Ask Liu Cixin how to write science fiction and the answer keeps returning to one thing: the idea. If a concept doesn't excite him, he won't write it, and he assumes readers would feel the same boredom he does.

The Direct Answer
Start with an idea that genuinely excites you, then build the story and characters around it.
Liu has said he writes only when an idea moves him, and that he plans the story around that idea before creating any characters (Nature, 2016).

Let the Idea Lead

Liu's advice mirrors his own method. He starts with a speculative idea strong enough to move him, plans a story around that seed, and creates characters last, to serve the story. He has been blunt about the filter he uses before committing to anything.

Write only when I have a creative idea that makes me excited and demands to be written. If I can't even be excited by an idea, readers will surely be bored as well.

Liu Cixin, Nature Future Conditional Q&A, 2016

For Liu, science fiction is a literature of ideas, and the idea about science or technology sits at the core of the story. He has said this method inevitably produces books where the speculative concept drives everything and character comes second.

Read Widely, Think Hard

His other recurring piece of advice is about input. Liu has called reading widely the cornerstone of creation, and he treats the frontiers of science, especially physics concepts that break common sense, as raw material for stories. The point is to gather more than you need, then find the concept that resists prediction and build from there.

None of this happens fast. Liu is a deliberate thinker who lets ideas accumulate before he drafts, an approach shaped by years of writing around a full-time job. See the full Liu Cixin writing routine for how that played out in practice.

What You Can Steal