Update README.md

README.md

@@ -14,10 +14,11 @@ capacity that the two experts can enable simultaneously with the current build.
 
 So in a sense, the experts will say what the settings are most likely going to be most optimized at, by making a quick soup.
 
-This should provide the necessary yields that I require, assuming I pick experts with relationally similar math.
+This should provide the necessary yields that I require, assuming I pick experts with relationally similar math. So... parameter narrowing soup for now, 
+eventually the system should be able to directly self-attenuate the parameters for the best suggested parameters at the get-go.
 
-So... parameter narrowing soup for now, eventually the system should be able to directly self-attenuate the parameters for the best
-suggested parameters at the get-go. 
+The models themselves will never be trained by the experts, only the params selected by what is most likely. The models will never see an expert opinion directly,
+nor will they be given gradients from anything expert-related. Everything in a vacuum.
 
 ## Flows, Routes, Patterns, Trajectories, Magnitudes, Etc
 Everything mathematically will have a represented flow attenuation mechanism specifically aligned to the curation of that math.