AI-NATIVE SERIES · PLAY-ENGINEERING, EPISODE 2 OF 4

Play-Engineering Runs on Fascination, Not Scores: Rewards Shrink the Fuel, a Future Self Grows It

By Paul Jialiang Wu · agentic-portfolio-lovat.vercel.app · 2026-09-25 · 13 min read

How to read the numbers: g and d measure how big a difference is — about 0.2 is small, 0.5 medium, 0.8 large. "SD" (standard deviation) is the same kind of unit.

The one question this episode answersWhat keeps a 12-year-old playing at something hard for years?

This piece does not answer how play turns into skill — that is episode 1, play finds the mechanism — or how to build what you are fascinated by, which is episode 3, or how to point it at a restored world, which is episode 4.

What this episode of play-engineering covers, and who it's for

Episode 1 said play-engineering is simple: play finds a mechanism, engineering names it and tests it on a second case. But none of that happens unless you keep coming back — for months, sometimes years. This episode is about the fuel. It's for students who have lost interest in something they used to love, for parents deciding what to reward, and for anyone who leads a team and wonders why incentives keep backfiring.

The sparring room from episode 1 carries on here, so let me extend it.

The picture: forms for the judges, or the fighter you want to become

In a martial arts club there are two ways to show up. You can practise forms so the judges give you a belt. Or you can train because you've seen the fighter you want to become — for me, as a college kid, it was Bruce Lee in films — and every session is a step toward him. Both people are in the same room doing the same drills. Only one of them stays after class to spar.

My rocket, and my crash

When I was twelve and thirteen, in middle school in China, I was near the very bottom of my class — failing almost everything. My mother, who is wise, said I was playing too much and not listening, and if I worked harder I'd catch up. So I did what everyone around me did: I forced myself. I still remember the dim classroom and the rows of heads bent over homework. By the end of the semester I'd climbed to about 16th–20th out of 50. Decent. And I hated it.

Then my father bought my older sister a book about the junior college class — kids of twelve to fifteen who had taught themselves and gone straight to a famous university of science and technology. My sister didn't read it. I did. And something switched on. If they can do it, I thought, why not me?

I found every textbook I could — maths, physics, chemistry — and taught myself. In less than one semester I went from the bottom of the class to one of the top three students in a grade of about 360. By the end of middle school I had taught myself all of high-school maths, physics and chemistry. Nobody was paying me. Nobody had to make me.

And then the crash, which is the most important part. At fifteen I jumped ahead to sit the national university exam early, and missed the junior-college entry. Soon I was too old for that door at all. The picture that had pulled me — the self I was becoming — was gone. I was still top of my class in three subjects, but by my final year I felt, looking back, a little depressed. I went to a good university, not the one I had dreamed of. (If a lost dream ever leaves you feeling low for weeks, tell an adult you trust. It helps more than you'd think.)

Twice in the same body, the same brain, the same school: once pushed by effort, once pulled by a future self. The pull was far faster — until the thing pulling me disappeared.

What the research says about fuel

Prizes can shrink the thing they're meant to grow. In one of the best-known studies, preschoolers who already liked drawing with markers were promised a "good player" award for drawing — and later chose to draw less than children who got no award or an unexpected one (Lepper, Greene & Nisbett, 1973, described). A meta-analysis of 128 studies found the pattern holds: rewards you expect for doing, finishing or performing a task reduced free-choice motivation, while positive feedback that tells you something true about your performance increased it (Deci et al., 1999). The argument isn't settled forever — some researchers found the effect weaker — but the design lesson is safe: don't promise prizes for an activity you want someone to love; give them information.

A future self works — if it comes with a plan. In a school intervention, 141 low-income eighth graders built vivid "possible selves", linked each one to concrete strategies, and practised treating difficulty as normal. Their grades, test scores and initiative improved compared with 123 classmates, and the effect was still there two years later (Oyserman, Bybee & Terry, 2006). The authors are clear that a dream on its own is not enough; what worked was the dream plus the plan.

Read my story through that lens and it stops being a mystery. The junior-college book gave me a possible self. The textbooks gave me a strategy. The rocket was the two together. The crash was the possible self being taken away — and no strategy can run without something to steer toward.

