The device has to work

Not “has to teach you something.” Not “has to push past what last year’s group already proved.” Work. Succeed. Clear the demo. Come home with a grade.

I was sitting in a medical device design meeting — scope, milestones, the usual — when both instructors landed on the same rule in different words. Build something that succeeds. A device that fails, even if the failure is interesting, is the wrong target.

I nodded. Then the rest of the day was that sentence on loop.

Because the sentence was not really about one module. It was Singapore’s education operating system, said politely in a project meeting.

This is not one bad rubric

Once “the device has to work” is the scoreboard, the rational move is fixed.

You do not take the risky sensor path.

You do not try the mechanism nobody in the batch has shipped.

You do not spend three weeks on an idea that might die in week four and leave you with a story instead of a checkbox.

You pick the small thing. The known thing. The thing that already works in someone else’s PDF. You stay inside the fence marked possible, and you treat the fence like a moral boundary.

That is not a quirky course preference. That is how a whole system trains people.

From primary school upward, the machine is excellent at one job: produce clean scores under high stakes. PSLE, streaming, subject combinations, GPA, scholarships, “good” modules, clean demos. The public story is merit and excellence. The lived game is risk management.

So of course university project courses sound like this. The instructors are not inventing a new culture. They are speaking fluent Singapore: do not fail in public, do not waste the attempt, convert effort into a mark that survives the transcript.

Do that for sixteen years and you do not get a generation of explorers. You get people who are world-class at not embarrassing the system.

Passing became the national product

I am not saying standards are the enemy. Singapore got rich partly because we are serious, disciplined, and allergic to sloppiness.

But somewhere along the way, passing replaced building as the main output of school.

The mark is the product.

The demo is the product.

The safe A is the product.

Being wrong in public on the way to something new is not a product. It is a liability. Parents feel it. Students feel it. Scholarship boards feel it. So project teams feel it too.

Look at a bioengineering batch long enough and the equilibrium is obvious without a survey:

Small. Safe. Predictable. Easy to defend in the Q&A.

Not: aim high, miss, and leave the lab smarter than the people who only ever built the safe version.

That is not laziness. That is a society reading its own incentives correctly.

A dead prototype can still answer real questions. Which joint shears first. Which assumption was cosplay. Which sensor lies when the patient moves. Which step in the pipeline is the actual bottleneck.

A polished device that only re-runs last year’s known-good path can answer almost nothing — and still win.

“It worked” and “it was worth doing” are different axes. Our education system is almost entirely built on the first one. Then we act surprised when graduates optimize for clean execution instead of new machines.

We hide the work, then wonder why nothing compounds

The other Singaporean habit shows up in student projects too: secrecy.

Previous groups treat their devices like contraband. CAD goes dark. Code disappears. BOMs vanish. Failure notes never existed. The next cohort starts half-blind and spends half the semester rediscovering solved problems, then calls it originality.

That is not how engineering compounds. Engineering compounds when prior art is visible — steal the solid parts, throw out the garbage, push one step past the last person.

Software figured this out in public. Open source makes copying easier — and attribution easier, double work harder, and the field faster, because five teams are not silently reinventing the same mediocre wheel under NDA cosplay.

Closed student builds do the opposite. Same brief. Same constraints. Same wheel. No trail.

If the national story is “innovation hub,” this is a tell. We are still better at protecting grades and face than at building a shared stack other people can climb.

Secrecy is not professionalism here. It is grade insurance with national branding.

You cannot slogan your way into inventors

Singapore loves the language of innovation. Deep tech. Medtech. Future economy. Talent. The next thing the world adopts.

Fine. But you cannot get that culture out of an education system whose daily religion is:

Safe over exploratory
Working over ambitious
Private over shared
Score over substance

That stack produces competent executors of known patterns. Necessary. Valuable. Not the same species as people who ship things other countries copy.

A society that punishes visible failure will not magically produce exploratory builders at twenty-two because one pitch deck said “innovation.” The pitch deck loses to the report book every time.

Open work is not charity. It is how a small country stops paying for the same lesson twice. Credit still exists when the record is public. It just stops depending on everyone else staying in the dark.

Change the scoreboard, not the posters

Keep the requirement that you can finish something. Then change what the system actually pays for:

how far you moved past prior work — including prior batches and prior schools
what you learned when it broke, with evidence, not vibes
what you left open: docs, designs, data, dead ends
whether the problem mattered beyond “this will demo cleanly on Friday”

Make last year’s artifacts the default starting point, not a secret. Grade the delta, not the reenactment. Teach students that a clean fail with a public postmortem is worth more than a private, tiny success.

Until then, students will keep doing the rational Singaporean thing: ship the small device that passes, hide the notes, protect the GPA, and call it education.

I am not mad at the meeting. I am mad that the meeting sounded like the whole country compressed into one sentence.

The device has to work.

Of course it does. That is how we were raised.