There's an old book called How to Lie With Statistics of which I'm very fond. Here we have a good example of an anti-nuclear argument that hinges on flimsy statistical assumptions--in this case, a silly definition of "delivered energy" which actually makes it appear that only the electrical generation field has to abide by the laws of thermodynamics.
In fairness the EIA perpetuates these silly myths by accounting for use of fossil fuels for transportation and heating as "delivered energy," even though the use of the primary energy in these applications is far from 100% efficient. This leads to numbers that give the impression that the electrical sector is massively wasteful, which is not at all the case. Look at Table A2 in the linked EIA document: the only "losses" accounted for are in the electrical field: 27.88 out of 101.9 quads of primary energy used in the US.
Grist promises that "Architecture 2030 will post a better answer on Grist next week." I certainly hope so, since so far all they've demonstrated is an ignorance of thermodynamics.