This is actually a brilliant answer.
Ballmer knows that not only is there no direct proof of intellectual
property being willfully and intentionally copied from Microsoft's
proprietary source archives into Linux, but Microsoft has legally
copied open source software under licenses such as BSD, and
incorporated this software into their kernel.
In fact, Microsoft lawyers have full access to the complete source code
of Linux and Windows and if there are attempts to contribute
proprietary source code from Windows into Linux, that the Linux
community is more that willing to remove the contested code. There
have been examples of this throughout the development of Linux, and in
many cases, when there are multiple candidates for the same function,
the "least risk" solution is usually used. If you take code created as
a freshmen class programming exercise and deploy that into Linux, it's
almost impossible to argue that this technology couldn't have been
intuitively derived by someone familiar with the state of the art. In
many cases, there can be a class of 20-30 students in a class, and
20-30 classes throughout the country in a given year.
By giving these classes a public and unrestricted specification, such
as specifications from X/Open, Posix, or FreeBSD Manual pages, teachers
across the country can get 200-300 different implementations of the
same "device", each different and created.
There have been a few contributions which were challenged, and they
were almost immediately replaced by completely different
implementations designed to solve the same problem.
On the other hand, there are plenty of amateaur lawyers who don't know
the history of Unix or Linux, who see the similar, or even identical
code, who assume that, because they are the same and THEY saw the
Windows code first, that the Linux code MUST have been stolen from
The irony is that Microsoft has successfully defended itself against a
number of patent lawsuits by citing examples of GNU and other Open
Source code, along with related documentation - often 20 years old,
that proved that Microsoft did NOT steal from the plaintiff - in fact,
they stole it from GNU.
The trick here is that because GNU did not patent their technology
(software patents were nearly impossible to get until about 1995),
Microsoft could study the source code, figure out HOW it worked, then
use that description to have a junior programmer (someone with 2-3 year
of experience in general programming) implement the same thing for
Microsoft based on the description, not the source code.
Then, when some amulance chaser of a lawyer tries to sue Microsoft on
behalf of some kitchen table software developer, hoping for a $billion
settlement, Microsoft trots out the REAL history, and the charges are
I would guess that there isn't a line of code in any Microsoft product
that doesn't have a well documented pedigree - for just this reason.
Linux is not significantly different. Every developer is required to
verify that, to the best of their knowledge, the code being contributed
is not proprietary, patented, copyrighted, or otherwise protected by
some company. Furthermore, because each contribution can be traced
back to it's original developer, it's possible to press criminal
charges against a contributor who wilfully and intentionally
contributes proprietary copyrighted or patented code and lies about it.
These archives are very public if not voluminous. Fortunately, most
of Linux was developed when archive technology was relatively cheap -
QIC-20 tapes were about $20 and the drives were less than $200, then
CD-Burners dropped to under $200 with media as cheap as $1/gigabyte.
Since the linux mailing lists and CVS repositories are mostly
plain-text, and easily searched and parsed, it's pretty easy to trace
any real violations almost immediately.
And most companies DO watch this development pretty carefully. There
is always the possibility that a criminal contribution could be made,
of code which is unpatentable, but protected only by nondisclosure
clauses. The failure to act on such a contribution could mean the loss
of all intellectual property rights to the contributed product - since
the implementation could later be intuitively derived.
If Ballmer openly said "Yes, Linux violates our intellectual property"
he knows that there would be demands for specifics. And the Linux team
would happily remove the contested code, and help prosecute the
Keep in mind, it's much easier to prove that Linux was intuitively
derived or at least that Microsoft failed to do due diligence in
checking for these violations - given that nearly everything in Linux
is published in source code form almost the minute it has been QA
Since Microsoft does not publish their source code, there is no way for
them to reasonably expect that Linux would check to confirm that the
code was taken from Microsoft.
It's very similar to trademark and patent registrations. When a
trademark is first registered, it can be challenged for a certain
period. If, after that period, there have been no direct challenges or
communications from someone else using that trademark as a trademark -
then the applicant is granted permanent and exclusive use of thaht
trademark. Ironically, Microsoft's use of Windows as a trademark was
almost immediately challenged. Legally, only the combination of the
words Microsoft and Windows together is trademarked. Still, Microsoft
loves to try to assert their exclusive ownership of Windows alone as a
Finally, yes, Microsoft does have intellectual property. They have
dancing paper clips, dancing puppy dogs, and thousands of little
context sensitive helps available to escort the user through most
situations - this, more than anything helped Windows to be accepted by
nearly 1 billion users across the globe. It even helped to make
"English" (actually the internationalized version used in most large
cities within the United States where millions of immigrants from
around the world have enhanced the vocabulary) as a second language for
many of these 1 billion people.
The irony is that the concept of context sensitive help itself dates
back to Emacs. Context sensitive help has been widely used ever since
then. Microsoft just made context sensitive help a requirement for
nearly every major object used in any Microsoft product.