What is needed to help the masses of developers out there work with Linked Data more effectively? They are already learning and using OOP, Object/Relational Mappers, RMI, JNDI and other similar technologies so it seems to me that Linked Data should easily be within reach. What are the barriers to adoption and what can we do to remove them?
My personal opinion is that the more Linked Data is seen as good RESTful web development practices the more uptake it will see. I think uptake is going to be sluggish as long as Linked Data is about having to convert your webapp to use a triple-store and offer up a SPARQL endpoint.
Something like Adenine (maybe not that extreme) could make things easier.
Adenine (Haystack, MIT) was a language that supported native RDF data model, and it could be "compiled to RDF" see Adenine: A Metadata Programming Language
More on Adenine in Karun Bakshi's thesis "Tools for End-User Creation and Customization of Interfaces for Information Management Tasks"
One thing is something Danny Ayers suggested a couple of years ago: Semantic Web in a Box (SWAB). There is a cached writeup here*. In a nuttshell: one-click install package, with software, tools, bundled examples, tutorials, etc.
For instance, someone starting to learn about webdevelopment can install a webstack like XAMPP, load some examples or point-and-click-install webapplications, and (s)he is ready to go. I would like something like that for semantic webdevelopment.
*Danny talks about a unified GUI; I'm not sure about that.
The hard part is often have to convince your IT dept to give you enough access to their web space so you can set up the server to do linked data properly. They can often get nervous about unfamiliar technology.
Personally I think it would be nice to see more linked data apps out there..ones that are really "living the linked data dream" to show people why they should care.