Same meter.
Two worlds.
Can the same falling electricity readings mean two different things? Build a counterexample, then see why a real blackout curve needs careful interpretation.
No account required · Calculator optional · Invented teaching model, not historical data
The same fall. A different story.
Each reading is constant over one hour in this invented example. Before looking below: does the fall tell you how much electricity people wanted but did not receive?
Demand falls to 45 GW.
The assumed delivery ceiling stays at 90 GW. All demand in both hours is served.
Model unserved energy, two hours
0 MWh
The ceiling falls to 45 GW.
Change the assumed second-hour demand. The meter cannot show electricity that is not delivered.
Applied input: 65 GW. The model outputs below have been recalculated.
Hour-two meter
45 GW
Model unserved energy, two hours
20,000 MWh
Reveal the interpretation after recording your own
Same two readings. Different unserved energy.
World A leaves 0 MWh unserved; World B leaves 20,000 MWh unserved under its assumptions. The readings alone do not tell you which world you are in.
A counterexample you can check.
In this simplified model, delivered power = min(demand, delivery ceiling). Unserved energy = (demand − delivered power) × 1 hour. 1 GW = 1,000 MW. All table inputs are invented assumptions.
| World / hour | Demand (MW) | Ceiling (MW) | Delivered (MW) | Unserved (MWh) |
|---|---|---|---|---|
| A / 1 | 60,000 | 90,000 | 60,000 | 0 |
| A / 2 | 45,000 | 90,000 | 45,000 | 0 |
| B / 1 | 60,000 | 90,000 | 60,000 | 0 |
| B / 2 | 65,000 | 45,000 | 45,000 | 20,000 |
This activity does not send your slider setting to GERT or count you as a participant. The JSON stays on your device unless you choose to share it. It records model parameters, not identity, learning gains, or verified participation.
The mathematical idea: non-identifiability.
Different hidden inputs can produce the same observation. The mapping from demand and delivery ceiling to a meter reading is many-to-one, so the reading alone cannot recover both inputs. This example establishes that logical limitation; it does not establish which mechanism caused any real-world change.
Real grids also involve transmission, generation availability, reserves, behavior, and changing conditions. The toy model leaves those out. Its chosen demand values are not estimates of demand during the Texas storm.
Inspect the actual winter-load curve →Host it with one shared screen.
- Minute 1: Show only the two readings. Ask what they do and do not establish.
- Minutes 2–3: Change World B to 70 GW. Have someone compute its two-hour unserved energy. Ask why the meter still reads 45 GW in hour two.
- Minutes 4–5: Try 45 GW. Ask why there is now no unserved energy in either world, then name one extra observation needed to distinguish the two worlds.
Facilitator answer and honest evidence record
At 70 GW, World B has (70,000 − 45,000) MW × 1 hour = 25,000 MWh unserved in hour two, and zero in hour one. World A has zero in both hours. At 45 GW there is no unserved energy in either world.
A demand estimate independent of the delivered-load reading, together with information about supply and interruptions, could help distinguish explanations; those inputs bring their own uncertainty.
After a real activity, record the date, number present, number who attempted the calculation, and one concrete misconception or critique. Keep host confirmation privately. Attendance and attempted calculations overlap; do not add them as different users. This is a teaching aid, not a learning-effectiveness study or certificate.
Found a confusing sentence or a broken assumption?
A specific counterexample or correction is more useful than a rating. Note your chosen demand, what you expected, and what happened. Downloading a model does not send feedback.
The review page links to public GitHub issues, which require a GitHub account. Do not post student names, private class records, or personal contact information. Original GERT material uses the repository MIT license. No institutional endorsement is implied.