# AI use cases for Falsify the Problem (ordinary-s/falsify-the-problem)

Repo: [ordinary-s/falsify-the-problem](https://github.com/ordinary-s/falsify-the-problem) | Use cases: 3 | Page: https://dedupedai.com/use-cases/ordinary-s-falsify-the-problem/

falsify-the-problem is an open-source Agent Skill, delivered as one SKILL.md file, that challenges a problem framing with cheap evidence before any solution is designed.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Test a proposed fix before anyone builds it: https://dedupedai.com/use-cases/ordinary-s-falsify-the-problem/test-proposed-fix-before-building/
- Find out why users stop returning before building reminders: https://dedupedai.com/use-cases/ordinary-s-falsify-the-problem/separate-forgetting-from-missing-value/
- Check for an input mismatch before blaming the model: https://dedupedai.com/use-cases/ordinary-s-falsify-the-problem/check-input-mismatch-before-blaming-model/

## How to test a proposed fix before anyone builds it with Falsify the Problem

Repo: [ordinary-s/falsify-the-problem](https://github.com/ordinary-s/falsify-the-problem) | Difficulty: Intermediate | Time: 15 minutes | Basis: From the README

Check whether a requested fix, such as a cache, targets the real cause of a problem.

**Who it is for:** An engineer whose team has proposed a specific fix, such as adding Redis, after an API slowdown, and who wants to check the framing first.

**The problem:** A solution is often proposed before the problem has been tested, so the team may optimize the wrong thing.

**The outcome:** A verdict (KEEP, WEAKEN, KILL, REFORMULATE, or INSUFFICIENT EVIDENCE) with a clear handoff status before solution design starts.

### Steps

1. Open a terminal in the root of your project.
2. `mkdir -p .agents/skills`
3. `git clone https://github.com/ordinary-s/falsify-the-problem.git .agents/skills/falsify-the-problem`
4. In the Codex prompt, type `$` and select Falsify the Problem, or use `/skills`. If it does not appear, restart Codex.
5. Paste the latency problem and the proposed fix into the prompt and ask for a verdict before any solutioning.

### Prompt to paste into your agent

```text
Use the falsify-the-problem skill in this repo on this problem: <the latency problem and the proposed fix>. Inspect the locally available evidence, separate observations from assumptions, and return one verdict with its handoff status. Verify that no implementation has started before the verdict is given.
```

### Caveat

The README's examples are illustrative scenarios, not measured results, and macOS and Linux installation has not been verified in the documentation.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/ordinary-s-falsify-the-problem/test-proposed-fix-before-building/

## How to find out why users stop returning before building reminders with Falsify the Problem

Repo: [ordinary-s/falsify-the-problem](https://github.com/ordinary-s/falsify-the-problem) | Difficulty: Intermediate | Time: 15 minutes | Basis: From the README

Test whether users forgot the product or never got value from it.

**Who it is for:** A product manager whose team sees low retention and wants to add push notifications right away.

**The problem:** Low retention can come from forgetting or from never experiencing value, and each needs a different fix.

**The outcome:** A tested framing of the retention problem and one cheap observation that separates the two explanations.

### Steps

1. Open a terminal in the root of your project.
2. `mkdir -p .agents/skills`
3. `git clone https://github.com/ordinary-s/falsify-the-problem.git .agents/skills/falsify-the-problem`
4. In your agent, invoke the Falsify the Problem skill by name.
5. Describe the retention problem and the proposed push notifications, then ask for the cheapest observation that would tell the two explanations apart.

### Prompt to paste into your agent

```text
Use the falsify-the-problem skill on this retention problem: <the retention data and the proposed push notifications>. Identify the current framing and materially different alternatives, then name one primary discriminating test. Verify that the output lists which observations are facts and which are assumptions.
```

### Caveat

The skill relies on the evidence you provide or that the host agent can access, and it does not measure retention itself.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/ordinary-s-falsify-the-problem/separate-forgetting-from-missing-value/

## How to check for an input mismatch before blaming the model with Falsify the Problem

Repo: [ordinary-s/falsify-the-problem](https://github.com/ordinary-s/falsify-the-problem) | Difficulty: Advanced | Time: 30 minutes | Basis: From the README

Test whether training and evaluation inputs differ before attributing poor results to model capacity.

**Who it is for:** A machine learning researcher whose experiment underperformed and who is about to increase model size.

**The problem:** Teams often assume the model lacks capability when preprocessing differences may explain the result.

**The outcome:** A verdict on whether the model capability framing survives, with a named test comparing the same saved inputs through both preprocessing paths.

### Steps

1. Open a terminal in the root of your project.
2. `mkdir -p .agents/skills`
3. `git clone https://github.com/ordinary-s/falsify-the-problem.git .agents/skills/falsify-the-problem`
4. In your agent, invoke the Falsify the Problem skill by name.
5. Describe the underperforming experiment and ask the skill to test the input path before any change to model capacity.

### Prompt to paste into your agent

```text
Use the falsify-the-problem skill on this experiment: <the experiment, the saved inputs, and the training and evaluation preprocessing code>. Test whether an input-path mismatch contributes before attributing the result to model capability. Verify that the output names the single primary discriminating test used in each round.
```

### Caveat

The skill can only compare inputs it can access, so results depend on the saved data and code you share with the host agent.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/ordinary-s-falsify-the-problem/check-input-mismatch-before-blaming-model/