You need to sign in or sign up before continuing.
Reflection — An Honest Take 8 min

Honest Take — Before You Begin

Honest Take — Module 1: How Humans Actually Learn #

The honest opening: you probably already know 40-60% of what is in this module. Make It Stick has been a popular book for a decade; "spaced repetition" and "retrieval practice" get dropped casually on any half-decent dev podcast. If that is you, read fast, skip what you know, and do not manufacture difficulty where there is none. The module is not asking you to become a cognitive scientist. It is asking for one shift: from "did I cover everything?" to "what will the reader still know in six weeks?" That single move — comprehensiveness to retention — is the whole module's payoff. Everything else is detail in service of it.

But here is what I predict for the engineers who haven't loaded this evidence base, which in my experience is still most engineers who write teaching content: Make It Stick will land like discovering an observability tool you have been ignoring for your whole career. You will read the testing-as-learning chapter and think, in roughly these words, "I have been writing tutorials for years without any of this." The sentence is correct, and you are not unusual — almost every engineer-author writes without the science loaded. The discovery is the easy part. The trap is what you do next.

Two earlier drafts of this curriculum disagreed about the trap, and the disagreement is worth having in front of you. One said: do not retrofit your past artifacts at all; the blog post that taught somebody once is not improved by being retroactively gilded with science, and the redesign impulse is mostly theater — apply the principles to your next artifact. The other said: the over-correction is retrofitting everything at once, and the discipline is to retrofit exactly one surface, measure it on a real learner, and only then expand, because a half-retrofitted portfolio teaches you to distrust the method. Both drafts were aiming at the same failure: engineering enthusiasm meeting new evidence and producing a sprawling, unfinished migration.

The checkpoint splits the difference correctly — audit one artifact now, write the re-engineering plan, and do the actual rewriting in M4 and M6 where it belongs. Diagnose first. Operate later.

The part the formal module says quietly and I will say loudly: the evidence base is solid in a way the course-creator industry largely ignores, and the reason it gets ignored is not that the evidence is wrong. It is that retrieval practice, spacing, and interleaving produce lower satisfaction scores in the short term, because they feel harder to learners than re-reading and watching. The methods that feel best retain worst. If you ever ship a course, this single fact will put you in direct conflict with your own feedback survey, and the discipline is to refuse the satisfaction signal in favor of the retention signal. Hold the Hattie effect sizes loosely — the methodology is contested in its details — but the direction of the evidence is not in serious dispute.

And keep one humility in your pocket: the science also says learning is integration with prior knowledge, which means your learners — who do not share your prior knowledge — need different teaching than the teaching that would work on you. If you finish this module thinking "great, I know how learning works," you missed it. You know how learning works for minds you can model. M2 is about the minds you can't.

Conclusion #

Take the one shift — comprehensiveness to retention — and let the rest be background that hardens into instinct. Audit one real artifact against the evidence base; write the plan; resist both the bulk-retrofit reflex and the gilding-the-archive reflex. The science is settled enough to act on and ignored enough that acting on it puts you in a small minority of engineer-teachers. That minority is where the compounding lives.

Predictions #

  • If you haven't read the cognitive-science canon: at least one moment of "I cannot believe I've been writing teaching content without this." Absorb it without spiraling into guilt; the guilt is not actionable and the plan is.
  • If you have read it: you will finish faster than the estimate, and the only durable artifact will be the evidence audit. That is enough. Do not pad.
  • You will want to re-engineer three or more surfaces in week one. The checkpoint asks for a diagnosis and a plan. The discipline is stopping at the plan.
  • "The illusion of fluency" will be the concept you quote to someone within thirty days — it is the stickiest idea in the module and the one most engineers have felt but never named.
  • Within three months you will catch yourself, mid-sentence, defending re-reading or highlighting as a study strategy to someone, and revise on the spot. That is the module working.
  • The Willingham line "memory is the residue of thought" will quietly restructure how you design examples — you will start asking what the reader has to think about to follow them, which is the question that was missing.