Describir: Can We Read Neural Networks? Epistemic Implications of Two Historical Computer Science Papers.