About the unverified marks on this pageFigures tagged unverified are PACE-local estimates that could not be confirmed against any published source, so they are labelled rather than quietly presented as fact. Every other factual claim on this page — statutes, standards codes, gene biology, primary-source citations — was independently checked; the full claim audit, including what came back wrong, lists every claim with its source.
Try it Tuesday · 90 minutes
Run this next week
90 minutes, and every student needs one crater somebody has already measured — Barringer, Tycho, anything with published dimensions they can look up. Minutes 0–10: the task is not to explore the simulator. It is to make the simulator reproduce a crater that already exists. Minutes 10–40: open the embedded crater-sim demo (or generate one with the prompt), feed in the impactor size, speed and angle the literature proposes for that crater, and set the computed diameter and depth beside the surveyed ones. They will not agree. Minutes 40–75: the residual — how large the miss is, which direction it runs, and what the simulation leaves out that would account for it. Worth chasing — whether the scaling law was ever fitted at that size, what atmospheric passage did to the impactor before it arrived, whether the target rock resembles the material the constants came from. Minutes 75–90: each student turns in a one-page calibration note — observed numbers, computed numbers, the size of the disagreement, and their best physical account of where it comes from. A simulation that matches an observation has taught nobody anything. The disagreement is the entire lesson.
The full assignment
Make it 3D print
Printed terrain from real elevation data the student selected: a crater, a rift, a volcanic feature. Print the same feature at two exaggeration factors and let students argue about which one misleads.
Build it AI chat · no code
An interactive explainer or small simulation of the process that formed the feature. Change one parameter, see the outcome change. Built by the student in an afternoon, no coding background required.
Try the demo · runs offline
Planetary Science demo
Interactive demo — “crater-sim”. It cannot be printed; run it in a browser at https://el3vate.vercel.app/demos/crater-sim/index.html
This demo runs entirely offline against canned data — no keys, no network, safe on conference wifi.
4 weeks, realistically
The 4-week version
- Wk 1Each student claims a feature and writes the formation question they want to answer; the feature’s elevation data defines the terrain to print.
- Wk 2Submit the terrain print file, including the second exaggeration factor, at the start of the week. Fabrication runs while students build the model.
- Wk 3Build the interactive simulation. Requirement: one parameter that, when changed, produces a feature that does not exist in nature, and an explanation of why.
- Wk 4Show and tell with the printed terrain and the running model. Peers try to break the model.
Starter prompt & what the model actually did
The prompt, run for real
Build a single self-contained HTML page with an interactive simulation of this planetary process: [describe the process]. Include sliders for the parameters that matter physically. Show the resulting landform update live. Label the axes and give units. Then tell me which parameter ranges produce results that are physically implausible, and why the simulation still lets me set them.
Here's a compact, self-contained page. Save as `.html` and open it.
```html
<!doctype html><meta charset="utf-8"><title>Impact Cratering</title>
<style>body{font:14px system-ui;margin:1rem;max-width:640px}
label{display:block;margin:.4rem 0}canvas{border:1px solid #999;background:#f4ead2}
output{font-weight:bold}</style>
<h2>Impact Cratering Simulator</h2>
<label>Impactor diameter L: <input id=L type=range min=1 max=2000 value=100>
<output id=Lo></output> m</label>
<label>Impact velocity v: <input id=v type=range min=1 max=72 value=20>
<output id=vo></output> km/s</label>
<label>Impact angle θ (from horizontal): <input id=a type=range min=5 max=90 value=45>
<output id=ao></output>°</label>
<label>Surface gravity g: <input id=g type=range min=1 max=25 value=9.8 step=.1>
<output id=go></output> m/s²</label>
<p>Final crater diameter D ≈ <output id=Do></output> m depth ≈ <output id=de></output> m</p>
<canvas id=c width=600 height=220></canvas>
<script>
const $=id=>document.getElementById(id), ctx=c.getContext('2d');
function calc(){
const L=+L_.value, v=v_.value*1000, th=a_.value*Math.PI/180, g=+g_.value;
const rho_i=3000, rho_t=2500; // projectile & target density
// Gravity-regime pi-scaling (Holsapple), vertical velocity component:
const vz=v*Math.sin(th);
const pi2=g*L/(vz*vz);
const Dt=1.16*Math.pow(rho_i/rho_t,1/3)*Math.pow(L,0.78)*Math.pow(vz,0.44)*Math.pow(g,-0.22);
const D=1.25*Dt, depth=0.2*D; // final rim-to-rim + depth
Do.value=D.toFixed(0); de.value=depth.toFixed(0);
draw(D,depth); Lo.value=L; vo.value=v_.value; ao.value=a_.value; go.value=g;
}
function draw(D,depth){
ctx.clearRect(0,0,600,220); const cx=300, sy=140, s=Math.min(1,500/D);
const w=D*s, d=depth*s;
ctx.fillStyle='#c9a15a'; ctx.fillRect(0,sy,600,80); // ground
ctx.fillStyle='#f4ead2'; ctx.beginPath(); // crater bowl
ctx.moveTo(cx-w/2,sy); ctx.quadraticCurveTo(cx,sy+2*d,cx+w/2,sy);
ctx.closePath(); ctx.fill(); ctx.strokeStyle='#7a5b1e'; ctx.stroke();
ctx.fillStyle='#333'; ctx.fillText('← '+D.toFixed(0)+' m →',cx-30,sy-6);
ctx.fillText('x = horizontal distance (m), y = depth (m)',10,210);
}
for(const el of[L_,v_,a_,g_]) el.addEventListener('input',calc);
const L_=$('L'),v_=$('v'),a_=$('a'),g_=$('g'); calc();
</script>
```
(Rename the `id`s if your browser complains about global name clashes; the code above assigns them explicitly.)
