I agree in the principle of open source moving up the stack, but it seems the majestic transition from infrastructure to application skips an important space that I think is more on the open source playing field: middleware. I will concede that this has always been an ambiguous term, but if you consider it to be the enabling software that provides the necessary abstractions for thousands of flowers to bloom, I think you can at least understand why it is distinguished from the same realm as an email client.

I perceive there is a general business culture with inertia that still expresses a need to see individual applications as line items on a budget summary with pressure-easing support contracts lined up behind them. They do care about detailed evolving requirements, but they have proprietary vendors lined up to eagerly take those particular orders. What they don't care about is what's under the hood making the functions function. Enter open source.

Consider JBoss. What problem does it solve? It may be better to ask: what problem doesn't it solve? With the emergence/resurgence of platform independence, I think middleware will be where the battle is won. For example, the real grist won't be Exchange vs. Evolution/Scalix/Thunderbird/etc. so much as it will be something like J2EE vs. .Net, AD vs. OpenLDAP, or Sharepoint vs. the inevitable OS alternative(s). Win the middle with a closed solution, and the application space will get strong-armed in winner-take-all fashion as we've seen before.

If, on the other hand, the open source movement takes the middle, then it's winner be all. Considering middleware is the heart of innovation, who do you think has the odds at taking the flag? Sorry Redmond, my money is with the renegades.