Exercise 7.12

"More data is better" applies to a fixed model $(H,A)$. However when we are doing early stopping, we are selecting models on a nested hypothesis sets $H_1\subset H_1\subset \dots$ determined by $D_{\mathit{train}}$, that’s because at each step, the $w_1$ is selected by choosing the one with minimal in-sample error. If we use the full data $D$, the $w_1,w_2,\dots$ will be different and as a result, the hypothesis sets will change even if we keep the step size $\eta$ the same.

That’s why the ’more data is better’ doesn’t apply here.