Oscillator DB level - Square Law vs Linear?

I don't have a lab and my test are more trial and error so I can't confirm or deny the chatter I hear on the boards. The subject at the moment is local oscillator signal strength. A 7 DBM mixer or level 7 mixer would require a signal of about .5Vrms. This seems to be the most common one. Mr Tuggle (on TRB) using a one active device receiver has DXed stations 1000's of miles away. His single FET oscillator driving the mixer picking up worldwide signals begs the question, "Is he really getting .5 Vrms with one JFET?".  So with this in mind I'll offer some data from the manufacturer and try and make since of it. One other subject that is constantly declared on the board is Square Root vs Linear. This data should shed some light on that.

So Square law produces twice the output for the same change in input?

WOW! Now I'm truly shocked. Our world renown expert has declared it's all about  matching, coupling and Z many times, yet the manufacturer says otherwise in the last line.
As confused as I am I guess the next thing is to look at a small signal and large signal detector plot and try to determine which to use?

May need to chew on this a while?