Exercise 3: Why This Chapter Deliberately Withholds the XOR Answer — Possible Solution ==================================================================== WHAT THIS CHAPTER ACTUALLY SAYS AT ITS OWN CLOSE ------------------------------ Per this chapter, "a single neuron... can only ever draw a straight decision boundary through its inputs, a flat line or hyperplane. No amount of retuning its weights changes that basic geometric limit. nn1-2 shows a real, concrete problem a single neuron structurally cannot solve... nn1-3 shows exactly why stacking a hidden layer underneath fixes it. This chapter deliberately doesn't resolve that mystery yet — it's worth sitting with the question first." WHY NAMING THE LIMIT WITHOUT SOLVING IT IS A DELIBERATE CHOICE ------------------------------ The chapter could have simply stated both the problem and its resolution in one place, but instead splits them across three separate chapters (naming the limit here, showing the concrete failure in nn1-2, and only then resolving it in nn1-3). This structure gives the reader a real, specific claim to evaluate on their own first — "a single neuron can only draw a straight boundary" — before being shown a worked example (nn1-2's own XOR problem) proving that claim concretely, and only after that, the actual fix (nn1-3's own hidden layer). Presenting the limit and the fix together would remove the reader's own opportunity to genuinely grapple with why a straight-line-only model would fail on some real problem, before being handed the answer. WHY "SITTING WITH THE QUESTION" HAS REAL PEDAGOGICAL VALUE ------------------------------ Per this chapter's own framing, understanding WHY depth is necessary (rather than simply being told that it is) requires first genuinely confronting a case where a single neuron demonstrably fails no matter how it's trained — a claim that's far more convincing and memorable once seen concretely (nn1-2's own worked XOR failure) than if merely asserted in the abstract, here in Ch.1, without evidence. Delaying the concrete demonstration by one chapter, and the fix by two, mirrors how the underlying historical problem was actually encountered and eventually solved — not resolved in the same breath it was first noticed. WHY THIS MATCHES THIS COURSE'S OWN BROADER STRUCTURE ------------------------------ This chapter's own roadmap table explicitly separates "nn1-2 / nn1-3 — The real historical limit a single neuron hit, and how depth fixed it" into two distinct roadmap entries, rather than a single combined one — consistent with treating the problem's discovery and its fix as two genuinely separate, worth-separating steps in the course's own overall narrative, not merely a stylistic delay. WHY THIS WORKS AS AN ANSWER ------------------------------ It quotes the chapter's own explicit deferral, explains why splitting the limit, the concrete failure, and the fix across three chapters gives the reader a genuine opportunity to engage with the problem before being handed its resolution, and ties this to the chapter's own roadmap table treating discovery and fix as separate, deliberately sequenced steps.