When the iterations on a given project or architecture start hitting diminishing returns, trying new things from scratch is often a good strategy. I assume the

When the iterations on a given project or architecture start hitting diminishing returns, trying new things from scratch is often a good strategy. I assume the commenter meant the quote for this context, but failed to make it explicit.