I just finished reading the essay, "How Red Hat Software Stumbled Across a New Economic Model" It is probably the first pro−linux article I've read that mostly seemed to make sense. I suggest you read it.
I believe the major arguement made to adopt linux is, since it is Open Source, you the user have the control to adapt it to your needs (assuming you can program or can find somebody that can and can afford his fees.)
It also makes the point (as do many other papers) that maintanence, support and distrubution are among the few known ways to make a living in the Open Source world.
The problem I see is that making a living is generally the reason most things get done and I don't currently believe all forms of software can support making a living under the Open Source terms. Sure a few people here and there will make an application just because they want to (ie, Gimp, Emacs, C++) but compare those types of applications to the tens of thousands of commercial applications across nearly all platforms.
Putting aside distrubution for the momment, It seems very clear that operating systems will always need support and maintanence. New hardware comes out (USB, IrDA, IEEE 1394?, Zip Drives, Sparq Drive, Digital Cameras etc) and all those products need support and its probably a good bet that support for many of those products is critical. Critical enough that if it doesn't exist, some company will put up the development money to hire someone to make the driver. In fact most hardware companies will probably fund driver support themselves. Similar things are true of new standards like support for HTML or DHTML or XML or Jini(tm) etc. etc..
The NASA and Fermilab examples from the article also make interesting points that Open Source appears to have solved because it is Open Source and allowed those labs to do things they probably could not have done otherwise.
My question is, is that always the case? For example Adobe makes a product called Illustrator. It's a 2D vector based drawing/design program. Would any of the current Open Source income models work for software such as this. Is any one company in enough NEED that they could fund the development of software like this. It's clear to me how Red Hat is able to do this with an operating system while using the GPL license. Read the eassy and it is spelled out pretty clearly. On the otherhand a product like Illustrator of the same quaility might never appear in an Open Source world because it is the monetary incentive of proprietary software and competetion with other proprietary software companies that drives the development of that product. No one company NEEDed that product so bad that they could afford the hundreds of thousands of dollars it cost to make and it's not clear that there are enough voluteers that have the DESIRE to create something as polished.
It is not so clear how the Open Source model promots the creation of end−user applications. By end−user applications I'm talking about product mom and pop would use and or the average office worker, soho home office user would use. I think there are some issues that make this type of software different from Operating Systems and also things like Office Billing Systems. Office Billing Systems are generally specific to a certain office and need maintanence when new procedures are added or the tax laws change. Those people MUST have their software updated and so they pay the maintanence charge.
Application users on the otherhand at somepoint, don't really need any maintanence. I'm sure some of people will disagree. I'll give you an example. Did Adobe Photoshop really need to add effects to their layering system in version 5.0? The answer is not really. No one was demanding that feature. Adobe added that feature to stay ahead of the competition and they do this because they are using the proprietary model where they need a way to get you to buy their product. There is extreme pressure to think of some new way to make a cool feature for Photoshop in that model. Under the Open Source model, this feature would likely not have been added until somebody not only had the idea but also had the talent and the time and desire to sit down and program it. Adobe, because they have to add features in order to keep having a new version to sell and to stay ahead of their competition, just needed money (which they have from selling propritary software) to hire some people who may not have wanted to add the feature but were willing to do it for the money. Compare that to the average user who does not have the 20K or more it would cost to pay a programmer to get such a feature added to an Open Source project.
Some of you will nodoubtedly point to the Gimp as a successful example of such a product but it is very arguable that the Gimp is not up to the level of the current version of Photoshop. It may get there but then the next version of Photoshop will come out with even newer and hopefully more useful features. There is no motivation to do the same for the Gimp except personal statisfaction of having created said featues but unfortunately, if the people responsible for the Gimp lose their interest or find another life (work, lover, religion etc) then the Gimp's progress will likely slow significantly.
Also, the Gimp is just ONE example. Where are all the others? If the Open Source model works so well, why don't we have every type commercial application of the same quality as the proprietary versions. We don't. Will it happen? I'm not so sure.
Some of you are going to claim this is FUD. It's not. What I want to know though is, is there a "natural" way to support the development of end−user applications in the Open Source world. The Red Hat article makes it clear that there is an incentive to making Red Hat brand Linux and even to supporting development of said OS. If we can find a model that works for applications then development will take off.
One last note: Today it was posted on Slashdot that some magazine somewhere put Red Hat Linux on a CD that was distrubuted with 70,000 copies of the magazine. I'd like to know what percentage of Red Hat's income is from distrubution of Red Hat Linux vs income from support. If it is signficant, all anybody has to do to kill Red Hat is convince AOL to put the lastest verison of Red Hat Linux on their ubiquitous AOL CDs and all of a sudden nobody has any reason to pay Red Hat for distrubution. That may not be so far fetched when DVD replaces CD as the normal distrubution method.