Finish Your Game Instead of Starting Over: 10-Day Plan
To finish your game instead of starting over, cut it down until one complete round can be played from start to end, and put everything else on a list called "not this game". The urge to restart usually comes from a project that has grown too big, not from one that is beyond saving.
Should you really start over?
Say you are making Comet Laundry, an invented game about washing the tails of comets. This is your third restart. Each new version began with a burst of energy, then slowed down in the same place: the boring middle, where the fun idea is done and the unglamorous work starts.
Before you open a new folder, answer these three questions:
- Is the main action fun with no art at all? If not, the idea is the problem. Change it, keep what you learned, and restart. This is the one honest reason to do so.
- Does the main action work? If yes, you have something a new project does not have yet.
- Is one part so tangled that two careful fixes have failed? Then rebuild that part. You do not need a new game for that.
A new project feels fast because nothing is connected yet. That feeling lasts about as long as it did the last two times.
What does "finished" mean?
Finished does not mean the game you imagined. For a first game, it means:
- A stranger can start it, play one complete round and reach an ending.
- Nothing in that round breaks.
- It is online somewhere that you can send a link to.
For Comet Laundry, that cuts forty planned comets to six, and it drops the shop, the crew and the story mode. They all go onto a list for the next game, which is not the same as deleting them.
What is the ten-day plan?
An hour or so a day is enough. The days are a guide, not a rule, but the order matters.
| Day | Job | Done when |
|---|---|---|
| 1 | List everything that exists and everything that is planned | The list is written |
| 2 | Sort it into Must, Nice and Cut | Cut is the longest list |
| 3 to 5 | Finish the Musts, one at a time | One round can be played to the end |
| 6 | Add a start screen and an ending | A round has a clear beginning and end |
| 7 | Test with a silent stranger | You have notes |
| 8 | Fix the three worst things on those notes | They are gone |
| 9 | Add a title, sounds and a short how-to | You are not embarrassed by it |
| 10 | Publish it on itch.io | You have a link |
No new ideas between day 3 and day 10. When one shows up, and it will, write it on the list and carry on.
Paste this to your AI
I keep restarting my game. Help me finish this one instead. Do not start anything new.
1. Read the project and draw a Mermaid diagram of one complete round, from the start screen to an ending, including how you lose.
2. Mark each step in the drawing as "works", "half done" or "missing". For "works", say what shows it: the place in the game, and what you tried. If you cannot tell, write "unknown" and do not guess. Show me the first spot where a player would get stuck.
3. Sort everything else in the project into Must, Nice and Cut. A Must is something the round cannot work without. Be strict, and aim for Cut to be the longest list.
4. Write not-this-game.md with everything in Cut, so that nothing is lost.
5. Give me a plan for the Musts, smallest first.
If I ask for a new feature before the round is finished, quote this plan and ask me if I am sure.
Or skip this and connect the MCP: your AI then finds all of this by itself.Connect the MCP
How to check it worked
- Read the Cut list. It should sting a little. If it does not, you kept too much.
- Follow the drawing to the first "missing". That is your next job, and nothing else is.
- Check one "works". Try it yourself.
- Ask for a new feature on purpose. Your AI should quote the plan back at you.
- Tell someone your publish date. Saying it out loud to one person keeps the plan honest.