Extracted from a video on effective writing, persuasion, and how to communicate ideas so people actually accept them. These principles apply directly to Project Imagination's game narration, marketing copy, and AI prompting.
Source: WRITING-SKILL.md (video transcript)
Most people think writing is about expressing themselves. That is the starting point for almost all bad writing.
Effective writing is not an act of self-expression. It is an act of engineering a specific desired experience in the mind of another person. The words are not the point. The effect the words have on the reader is the point.
Most people focus on the transmission (syntax, word choice). Masters of this craft focus on the reception (what happens in the reader's mind).
Example: Som Parik's cold email got him hundreds of interviews. It wasn't good because it expressed who he was. It was good because it perfectly understood who he was writing to. He was executing a program in the mind of his target audience.
Before you write anything, deeply understand who you're writing for. What do they believe? What do they value? What are their psychological needs, not just their technical needs?
Parik understood that startup founders don't optimize for process - they optimize for conviction. Their greatest hope is to find someone as obsessed with their company as they are. His email wasn't a skills list. It was a mirror reflecting the founder's deepest desire back at them.
Example: "I don't have many hobbies outside coding. Building is the only thing I'm good at." - This wasn't self-deprecation. It was signaling "I am the ideal obsessive engineer you dream of hiring" to a YC-style founder audience.
The secret to all successful storytelling is world building. Either you fit into a world the audience already lives in, or you build a new one. But even new worlds need familiar anchors - you cannot start from complete scratch.
There are two types:
- Fitting into an existing world (easier) - Understand the audience's world and speak within it
- Building a new world (harder) - Take elements from a familiar world and spin something new on top
Example: Dune borrows Arabic and Islamic terms (Mahdi, Lisan al-Gaib, Muad'Dib) to anchor its world in something readers can connect with. Then it adds sandworms, spice, and entirely new lore on top of that familiar foundation.
Every world, no matter how complex, is built from simple atomic units. These are foundational points that require the least amount of new information for the reader to accept. They are points of common ground - often so obvious they feel unnecessary to state.
Once the reader has nodded along to your atomic units, you can build more complex structures on top.
Examples of atomic units:
- Business proposal: "We all want the company to grow, right?"
- Health article: "Everyone wants to feel healthy and have more energy, right?"
- Political argument: "We all want our children to live in a safe community, right?"
Once they nod, you can introduce: "One of the biggest obstacles to our growth right now is our outdated sales process."
Words like "super lean teams" and "work across the stack" are not normal words most people use in applications. They are in-group language - signaling "I belong. I know your words. I'm a member of the tribe."
When you use the specific vocabulary of your audience's world, you instantly build trust and bypass skepticism.
Example: Parik used YC startup lingo deliberately. Every YC company uses specific jargon. By speaking their language, he signaled membership in their world before making any ask.
Going micro makes worlds believable. Inconsistency breaks immersion. The more consistent detail you add, the more rooted and real the world feels.
Example (Dune): The sandworm spice lifecycle, the hooks used to ride worms by prying open a segment to expose sensitive flesh, the "sand rider" title as a rite of passage, spice as a psychoactive substance - each detail is internally consistent and adds depth.
Counter-example (Superman): Superman catches falling people at high speed (should kill them - explained away as a "protective barrier"). Clark Kent's glasses (explained as "hypnosis glasses"). These lazy fixes break immersion because they're inconsistent with the world's own rules.
Setting up elements early that pay off later builds a richer world and makes the eventual reveal more powerful.
Example (Marvel): Thanos was shown in end-credit scenes across many movies before his full appearance. His first real showing had him beat the Hulk - one of the strongest established characters - instantly establishing him as a credible threat. All the small sequences before the final movies were setting up parts of the universe for the final reveal.
People favor information that confirms their existing beliefs. The most effective persuasion starts by reinforcing what the reader already believes, then gently introduces a modification.
You're not selling an argument. You're selling an identity the reader is already desperate to buy.
Example: Parik's email told founders: "Your worldview is correct. These are the kind of employees you should be looking for. I am one of them." The founder's identity as a leader of a worthy mission was reinforced. He was selling an identity, not a skill set.
You cannot change someone's worldview overnight. Start with what they believe, get them nodding, then gradually introduce your modification. Step by step.
Example: Big Indian investors didn't care about content creation. The pitch started with their world: "You need billion-dollar companies. You need good product and good GTM (go-to-market). Meta and Google ads are expensive and saturated." They're nodding. Then the pivot: "There's a section of society that doesn't have a cost of acquisition - content creators. What if you stole their playbook?" Then ground it with examples: Physics Wallah (content creator + billion-dollar company), Mr. Beast + Feastables, Logan Paul + Prime, Kylie Jenner.
Without examples, your story floats. Examples ground your modified world and make it real. The more examples, the better. When you don't have concrete examples, use a prototype - just root it in some reality people can touch and feel.
