Challenge 2: Explain Why Re-Measuring After Step 2 Matters — Possible Solution ==================================================================== WHY RE-RUNNING LIGHTHOUSE AFTER STEP 2 IS A GOOD IDEA ---------------------------------------------------------- Deferring the render-blocking script changes WHEN the browser discovers and starts fetching every OTHER resource on the page — with the script no longer blocking parsing, the HTML parser reaches further down the document sooner, potentially discovering the ad slot's embed code, other images, or additional scripts earlier than it did before. This is exactly the kind of change Chapter 2 warned about with its "the LCP candidate can change mid-load" point: the relative TIMING of when different elements become visible can shift once one bottleneck is removed, even though nothing about those OTHER elements was directly touched in Step 2. Concretely, this could reveal: - A NEW LCP candidate: if some other large element (that previously loaded even later, hidden behind the same render-blocking script) now loads faster than the hero image for some reason, it could briefly or permanently become the new largest-visible-element candidate, something Step 2's fix never anticipated or specifically addressed. - A CLS issue exposed earlier: if content that used to appear late (after the blocking script finally finished) now appears sooner, any layout shift it causes might now happen at a different, more noticeable point in the loading sequence than it did in the original audit. - An INP measurement affected by timing shifts: script execution order and timing changes could affect exactly when certain long tasks occur relative to a user's first likely interaction, subtly changing the INP picture even though Step 4's INP-specific fixes haven't been applied yet. WHY ASSUMING THE REMAINING ISSUES ARE UNAFFECTED WOULD BE A MISTAKE ------------------------------------------------------------------------ Treating LCP, CLS, and INP as three completely independent, sealed-off problems — where fixing one has zero effect on the others — ignores that all three are being measured on the SAME page, competing for the SAME main thread and the SAME resource-loading timeline. A change made specifically to help LCP can shift the underlying conditions the CLS and INP measurements from the ORIGINAL baseline were based on, meaning the plan for Steps 3 and 4 should ideally be informed by a FRESH measurement taken after Step 2's change, not by the original Step 1 baseline alone. WHY THIS WORKS AS AN ANSWER ------------------------------ This directly reflects the capstone's own closing tip box: "fixes can interact" and re-auditing after each round is what catches problems a single, one-pass fix-everything-then-measure-once approach would miss entirely — this challenge is asking specifically about the FIRST instance of that interaction risk, right after the very first fix in the sequence.