The experiment desk
Make chance
a little more tangible.
Run a local number experiment. See how a digit becomes a size and a set of colors, then compare the sample with the model.
Run a number experiment.
Local educational simulation. Outcomes are generated in your browser using the stated model. They are not live results or predictions of an external game.
Interactive controls require JavaScript and the browser random-value API. The model explanation and classification table below remain available.
Ready. No outcomes generated.
Nothing runs automatically.
Replay restores the last completed dataset without drawing again. Reset clears it. Reloading clears all in-memory outcomes.
Observed frequencies
No draws yetAll counts and percentages use the complete current run, not just the recent rows below. Expected digit frequency: 10%.
| Digit | Count | Observed | Distribution |
|---|---|---|---|
| 0 | — | — | |
| 1 | — | — | |
| 2 | — | — | |
| 3 | — | — | |
| 4 | — | — | |
| 5 | — | — | |
| 6 | — | — | |
| 7 | — | — | |
| 8 | — | — | |
| 9 | — | — |
| Group | Theoretical | Count | Observed |
|---|---|---|---|
| Small | 50% | — | — |
| Big | 50% | — | — |
| Red | 50% | — | — |
| Green | 50% | — | — |
| Violet | 20% | — | — |
Color memberships overlap: 0 is Red + Violet, 5 is Green + Violet. Color percentages can sum to more than 100%.
Longest observed Big or Small run: —. A description of this sample, not a forecast.
Recent local outcomes
Latest first. Shows at most 20 outcomes; statistics above use the entire run.
- No local outcomes yet.
How outcomes are generated
The sampler uses your browser’s random-value API. It accepts unsigned 32-bit values below 4,294,967,290 and maps their remainder after division by ten to a digit. Rejecting the final six possible values avoids modulo bias under a uniform word source.
This implements a model of independent draws. A browser generator is not proof of mathematical independence or certified external fairness. No operator is contacted and no prediction is made.
Run, pause, and inspect
Batch samples contain 1–10,000 draws and run in bounded chunks so you can cancel. Timed experiments use five-second teaching rounds, with 1–100 rounds per run. The timer changes the pace, not the model. Timed progression pauses while the page is hidden and does not invent a background catch-up history.
Replay restores the last completed recorded dataset without drawing again. Export downloads that completed dataset as CSV with the model, mode, sample size, digits, and classifications. Cancelling does not record the partial run. If a previous completed dataset exists, replay/export still refer to that last completed run. Reset clears both the visible run and the recorded dataset; reload clears local run history.
Reading the classifications
The table is the selected example ruleset. A digit has one size, but 0 and 5 have two color memberships. Other versions may use different rules.
| Number | Size | Color membership |
|---|---|---|
| 0 | Small | RedViolet |
| 1 | Small | Green |
| 2 | Small | Red |
| 3 | Small | Green |
| 4 | Small | Red |
| 5 | Big | GreenViolet |
| 6 | Big | Red |
| 7 | Big | Green |
| 8 | Big | Red |
| 9 | Big | Green |
Observed is different from expected
In 20 draws, the expected count for a particular digit is 2. It need not appear exactly twice. Counts use the entire run, even though the recent-history list shows only the latest 20 results. The historical-analysis guide explains sampling variation.
Local educational simulation only. No live results, external predictions, accounts, or financial transactions.