Example: When trying to convince people that AI avatars work for content, there were no previous success stories. So a low-resolution prototype was made. Once it succeeded and became a case study, hundreds of other creators could point to it and say "it worked for him, therefore I can do it."
Present the reader with two ideas they believe are both true, but which are in conflict. This creates tension they need to resolve. Your idea becomes the resolution.
Don't brute-force facts that contradict their world. Create the tension, then release it.
Example: "You're saying code is more valuable than content. Okay. But how many apps have you downloaded last year versus how many new creators have you followed? If funding and salaries come because there's usage, and you're consuming far more content than downloading new apps..." The reader's own beliefs create the conflict. The resolution naturally follows.
The true test of a storyteller: without telling a lie, how far can you stretch a person's brain from what they currently believe? The amateur brute-forces with contradicting facts. The skilled persuader validates first, then gently bends.
Example: "You are right to be skeptical of new marketing fads. Most of them are waste of time and money. That's why it's important to look at the few that are grounded in fundamental principles." - Validates skepticism, lowers defense, then introduces the new idea.
Writing simply is not dumbing it down. It is a sign of respect. The brain has an immune system that rejects new ideas and unfamiliar words. The less energy the reader spends decoding your writing, the more the idea gets absorbed.
Use ordinary words. Write clean sentences. Omit needless words. You are giving the reader a clean, efficient program that runs smoothly in their operating system - instead of asking them to download yours.
Why this matters: Very good video editing drops the brain's immune system. The same principle applies to clean, simple writing. The easier you make it, the wider you scale, because you fit into everyone's worldview.
The greatest enemy of clear writing. You cannot imagine what it's like for someone who doesn't know what you know. You must constantly ask: What does my reader know? What is the simplest way to say this? What context am I assuming?
Example: A person who earns 100 crores a year, when speaking on podcasts, always uses "5 lakh" in his examples - because that's what the audience relates to. If he used his real numbers, nobody would connect. It wouldn't be rooted in their world.
Another example: Zerodha founders were making thousands of crores per year in dividends - nobody cared. The minute news said "100 crore salary," everyone lost their minds. People relate to "salary" more than "dividend."
Most writers treat all their points as equal and present a laundry list hoping something sticks. This is the "engineering answer syndrome" - listing many points hoping for part marks. In real life, there are no part marks.
The highest leverage point is the single core assumption upon which the entire opposing view rests. Find the one flawed premise at the root.
Example: Hiring a CFO at a high price. Instead of listing 5 reasons from his CV, the referrer said: "Bro, this is a gut decision. Instead of falling short by two-three lakhs, just close one more deal. Having a CFO means you close your eyes and go faster." One line. That converted.
Another example: "AAA games cost $100 million" - break it down: $25M dev + $75M marketing. Dev cost is driven by US salaries at $200K. Indian starting salary is one-tenth. So the entire dev budget in India could be one-tenth. Attack the root premise, not 10 surface symptoms.
The title and first sentence heavily influence how everything after it is received. They set the frame for the entire piece.
Example: "An app that does RAG with LLMs" - boring, technical, no one cares. "Rahul is using this app to prepare for his exam in 5 minutes" - immediately shows a compelling human benefit people can understand.
Humans are wired for stories, not numbers. A concrete story is more memorable and more persuasive than any statistic.
Example: "92% of customers are satisfied" - nobody cares. "Himish used to spend all his Friday afternoon updating spreadsheets. Now that process is automated and last Friday he closed two extra deals. That's five lakhs extra." - That's tangible.
- World building (#3, #6, #7): Each chamber needs consistent micro-details, in-group language, and foreshadowing to feel real
- Atomic units (#4): Start each chamber with grounding details the player can latch onto before introducing the strange or complex
- Cognitive hospitality (#13): Keep narration clean and simple - the player's imagination does the heavy lifting, not verbose prose
- Consistency (#6): If a chamber establishes rules (e.g., "the door only opens when spoken to"), never break them without narrative reason
- Stories over stats (Bonus): Show consequences through narrative, not numbers. "Your torch flickers and dims" hits harder than "Torch: 40% remaining"
- Audience first (#1, #2): Every post should be written for the reader, not about the product
- In-group language (#5): Use the vocabulary of each platform's community (Reddit speaks differently from LinkedIn)
- Frame shift (#9): Start with what the audience already believes ("You've played hundreds of games with amazing graphics"), then pivot ("What if the best game had no graphics at all?")
- Anchor with humans (Bonus): Lead with a person and a story, not a feature list
- One core argument (#15): Don't list 10 features. Find the one thing that makes someone say "I need to try this"
- Prompting IS world building (#3): Every prompt creates a universe for the model. Vague prompts create generic worlds. Detailed prompts with specific rules, examples, and constraints create rich, specific worlds
- Atomic units in prompts (#4): Ground the AI with clear foundational rules before adding complexity
- Examples ground the world (#10): Few-shot examples in prompts are like case studies - they show the AI what the world looks like in practice
- Set laws, not suggestions (#6): The prompt is the physics of the AI's universe. Be specific about what should and should not happen