonder to see some bad stuff here or
there because people are too different, work and interact differently and
managing them is a whole different story. And I think this applies not only
to MS and Vista.
> [1] The reason being that extra members can catch the inevitable
> goofs, even if only by being a bear [2] to which the problem can be
> explained.
Unfortunately, big teams aren't for spotting bugs, they're for making 'em.
> [2] Single unit development teams can benefit from a large teddy
> bear to which the problem can be explained. They rarely interrupt,
> never criticize nor categorize actions as idiocies, and never never
> squeal on you [3].
It's one thing if you're directly interacting with someone on whose
work/code you depend or are on the same branch in the hierarchy and it's a
different thing if you're in different teams with own goals and so forth and
the common management point is far up the tree. You have to escalate your
issues with someone else's ugly code or architecture in order for them to
reach the target and make a difference for you. And some out there will be
pushing back. It really seems like when there're too many owners, there's no
owner, nobody cares for the common success as much as they do for their own
component they're directly responsible for.
Alex
- Raw text -