Step 1 — The Idea
Every game starts with one sentence. If it doesn't fit, we don't build it. Here's how to find the sentence.
The first thing we do with any new game is try to kill the idea.
If we can describe the game in one sentence, it has a chance. If we need a paragraph, it's too complicated for mobile. Players won't read.
The one-sentence test
# Mole Bash Whack moles to the beat.
# Mirror Match The mirror moves with you.
# Dodge or Die Endless dodging arena.
✓ All four ideas fit in one line. ✓ All four shipped. ✓ All four playable in 5 seconds.
/the-rule
One sentence or it's dead.
Mobile players decide in 5 seconds. If we need a paragraph to explain the game, the player already left the App Store page.
Long ideas hide complexity. Short ideas force focus. The discipline of fitting your idea into one sentence reveals what it's actually about.
Examples of bad ideas
✗ "An RPG where you build relationships while crafting items in a fantasy world with rotating seasons..." → too many words. Too many systems.
✗ "A puzzle game with new mechanics every 10 levels." → what's the core mechanic? Unclear.
✗ "Like Wordle but with shapes." → ok one sentence. But "like X but Y" usually means we don't have an idea.
/red-flags
Watch for these patterns.
If your idea has "and" or "with" connecting multiple systems, it's two ideas. Pick the stronger one.
"Like X but Y" is often a sign you don't have a real hook — you're just remixing.
The 5-question filter
After the one-sentence test, we ask 5 questions. If any answer is unclear, the idea goes back to the drawing board.
✓ Pass all 5 → start building ✗ Stuck on any → re-discuss
# Question 4 is the most important.
It surfaces what could kill the game
before we spend a week building it.
/filter
5 questions, then go or no-go.
This kills 80% of ideas in the first hour. That's a feature, not a bug.
Better to throw away a bad idea on day 1 than to ship a confused game on day 7.
Real example — how BOP passed
Q1. Has anyone shipped this? Yes — but no one with bosses. ✓ differentiation
Q2. Hook in 5 seconds? Ball falling. Tap = it bounces. ✓ instantly clear
Q3. Platform? iOS first. Mobile-native feel. ✓ decided
Q4. Riskiest assumption? Boss fights need feel different. ✓ flagged for design phase
Q5. One input? Yes — single tap. ✓ mobile-friendly
All 5 pass → green-light to build
/case-study
BOP took 12 minutes to validate.
From "I had this idea on a walk" to "we're building it" was a 12-minute back-and-forth with the AI. No 30-page design document. No mood boards.
The 12 minutes saved us 3 days of work that would have died in week 2.
Common pitfalls
✗ Falling in love with the idea before testing it
✗ Adding features to "make it more fun"
✗ "What if it also had..." (kills it)
✗ Skipping question 4 because you're excited
The discipline is harder than it sounds. Killing your own idea on day 1 feels bad. Shipping a confused game on day 14 feels worse.
What you can copy
If you have a game idea right now:
- Write it as one sentence. Out loud. To yourself.
- If it doesn't fit, cut something.
- Ask the 5 questions. Be honest.
- If you stall on question 4, the idea isn't ready. Iterate.
That's the whole step. Tomorrow we discuss it with the AI.