Reading programs for low-income populations often give disappointing results. Failures may be partly due to a neglect of practice in decoding letters. Visual stimuli are best learned symbol by symbol, with pattern analogies and much practice to unite smaller components and speed up identification. The prerequisite for comprehending volumes of text is parallel-processing of letters. This happens when the letter-by-letter decoding function moves to the visual word form area of the brain, which recognizes words as if they were faces. Strangely, people may become fluent readers in transparent orthographies without knowing the relevant language. To teach the poor, governments and donors should promote instruction of individual letters and independent student practice with teacher feedback; then, students may achieve parallel processing. Teachers would emphasize comprehension and writing after the attainment of fluency. This methodology requires a rather lengthy textbook for every student, but it simplifies classroom activities and teacher training. To attain the Sustainable Development Goals of 2030, teachers must base instruction on specific scientific studies rather than "best practices".