Raw transcript — How to Speed Read
Source: https://www.youtube.com/watch?v=ZwEquW_Yij0 Duration: 9m36s Captured: 2026-04-19
Full clean transcript at /tmp/yt-backfill/speed-read.txt (1,681 words). Per copyright policy, raw transcript stored externally for internal reference. Re-download via:
yt-dlp --write-auto-sub --sub-lang en --skip-download --sub-format vtt -o "/tmp/yt-backfill/%(id)s" "https://www.youtube.com/watch?v=ZwEquW_Yij0"
python3 ~/.claude/scripts/vtt-to-text.py /tmp/yt-backfill/ZwEquW_Yij0.en.vtt
Structural outline:
- [00:00–01:00] Premise: most “speed reading” is pseudoscience, but a few mechanical tricks do work without sacrificing comprehension. Demo books: standardized formatting, mostly text
- [01:00–02:00] Step 1 — measure. Count words per line on a representative page, multiply by lines per page = words per page. Read one minute at normal speed for baseline WPM
- [02:00–03:00] Peripheral vision argument. Most readers track word-by-word edge-to-edge; they waste the visual margin. Indent one word from each side as a hard boundary; eye still picks up margin words via peripheral vision
- [03:00–04:00] Indent further (two words) once comfortable. Goal: track only the middle third of the page, dropping straight down. This alone can roughly double WPM with no comprehension loss
- [04:00–05:00] Saccadic eye movement — eyes don’t smooth-track, they jump fixation to fixation. Reduce regressions (back-skips) and reduce fixation count per line = fewer milliseconds per page
- [05:00–06:00] Use a pacer (finger, pen). Trace under each line, force two fixations per line, prevent regression. Solves the “I keep re-reading the same line” tired-eye loop
- [06:00–07:00] Re-measure WPM after 10 minutes of pacer + indent practice. Most readers see a 50–200% lift on first session
- [07:00–09:36] Caveats: not for everything (poetry, dense technical, contracts). Use selectively. Comprehension claim is empirical — re-test yourself, don’t trust the technique blindly
Quotes ≤15 words each per copyright policy.