I can tell you why one of them fell out.
Firstly, I did not feel that there is enough consideration and effort going into optimization required to realistically operate nodes in 'lower echelon' environments. The bulk of the effort I detected was directed toward denying that there is a potential issue looming.
Secondly it's a bitch to compile bitcoind on anything but the one 'true' platform. I've got to futz with this every time I wish to upgrade. (I've not checked to see if anyone has developed a functional auto-configuration system yet...but I doubt it.)
Thirdly, and most importantly, I've got other interests and responsibilities.
Next time I run a node it will probably be on a dedicated micro instance in AWS or some similar platform. That probably won't be until I have the time and inclination to actually do something positive toward support of the project rather than just whine.