The second fuel: maths as a story

After college I wanted to study abroad, and my English was bad — I failed my first two tries at the graduate exam. I went back and lived for a while with a college friend who was a brilliant mathematician. Over beers, he told me something that changed how I read everything. In my memory of it, he said that reading mathematics was like reading a storybook: the theorems are characters, and he was having a conversation with them about the universe.

I started reading maths that way — as a conversation with someone wise — and the subject I had drifted away from became beautiful again. It's worth being honest about the evidence here: teaching maths as storytelling is a well-developed idea among educators, but I found no controlled study measuring its effect. So it goes in this episode as a lived mechanism, not a proven one — and as a mission option, so you can measure it yourself.

Five whys: why did a failing twelve-year-old reach the top of his grade in one semester?

1. Why did my rank jump? Because I studied for hours that nobody required. Evidence: my own account; one person, told as a story.
2. Why did I study without being made to? Because I had a vivid picture of who I could become — the junior-college kids. Evidence: possible selves predict effort when they feel real and reachable (Oyserman et al., 2006).
3. Why didn't the earlier "work harder" push do the same? Because effort for a score is the belt-for-showing-up kind of motivation, and expected rewards shrink free choice. Evidence: d = −0.40 across 128 studies (Deci et al., 1999).
4. Why did the picture turn into progress instead of daydreaming? Because it came with a strategy — every textbook, in order. Evidence: the intervention that paired selves with strategies improved grades for 141 students.
5. Why did it collapse later? Because the future self became impossible, and a strategy with nothing to steer toward stops. The root: the fuel is a possible self plus a strategy; scores are the exhaust — and rewarding the exhaust drains the fuel. Evidence: effects mediated by change in possible selves, still present at two years (evidence.yml row oyserman_treatment).

Operate — the thing that runs

In one sentence: every number on this page is checked by a robot before the page goes live. In more detail: the mission is the operating unit: a card, a daily number, a weekly piece of feedback. On this site, every research figure on the page is pulled from content/series/play-engineering/evidence.yml and re-checked by scripts/check-play-engineering.mjs on every build; the reader results are counted by scripts/play-missions.mjs from the comments.

Feedback — the right number, and its blind spot

Free-choice minutes are the measurement the motivation research itself uses: not what you say you feel, but what you do when nobody's watching. If week 4 is higher than week 1, the fuel grew. The blind spot: minutes can rise for reasons that have nothing to do with the card — a holiday, a new friend in the same hobby, a rainy month. One person's numbers are a story; many people's numbers, in the comments, start to separate the card from the weather.

Feedforward — decided before the results

• If most readers' minutes fall, the card is too vague — the next version adds the ingredient most often skipped: the "when it gets hard" sentence, practised out loud.
• If the maths-as-story twist raises minutes for most who try it, it becomes its own mission in a follow-up, because that effect is unmeasured in the research and readers would be measuring it first.
• If an adult reports that prizes helped, that's welcome data — some studies found the effect weak — and it goes into this page's next update, not into a footnote.

Where this episode sits in the series

Same story, five exits

Build it in 30 minutes

• Input: one skill and one person who is trying to learn it.
• Output: a future-self card and a 30-day line of free-choice minutes.
• Invariant: no promised, tangible reward for doing the activity.
• Stop rule: if minutes fall three weeks running, rewrite the card with the person — the picture isn't theirs yet.
• Minimal record: date, free_choice_minutes, feedback_given.
• Deliberately missing: a comparison week with a promised prize. Don't run it on a child you love; the research already has.

5S self-score, with evidence

References

1. Deci, Koestner & Ryan (1999), Psychological Bulletin 125(6), 627–668 — rewards and intrinsic motivation (meta-analysis).
2. Oyserman, Bybee & Terry (2006), JPSP 91(1), 188–204 — possible selves and school outcomes.
3. Lepper, Greene & Nisbett (1973), Journal of Personality and Social Psychology 28(1) — described, not quoted.
4. Zazkis & Liljedahl (2009), Teaching Mathematics as Storytelling — a practice book; no controlled outcome data found.

Related

• Episode 1 — What is play-engineering, and what can play do that drills can't?
• Episode 3 — How does a kid turn a sci-fi artifact into something they can build and test in 30 days?
• Episode 4 — How does a team of kids and elders turn a vision of restoration into a plan they can run and measure?