[calmcoconut]

Summary

Audit and fix of Chapter 9 multiple choice question distractors. Seven questions had one or more wrong-answer choices that were implausible, invented, or nonsensical — making the correct answer too easy to eliminate to by process of elimination rather than genuine knowledge.

Questions Changed

Q#	Topic	Old Weak Distractor	New Distractor
Q4	Unbounded delays	"OS injects deliberate delays to optimize CPU cache" (nonsensical)	"OS throttles per-process NIC throughput" (real concept, plausible trap)
Q5	Packet vs. circuit switching	"RDBMSs are incompatible with circuit-switched networks" (non-sequitur)	"Circuit-switching adds per-connection setup overhead" (true but misdirecting)
Q8	Clock drift	"Routers strip NTP flags from TCP headers" (implausible)	"NTP daemon bug accumulates offset across sync cycles" (plausible, confusable with drift)
Q10	Logical clocks	Option B was a vague, obviously-wrong placeholder	Replaced with: "logical clock uses consensus rounds to assign unique timestamps" (sounds authoritative, subtly wrong)
Q17	BFT in datacenters	"BFT requires GPS atomic clocks" (confusingly echoes Spanner TrueTime)	"BFT requires independent implementations per node" (real argument from literature)
Q19	Crash-stop model	"Transient amnesia model" (invented term)	"Byzantine fail-arbitrary model" (real term, meaningfully different)
Q21	Eventual consistency	Options C & D were instant eliminations ("Byzantine property", "deterministic property")	Option C replaced with linearizability definition (classic confusion trap)

What was NOT changed

• Q1, Q2, Q7, Q11, Q13, Q14, Q15, Q16, Q20, Q22–Q27 — distractors were already technically grounded and plausible.
• All correct indices, explanation text, modelAnswer text, flashcards, and all other non-MC data are untouched.
• No behavioral or UI code was modified.

Testing

Manual review of all 29 questions against the chapters_fallback/chapter_9.txt reference material. No structural changes to the question objects (same keys, same format).