meanwhile, my full nodes sit here totally unstressed and under-utilized. 
i thought gmax et al said "large" blocks were going to collapse my nodes?
I've been reading this thread since a long time, and mostly enjoyed the economic insights it used to be about. However, I can only agree with those who see cypherdoc's reputation fading with his supposedly technical comments like the one above. It has been pointed out repeatedly and should be clear as day - currently we have the 1 MB limit you are complaining about. That's precisely why your nodes are "unstressed and under-utilized". From the current stress on your nodes, you can at best guess very vaguely at what they would be doing with larger blocks. I don't see why that's an argument you make in favour of increasing the blocksize. (Same as your comments about "full" blocks that were debunked by others above.)
no, memory is not just used for 1MB blocks. it's also used to store the mempools plus the UTXO set. large block attacks have the potential to collapse a full node by overloading the memory. at least, that's what they've been arguing.