I think that the "publish or perish" mentality holds people back from trying to get their kernel code accepted. It's only important for most researchers to publish - not to do anything with it.
Some people found it really strange when I wanted to take the DCCP code and help put it into the kernel. Thankfully I persevered.