After reading Paul Krill's article on InfoWorld about Eclipse IDE At A Crossroads, I have to disagree with some of the main points made. Paring down the Eclipse projects is a totally different thing to dropping features in the IDE. The JDT is one of the most successful and popular parts of Eclipse, as is very apparent from the demographics from EclipseCon 2008.
Of course, some projects that exist right now in the Eclipse ecosystem will change - whether they disappear or merge with others - this is the nature of evolving software. To suggest that the JDT will have to be reduced would be wrong. As discussed in the article, there's people happy with Eclipse and how it is architected. I believe these to be the majority of users.
Eclipse is based on plugins, so if something isn't required it isn't loaded. Developers have the freedom to add and remove bundles that they require to do their job. There would be no reason to pare down the JDT when this flexibility exists.
Another point raised was the learning curve of Eclipse. While I'll admit that this exists when developing with some of the APIs, as an IDE Eclipse provides first class user assistance. When you start up first you are greeted with a welcome page and links to help documentation and tutorials.
What do you think? Does the JDT need to be pared down, or is it giving you what you need?