Was anything about cromwellian's statement inaccurate, such that it should have been downvoted? I'm pretty sure I understand everything being discussed here but I don't understand THAT.
Similarly, citing the 850 ISP of the experimental NERVA engines is equally out of context.
Let me qualify the SSME claim as "highest Isp/Thrust of any widely used engine". Prototypes or engines that haven't flown much IMHO aren't able to make this claim.
The NK-33 may beat a Merlin 1D, but what much do they cost to make and what is the reliability? If you make 100 NK-33s and 100 Merlin 1Ds, what's your total budget, and how many NK-33s will fail compared to the Merlin 1D?
One thing I get with SpaceX, is that their engines are not designed to push the performance envelope. They want them to be cheap and reliable.