Build-and-test defense
This is a live exam at the bench. The student builds a simple working device — a circuit that lights a bulb, a ramp that rolls a cart, a lever, or a spring setup — then measures how it actually behaves and writes the numbers down. Then the guide starts asking: why did you build it that way, what does this measurement show, why is your data honest even on the run that went wrong. There is no worksheet to copy and no picture to look up: the student stands next to the thing they built and defends the design and the data out loud.
| Criterion | Not yet | Approaching | Mastered |
|---|---|---|---|
| Working device | The device does not work — the circuit never closes, or the cart will not roll. | The device works some of the time but is unreliable or needs constant fixing. | Builds a device that works reliably and can make it run again on demand. |
| Design reasoning | Cannot say why the device is built the way it is. | Explains part of the design but cannot connect the parts to how it behaves. | Explains each design choice and how it shapes the way the device behaves. |
| Measurement & technique | Takes no measurements, or reads the tool wrong and forgets the units. | Measures the device but mixes up units or reads the scale carelessly. | Measures with the right tool and correct units, and repeats a reading to check it. |
| Honest data | Changes or hides numbers that did not match the hoped-for result. | Records the data but smooths over a run that went wrong. | Records every reading as it happened — including the runs that surprised them — and keeps them. |
| Oral defense under questioning | Folds at the first follow-up or repeats a memorized line that does not fit the build. | Answers some follow-ups, falters when asked to justify a choice or a number. | Handles unrehearsed follow-ups about this build with sound, on-the-spot reasoning. |
“I built a series circuit with a battery, a switch, and one bulb. It lights every time because the loop is complete — open the switch and it goes dark, which proves the current needs a full path. I measured the brightness the same way each try, and I wrote down the one run where a loose wire made it flicker instead of pretending it worked.”
“I made the bulb light up. I’m not really sure why it works — I just connected the wires until it came on. I didn’t write down the times it didn’t work.”
This assessment is AI-proof by design: it happens at the bench, with a real device the student built and real measurements they took, in real time. No chatbot can close a circuit, roll a cart down a ramp, or defend a number it did not measure. Mastery is shown by building and defending — not by submitting.