Methods · Perspective, simulation & context · Short guide

Counterfactual simulation

Change one variable in a historical or systemic scenario and reason out what follows: training causal discipline instead of storytelling sprawl.

When to use this: When you want to go deeper
Learning impact: Reliable gain
Learning evidence: Credible with caveats
Where you are with the topic: Applying in new contexts
How often to use it: Worth repeating often

When to use it

You understand what happened, but not which factor actually did causal work.
You want to test a claim of the form “if X had been otherwise…”
Your essay is descriptive and you need a disciplined what-if, not fan fiction.

What to ask AI

Type something like: “Counterfactual: change ONLY one variable, the harvest fails in 1845 Ireland. Hold everything else at period constraints. Help me reason consequences; do not invent a novel.”

Ask it to keep one change, track downstream effects, and mark where the chain becomes speculative.

You propose the warrant for each link; do not outsource the whole causal story.

How it helps

A careful counterfactual isolates one change so you can see which later events depended on it.

Multiple changes at once identify nothing: you cannot tell what did the work.

The method trains causal discipline, not alternate-history entertainment.

How to run it

State the factual baseline in a few sentences first.
Change exactly one variable.
Reason forward under period constraints; forbid hindsight tools.
Mark each link as necessary, enabling, or speculative.
Stop when speculation outruns anything you could evidence.

Example exchange

You

Baseline: penicillin reaches battlefield medicine in WWII. Counterfactual: clinical use delayed ten years,one change only. What follows for infection deaths?

First-order: wound sepsis stays higher in that window. Second-order effects get speculative fast,doctrine, logistics, and other drugs also move. Where do you want me to stop inventing?

You

Stop after the first-order sepsis claim unless I bring evidence. Don’t invent geopolitics.

Agreed. Anything about wars “ending differently” is story unless you bring sources. What’s your one warrant for the sepsis claim?

Copyable prompt

Baseline: [FACTUAL SCENARIO].
Counterfactual change (ONE): [ ].
Help me reason, but enforce: no additional silent changes; demand mechanisms;
make me rank outcomes; stop me if I smuggle a second change. Afterwards ask what
this implies for the real-world causal claim [CLAIM].

The Tell

You know this method has flipped when your “analysis” is just a cooler timeline with extra plot twists.

If you changed more than one thing, you learned nothing about causes.

One change, marked speculation, or it is fiction.

Principle evidence

Strength of the underlying learning idea, not a claim about AI products.

The underlying learning idea is rated moderate. Counterfactual reasoning is used in history education and causal reasoning training: change one variable, trace consequences. Related to causal model building.

AI delivery evidence

Whether an AI tutor delivers this method well is a separate question.

Principle moderate; AI causal chains can be glib, speculative.