> Essentially all we've done is pushed the basic model proposed by linear regression to it's absolutely limits [emphasis mine]

Where's the data to show that we've hit the limits?

It's obviously not linear regression. You mean gradient descent?