How I Stress-Test Fiction
While working on a political thriller, I realized that making fiction feel real is less about collecting facts and more about challenging the right assumptions.
Because my story involved intelligence work, terrorism, government decisions, and the Middle East, I spent a lot of time understanding the political, legal, historical, and practical background needed for each scene.
One of the biggest things I learned is that research can easily become a trap for fiction writers.
You could spend hours finding out how intelligence agencies operate, how governments respond to terrorism, what laws might be relevant, or how countries deal with one another. Yet that research doesn't necessarily make a situation feel realistic.
Eventually, I broke the process down into three main questions:
- Is it actually possible?
- Is it possible that this event, decision, technology, political arrangement, or method of operating could actually exist?
- Does it seem logical?
- Even if it were possible for something to happen, would the characters actually make those choices given what they know, what they want, and the risks they have to face?
- Will the reader accept that?
- This question is unlike the first two, and even if it is realistic, it may still seem wrong to the reader if the writer has not provided sufficient context.
The third question proved especially useful to me during revision.
I also experimented with the use of AI in this process; although I didn't use it to write the novel or develop the story, I found it was most useful for putting my manuscript through its paces—stress-testing it.
For example, I would describe the situation around a scene and ask questions like:
- What assumptions am I making here?
- What would a skeptical reader question?
- Is there an obvious logical gap?
- Does the character have sufficient information to make this choice?
- What other course of action would a reasonable person take?
- Does the political or operational logic add up internally?
The responses I received were all over the place, but that proved to be useful.
At times, the AI highlighted a weakness I had missed. Other times, it proposed inappropriate ideas because it either overlooked relevant context or didn't understand how people or organizations actually behave.
This made me think carefully about each suggestion instead of just accepting it.
For me, the most helpful way to use AI in writing is to treat it not as an expert, but as a tough reader who challenges my work.
The writer still has to decide:
- whether the criticism is valid;
- whether the research supports it;
- whether fixing the issue improves the story;
- and whether realism is in fact serving the narrative.
I get better results when I look into specific problems in my manuscript rather than trying to learn everything about a topic at the beginning.
So instead of thinking, "I need to understand intelligence operations," I ask much more focused questions:
"How would information realistically move between these two organizations?"
"What legal or political obstacle would prevent this action?"
What specific knowledge could this character have at this exact moment?
What repercussions would this decision have?
These kinds of questions lead to research you can really use in your story.
I'd love to hear from other self-published fiction writers about how they approach this:
How do you check if a scene feels believable, not just accurate?
When you use AI as you revise, does it help you more in generating new ideas or in questioning the ideas you have already written?