There are no hard figures. If you take say 10 software development companies, how many window and linux boxes they had say begining last year and how many do they have now? This will give us a fair idea of whether or not, Linux is acutally gaining ground in the Windows segment.

I believe that all the Linux and for that matter Mac people in the world spend way too much time on surveys and whining. If Linux is a better system it will eventually gain an undeniable market share. I am not going to go out and buy a copy of linux because I want to keep up with the penguins. Linux is a very flexible system that allows geeks(like me and you ) to manipulate our environment. But never ,never ,never tell me that it is more secure than anyother system out there. look around the internet for security holes and security flaws in OS's you will find that there are as many in linux as there are in windows. Mac on the other hand does not have as many they actually have the least. But once again there are not as many people using Mac's as there are people using Windows OS's. and therefore not as many viruses being forged for that OS If you want to get something out there are you going to use Windows,Mac,Linux to deliver your message. Duh everybody know the answer to that question I believe that all the other OS's are riding the wrong train tell why it is better not that it is more secure. Because if someone takes over the spot of Windows as the next giant or if they parallel Windows that system will have to put up with hundreds of viruses and infections and scrutiny just like windows.

Just look at server market share numbers. Both Linux and Windows are rapidly gaining share, while UNIX is losing. It seems pretty clear that Linux is eating into UNIX share, not Windows.

Jereon and Michel both make interesting observations. Here's another data point. Microsoft's overwhelmingly most successful market is on desktops. The greatest penetration of Linux, outside of developers, is in the server realm. I work for one of the big companies jumping on Linux somewhat. But, the target marketplace for Linux from my company as well as from IBM is large blade based clusters and so forth. The only people I've seen with Linux (or less often FreeBSD) on their desktops are a very small minority of developers, all of whom used to work on our SVR4 based kernel...inherited from Tandem days. Microsoft has more than an OS stranglehold on the desktop. Word, Excel, Powerpoint etc are the standard tools for my company. Personally, I don't like Microsoft Office, but strangely, the younger software folks seem to love it. As always, one opinion, YMMV.

There is more wrong with the claim.
In my experience most people changing to coding for Linux in the past coded for commercial Unix distributions and not for Windows.
Therefore the "loss" of coders to the Windows community is even lower than Michel makes it.

The 50% stated was probably derived at using a non-representative group of people.
If for example they queried only hobbyist (quite possible) the only people who would have used another Unix in the past are the ones switching from for example FreeBSD to Linux.

Also not taken into accound is that most people who start coding for Linux will not stop coding for other OSs.

IF Linux were indeed to gain marketshare so quickly, then sales of Windows would be down severely (simply because computer sales as a whole are down, add the supposed loss of sales due to decreasing marketshare...).
In fact sales of Windows seem stable despite decreasing computer sales. While this is in part due to replacement of existing installations with new versions and increased anti-piracy efforts, there must be more to it.

I´ve yet to see a larger company using Linux exclusively instead of Windows (instead of maybe on a few servers, and then often in place of Solaris or AIX, NOT in place of Linux).
I my experience Linux is still for the most part an OS used by hobbyists and for small servers (in places where larger commercial Unix boxes would be overkill).

The mathematics are wrong, since they do not respect the rules of inference.
To wit: assume there are 1000 WIN developers and 100 UNIX
developers. Then along comes LINUX and 10 people start developing...
if (article's numbers) 50% used to code for WIN (5 people), then WIN lost 1 in 200 of its coders. If 30% used to code for UNIX, 3 dev's, then UNIX lost 1 in 33 of its coders, or an impact felt six times as strongly.
So the conclusion 'WIN lost in absolute number more of its coders so WIN bleeds more than UNIX' violates logic.
I have no idea of the real amounts (or fulltime jobs per year) of developers for the platforms, but you essentially need those data to draw a conclusion. If not, it's data torturing (yep, they confess eventually).