**Physically implausible ranges — and why the sliders still allow them:**
- **Velocity below ~11 km/s**: slower than typical planetary escape/encounter speeds. Real impactors arrive at 11–72 km/s (Earth). Low values are physically possible only for grazing captures, so results there understate cratering.
- **Very shallow angles (<~10°)**: the vertical-velocity term collapses, giving tiny craters. Real low-angle impacts instead produce *elongated* craters and downrange ejecta — a geometry this bowl model doesn't capture, so the number is wrong even if finite.
- **Large diameter + high velocity + low gravity together**: pushes energy into the *strength* or *complex-crater* regime (central peaks, terracing). This uses only **gravity-regime pi-scaling**, so it silently extrapolates past its validity.
- **g near 1 or 25 m/s²**: fine for planets, but combined with tiny L you enter the strength regime where cohesion, not gravity, sets crater size — ignored here.
The sliders allow these because the tool computes a *continuous scaling formula* with no regime gate: every input yields a finite number regardless of whether the underlying physics still applies. That's a deliberate teaching feature — you can *see* the model break, which is more instructive than clamping. A production model would switch scaling laws (strength vs. gravity, simple vs. complex) at the relevant thresholds.
- Self-listed implausible regimes (velocity below 11 km/s, angle below 10°, strength regime)The model named exactly the parameter ranges where its own result is physically wrong — the model-limits reasoning the assignment grades, produced unprompted.
- Named its scaling law: 'gravity-regime pi-scaling (Holsapple)' — the attribution checks out, the constants were not checkedPi-group scaling with separate gravity and strength regimes is a real approach and is correctly attributed. We closed one number and could not close the rest: the gravity exponent −0.22 matches the β = 0.22 commonly adopted for dense rock (Schmidt & Housen 1987), but the leading constant 1.16 and the exponents 0.78 and 0.44 were not confirmed against the primary literature, and the assumed densities rho_i = 3000 and rho_t = 2500 kg/m³ are stated with no source and tied to no particular body. A correct citation is not a correct formula, and a student should pull the paper before trusting any number this code returns.
- No regime gate: 'every input yields a finite number regardless of whether the physics still applies'The model explicitly admits the tool computes past its validity — the exact failure the exercise exists to expose.
- Low-angle geometry error acknowledgedIt notes that real low-angle impacts make elongated craters its bowl model cannot show, so the number is 'wrong even if finite' — a precise statement of the physical limit.
Assessment
Rubric
| Criterion | Weight | What it assesses |
|---|---|---|
| Physical soundness of the question | 15% | The formation question is well-posed and answerable with the model's variables. |
| Finding the breakdown | 35% | Identifies a parameter regime where the model's result is physically implausible and demonstrates it. |
| Explaining the limit | 30% | Explains what physics the model omits or extrapolates past, not just that the number looks wrong. |
| Communicating with units | 20% | Axes, units, and scales are labeled so the result is interpretable. |
What this replaces
Swapping it into a real course
Replaces
The problem set of cratering or orbital-mechanics calculations.
What is lost
The hand-computation practice with the governing equations.
What is gained
Students reason about where a scaling model breaks and why — the judgment a working scientist needs — instead of grinding through plug-and-chug problems, and they get a tangible landform to argue over.
Where AI is bad at this
The failure your students should catch
A model builds a physics simulation that returns a confident number for every input, including regimes where its scaling law no longer applies — extrapolating a gravity-regime crater formula into the strength regime, ignoring that low-angle impacts make elongated craters, and never gating physically impossible parameters. The clean output invites the student to trust results the underlying physics does not support.
Budget & logistics
What it costs to run
- Instructor prep2 hours
- Class time90 minutes
- Per-student cost$0 for the Tuesday version; roughly $6–14 per student in filament unverified for terrain prints at two exaggeration factors in the four-week build.
- Fabrication file dueFirst day of Week 2 — submit the terrain print file (both exaggeration factors) at the start of the week the feature is claimed. PACE quotes 7–10 business days (up to 14 calendar days), so it clears before the week-4 show-and-tell.
- Calendar dependencySimulation work needs no lead time. The terrain comes from the elevation data of the feature claimed in week 1, so its print file goes in at the start of week 2 and the up-to-14-calendar-day turnaround clears before the week-4 show-and-tell.
Three sizes
Scale it to the time you have
One session
The 90-minute calibration session producing a one-page note on why the simulator misses a surveyed crater; no fabrication.
4 weeks
The seeded four-week plan: claim a feature and question, submit print files at two exaggerations, build the simulation with a documented implausible regime, then a show-and-tell where peers try to break each model.
One semester
A full project — a self-selected feature from real elevation data, printed terrain at two exaggerations, an interactive process simulation with a documented implausible regime, and a peer session trying to break each model.
Reserved · live build
This space is intentionally empty. During the Day 8 session it will be filled in live — fill the liveBuild field in content/planetary-science.json and rebuild.
Tell us what happened
Run it, then say how it went
If you try this — the 90-minute version, the prompt, any part of it — send back what you tried and what happened, especially anywhere the model was confidently wrong. That is the material the next session is built from.
Email feedback on Planetary Science →
Opens a draft in your mail client, already addressed and titled. No form, no account, nothing to sign up for.