Teaching children to read

The program (what a school should actually do)

Reading is the closest thing education has to a solved problem at the level of method — the disagreements that remain are about magnitude and durability, not direction. A founder can commit to the following with high confidence:

Ages 4–7 — build the code.

• Teach grapheme–phoneme correspondences explicitly and systematically from the start. Do not use whole-language, "balanced literacy," or three-cueing (guessing words from pictures/context). This is the single most defensible instructional choice in the building. (evidence: phonics)
• Fold phonemic awareness into the phonics work — always tied to letters, briefly (on the order of 10–15 hours, focused on blending/segmenting). Skip standalone oral "sound games" curricula; speech-only PA doesn't transfer to reading. (evidence: phonemic awareness)

Ages 6–9 — build fluency through volume.

• Have children read connected text aloud with feedback, and read a lot. The active ingredient is reading volume, not any special technique — don't over-invest in rereading the same passage, and don't treat silent independent reading time as an instructional lever (it has no proven effect on achievement). (evidence: reading fluency)

Throughout, K–8 — the comprehension engine is knowledge, not "comprehension skills."

• This is the part most schools get wrong. Once decoding is fluent, comprehension tracks vocabulary and background knowledge, which are built slowly and cumulatively. Adopt a coherent, knowledge-rich, content-heavy curriculum (history, science, geography, arts, literature — deliberately sequenced) and treat that as your reading-comprehension program.
• Do not run generic "comprehension strategy" units as a substitute. Teaching strategies (summarize, predict, self-question) is worth a handful of sessions — it's a one-time trick, not a trainable skill — but it does ~nothing whole-class at scale and cannot replace content. (evidence: comprehension strategies, background knowledge & vocabulary)

For struggling readers — intervene early, with the code.

• Use explicit, code-based 1:1 or small-group tutoring, concentrated in K–2 (intensive code instruction has a closing age window). Prefer cheaper code-aligned tutoring over branded meaning/cueing programs like Reading Recovery, which show strong short-term gains but no credible evidence of durable benefit at high cost. (evidence: Reading Recovery & Tier-2 tutoring)

What to refuse to spend money or time on

• Whole language / balanced literacy / three-cueing — never beat explicit phonics in a trial.
• Standalone phonemic-awareness ("ear training") curricula — don't transfer without letters.
• Sustained silent reading as an instructional intervention — no experimental support (this is not an argument against reading for pleasure, only against counting it as a lever).
• Comprehension-strategy programs as the comprehension plan — a saturating trick, ~null at scale.
• More "systematic" or branded phonics over ordinary explicit phonics — adds ~nothing.
• Reading Recovery at its price point — unproven durability.

The hereditarian bottom line for a founder

Reading is ~70% heritable by school age, so most of the variance in which children read best is genetic — but that is not an argument against teaching, it's the reason teaching matters: literally nobody reads without instruction, and good instruction moves the population mean even as it raises the heritability of the remaining differences (better teaching lets genetic potential express — it's floor-raising, not gap-erasure). Two practical consequences:

1. You can make every child a competent decoder; you cannot equalize them, and you can't raise g. Aim your reading program at getting every student over the decoding threshold efficiently, then pour years into knowledge and vocabulary — the slow lever that actually compounds.
2. Early gains fade by default. A great phonics year shows up strongly at age 5–6 and washes toward zero by age 11 for typically-developing kids (it persists mainly for the at-risk tail). The durable payoff comes from the cumulative knowledge-building curriculum, not from any single intervention — so build for the long game, not for a test-score bump.

Confidence

Open questions a founder should watch

• The durable comprehension lever (knowledge-building) rests substantially on one strong lottery study — theoretically compelling, thinly replicated. Watch for confirmation before betting the whole curriculum on the strongest version of the claim.
• No method reliably prevents the early-gain fadeout; the cumulative-curriculum bet is the best available answer, not a proven one.