Digital Lab · Learning

Orbit School — learning orbital mechanics by getting it wrong first.

A small 3D spaceflight game in the browser. Four missions, each built around a prediction that most people get wrong. This page covers why it's built the way it is.

The teaching problem

Orbital mechanics is a well-known case where everyday intuition is wrong. On the ground, you go up by pushing up and catch someone by speeding up. In orbit, pushing up mostly tilts your path, pushing forward raises the other side of the planet, and to catch a spacecraft ahead of you, you slow down.

Telling people the rule rarely changes those intuitions. Intuitions that come from years of everyday experience tend to survive a correct explanation. What changes them is a moment where your own expectation fails in front of you. So the game is designed to make that moment happen on purpose, safely, and then explain it straight away.

Predict → Observe → Explain

Every mission uses the Predict–Observe–Explain sequence (White & Gunstone), a classic structure from science teaching:

The commitment in step 1 matters most. A vague "I'd probably speed up" is easy to quietly revise afterwards. A clicked answer that's shown back to you ("You predicted 'Fire forward'. The answer was 'Fire backward'.") turns an experience into a contradiction you have to resolve. The wording is on purpose: "That surprise is the lesson, so here's why." A wrong prediction is never marked red and never costs anything.

Feedback before consequences

The hardest thing about real orbits is the delay: a burn here shows its effect half an orbit later, on the far side of the planet. That delay is exactly why intuition never gets corrected. The game removes it. The orange line is the orbit you'd follow if you stopped firing, and it's recalculated every frame from your position and velocity. Hold "Forward" and you see the far side of the line swing outward while your ship barely moves.

• Several views of the same thing. The orbit line, the AP/PE markers (highest and lowest points) and the numbers in the corner all describe the same state. A learner who doesn't read numbers can watch the line; one who doesn't trust the picture can watch the AP number climb.
• Warnings before failure. If your orbit is about to hit the planet, the line turns red before you crash, so the cause and the effect show up together.
• Notes that react to what you do. Hold "Away from planet" during mission 1 and a note appears: "Pushing up or down mostly tilts the orbit sideways. Watch the AP number: it hardly moves." The feedback targets the likely misconception at the moment it happens.

Guided discovery, not a sandbox

Pure "go and explore" discovery has a poor track record for learning concepts (Mayer's 2004 review is the usual reference). Learners often explore without finding the principle. So the open-ended part, free flight, comes last. Before that, every mission:

• has one goal you can see (a dashed ring, a station, an orbit that has to open up);
• tracks progress with one bar and one number, so you can tell whether your last action helped;
• gives a hint after 30 seconds or when the fuel runs out, never before. There's room to struggle a bit first, but not to stay stuck.

Self-explanation prompts

Every debrief ends with a "Think about it" question and a hidden answer, for example: "Your new circular orbit is slower than the one you started on, although you only ever fired forward. How?" Prompting learners to explain things to themselves is one of the best-supported ways to deepen understanding (Chi et al., 1994). Hiding the answer behind a button is a small nudge to think before reading.

A second layer: doing it efficiently

After each mission the game shows how much fuel (Δv) you used next to what an efficient flight needs. Winning a mission proves you know the right direction; matching the efficient figure shows you also know the right place and amount. Mission 2 is the clearest example: firing forward at the low point instead of the high point technically raises your orbit, but it wastes fuel and overshoots the ring. The game says so while you're flying.

Why 3D, and why not more

The physics is two-dimensional: every orbit here lies in one plane. 3D still helps, because the key ideas are spatial. "The other side of the planet" means something when the planet is a sphere you can look around. The tilted camera makes near and far sides of the orbit readable. A globe with clouds and an atmosphere also makes the scale feel real (the readouts use real Earth numbers: 7.1 km/s at 1,600 km).

What the game avoids: a free 3D camera that can lose the orbital plane (the tilt is limited), inclination changes, and anything needing a gamepad. Every one of those would show off more 3D and teach less.

Using it in a class

It works well projected at the front of a room for about 20 minutes, age 12 and up:

1. Show the prediction question for mission 1 and let everyone vote by hand before anyone touches the controls.
2. Let pupils who disagree convince a neighbour, then vote again. This is Mazur's Peer Instruction, which works well with exactly this kind of misconception question.
3. Fly the mission live. Read the debrief together, and leave the "Think about it" question open as a discussion prompt.
4. Repeat for missions 2–4. Mission 3 ("to catch up, slow down") usually starts the best argument.

On laptops or tablets, pupils can play alone. The touch controls are big buttons, and nothing is saved anywhere except which missions are done, kept in that browser.

Under the hood

One HTML file plus a self-hosted copy of three.js r178 (MIT licence). No build step, no external requests, no tracking. Units follow Earth: one length unit is the planet radius (6371 km), GM = 1, so one speed unit is 7.905 km/s. The ship and the station are integrated with symplectic Euler in substeps of 0.004 time units, which keeps orbits stable at 16× time warp. The orange prediction line is not simulated forward. It's the exact Kepler conic from the current position and velocity (eccentricity vector, semi-latus rectum), so it costs nothing to redraw every frame and is always correct. Orbits are drawn as flat ribbons in the orbital plane because WebGL lines are stuck at one pixel wide. The planet's texture is value noise sampled on the sphere itself, so there's no seam.

References

White, R. & Gunstone, R. (1992). Probing Understanding. London: Falmer Press. (Predict–Observe–Explain)
Mayer, R. E. (2004). Should there be a three-strikes rule against pure discovery learning? American Psychologist, 59(1), 14–19.
Chi, M. T. H., de Leeuw, N., Chiu, M.-H. & LaVancher, C. (1994). Eliciting self-explanations improves understanding. Cognitive Science, 18(3), 439–477.
Mazur, E. (1997). Peer Instruction: A User's Manual. Upper Saddle River, NJ: Prentice Hall.

Support this — everything here is free.

R Donate via Revolut