What do you mean by removing as many CMS features? Why? End users use it as a cms. People create sites with DNN because they want to concentrate their efforts on adding content and not deal with layout, html and web design stuff. Many end users do not know html and a system like DNN makes sense to them.
I can't speak for Frank here, but the way I read his comments was on keeping the DNN 'core' as small as possible, and enabling wide variety in using whatever CMS features the end-user wants. It's about not baking a particular way of showing content into the core code. Now, I think the problem here is the use of the world 'Core'. To me, core means the set of code in the DNN library - anyhing that is not a Http Module, provider, or DNN module. This is actually a pretty small subset of code. To other people, 'Core' probably means anything downloadable from dotnetnuke.com, or anything that a 'core' Team Member wrote.
I don't think Frank was saying 'throw out the CMS parts'. I think he was saying 'keep it open and allow for pluggable CMS parts'. DNN took a step along these lines a while back by not installing every module available on a base install. This lowered the default installation size, complexity and time and reduced the amount of unncessary features that most people have in their site. Even Microsoft is doing this with Vista now - to get what used to be standard features in Xp and it's forefathers, you have to go and manually choose them to be in the system. This lowers the resource usage, allows for implementation of other, third party features, lowers the security exposure and in general makes everyone happy. The same philosophy should and is being applied to DNN, in my opinion.
So bring on the new UI widgets, the CMS features - the whole lot. But let the core team focus on a fast, secure and extensible web application platform, and let the myriad of third party developers, open souce contributors and anyone else add to the total feature set of DNN as a CMS.