The algorithm begins by reading for structure and noting hedges, then marking the specific line numbers that make the claim. Write a short neutral paraphrase in the margin before interrogating each choice to find what words must be true. Apply distractor labels like TS (too strong) or OS (off-scope), compare tone and scope, and when two options remain, prefer the minimal claim the text clearly supports. Finally, lock your answer by pointing to line numbers before bubbling.

Clean elimination increases expected value even when unsure, because removing two distractors converts a 1-in-4 guess to a 1-in-2 guess. If you allot 13 minutes for a 10-question passage (1.3 minutes per question), force a decision among remaining contenders once a question exceeds double that time, then mark it for revisit. This preserves time for questions you can fully prove while maintaining overall accuracy.

For Main Idea questions, reject choices that only summarize a detail or add unstated motives. For Detail questions, accept only what the referenced lines literally say and eliminate any answer that shifts time, agent, or degree. For Inference questions, pick only what is highly likely given the lines and eliminate options requiring new assumptions. For Words in Context, reread surrounding sentences and eliminate dictionary definitions that don't fit the author's usage. Paired Evidence questions reward the choice that best proves your previous answer.

Effective practice involves timed drills where you label each wrong choice's flaw (TS, OS, CON, PAR), copy exact words from the passage that conflict with it, and rewrite the correct answer in simpler language matching the lines. Distractor-only drills work well: read only answer choices without the passage, predict what flawed logic each uses, then read the lines to confirm. Evidence pairs can also test your ability to immediately link an answer to the best supporting lines.