# What should your first game be? Simple types to start with

> Your first game should be small: a single action, like tapping or swiping, a single screen, and rounds that last a few minutes. One-tap games, simple puzzles, and endless runners fit that rule. Pick the type you have played the most. And cheap to make is not easy to sell: of the 965 games that play themselves released on Steam in 2025, 27 reached 1,000 reviews.

Summary for when you're short on time

- A good first game has a single action, fits on one screen, and has short rounds; the dream game can wait.
- Flappy Bird, a one-tap game, was made by a programmer over a holiday weekend in 2013 and reached the top in more than 100 countries.
- 2048, a number-merging game, had its first version made in a weekend, based on a similar game that already existed.
- Subway Surfers, a game where you run nonstop, took about five months, with two studios working together.
- Balatro, a card game, took more than two and a half years: simple rules do not mean little work.
- Of the 965 games that play themselves released on Steam in 2025, only 27 reached 1,000 reviews, close to the store average.
- In our view, the type you have played the most is the one you can judge best.

Your first step: Write down the names of three simple games you have played a lot and note what the player does in each one.

By Daniel Novi. Published October 14, 2026. Updated October 14, 2026. Address: https://www.gamearchitectsai.com/en/start-here/what-should-your-first-game-be

## What is the best type of game to start with?

A small game that passes three questions. Does the player do a single action, like tapping, jumping, or swiping? Does the game fit on a single screen? Does a round last a few minutes? If the answer is yes to all three, it is a good first game (our take).

It sounds like very little, and that is on purpose. Your first game is there so you reach the end: decide what the game is, test it, fix it, and see it working from start to finish. Every extra character, level, or screen is one more thing to make and fix, and a game that never gets finished never teaches you the last stretch of the road.

This holds with AI too. It writes the code, which is the set of instructions that make the game work, from what you describe in plain English. But deciding the size of the game is still your choice. See [what the AI does and what stays with you](/en/start-here/make-a-game-without-coding).

## Why not start with your dream game?

Because the dream game is usually huge: a whole world, dozens of levels, a long story. As a first game, it is too big, and the risk is giving up halfway.

Flappy Bird shows the other path. In April 2013, Dong Nguyen was 28, lived with his parents in Hanoi, Vietnam, and worked programming location devices for taxis. Over a holiday weekend, he made a little phone game. He wanted a simple, challenging game, and he expected to earn a few hundred dollars a month from ads[1].

How it plays: the bird moves to the right on its own and keeps falling; it only goes up when you tap the screen. Each pair of pipes it gets through is worth one point, and hitting a pipe or the ground ends the round[2]. According to Wikipedia, the first version was much easier, the creator found it boring, and he made the game harder[2].

The game sat forgotten in the store for eight months. In February 2014, it was at the top in more than 100 countries and had passed 50 million downloads[1].

Three caveats before drawing a lesson from this. Flappy Bird was not made with AI, and its creator already worked in programming. It is an exception: most small games never come close to that. And the story tells you how small a game can be, not how successful it will be.

The dream game does not go in the trash. In our view, it goes on the top shelf, for after you finish one or two small ones.

## What are the simple types of games to start with?

They are types with a central rule you can explain in one sentence. Below, each one is explained before the name it goes by in the field, with a famous example. None of the examples was made with AI.


### Tap the screen and nothing else: the one-tap game

The player has a single button: tapping the screen. The game does the rest on its own. Flappy Bird is the classic example[2]. Geometry Dash is similar: a figure runs on its own and you tap to jump over obstacles. Its creator, Robert Topala, says there was no detailed plan: the game started as a square that could jump and crash, and he kept adding things until it was good[3].

**Why it is good for starting out:** one rule, little art, short rounds. **Where the work is:** getting the difficulty right, which is what makes people want to play again.

### Move pieces until it is solved: the puzzle

The player moves pieces on a screen to reach a goal. In 2048, you slide numbers on a grid and merge the matching ones, until you reach 2048. Gabriele Cirulli was 19 when he made the game, over a weekend in February 2014. He said himself that he based it on an earlier game with a similar idea, 1024. 2048 passed 100 million games played, and its creator said he had earned a few hundred dollars in donations[4].

**Why it is good for starting out:** it fits on one screen and needs little art. **Where the work is:** coming up with challenges that are neither easy nor impossible.

### Run nonstop and dodge: the endless runner

The character runs all the time, the path never ends, and the player only jumps and dodges. In the field, this is called an endless runner. Subway Surfers is the best-known example: the character never stops running, and the options are to jump and to dodge[5].

Subway Surfers is a good reminder of scale. Its creators were animators, and the game was made together with another studio, Kiloo, in about five months: design work started on December 20, 2011, and the launch was on May 24, 2012[5].

**Why it is good for starting out:** one action, a clear rule. **Where the work is:** the art. Backgrounds, the character, and obstacles are on screen all the time and have to match.

### Jump from one platform to another: the platform game

The character walks sideways and jumps between platforms, avoiding pits and enemies, until the end of the level. It is Mario's type of game, and how Mario began is told in [famous games that started small](/en/stories/famous-games-that-started-small).

**Why it is good for starting out:** everyone gets it right away. **Where the work is:** each level has to be designed and tested, and the jump has to feel right. And it is a crowded type: in a study with Steam data, 2D platform games had the worst chance of all of reaching 1,000 reviews[8].

### Build the right hand: the card game

The player gets cards and chooses which ones to use to hit a target. Balatro is a poker game with new rules: you play poker hands to score points, and special cards, the jokers, bend the rules in your favor[6].

The rules fit in one sentence, but the work does not. Balatro was made by one person, over more than two and a half years, and started as a side project[6].

**Why it is good for starting out:** little movement on screen and simple art. **Where the work is:** balancing the cards so that no combination gets too strong.

### The game that plays itself: the idle game

The game runs on its own and the player only decides where to spend what it earned in order to grow faster. In the field, this is called an idle game, or incremental game. A 2026 study of game types on Steam points to it as the cheapest to make, by a wide margin[8].

**Why it is good for starting out:** little action on screen, little art. **Where the work is:** making the numbers go up at a pace that makes people want to come back. And the warning is just below.

### One step up: you only move, and the weapons attack on their own

In Vampire Survivors, the character's weapons attack on their own and the player controls where they go, in a level that never ends[7]. The first version for sale took about a year, and its creator spent about £1,100 on art, music, and other game materials[7]. It is simple to play, but there is a lot more on screen than in a one-tap game.

### The types side by side

| Type | What the player does | Art | Where the work is | Example |
|---|---|---|---|---|
| One tap | Taps to jump or fly | Little | The difficulty | Flappy Bird[2] |
| Puzzle | Moves pieces on a grid | Little | The challenges | 2048[4] |
| Endless runner | Jumps and dodges nonstop | A lot | Backgrounds and character | Subway Surfers[5] |
| Platform | Walks and jumps between platforms | Medium | Each level and the jump | Mario |
| Cards | Picks cards to hit a target | Little | Balancing the cards | Balatro[6] |
| Idle | Decides where to spend what the game earns on its own | Little | The pace of the numbers | PokeIndie study[8] |

The "Art" column is our take, based on how each game works.

## Which types of games are better left for later?

The ones that multiply the work by their very nature, even with a simple idea. In our view, there are three.

- **Online multiplayer games.** Each player is on a different device, and everyone has to see the same thing at the same time. Jonathan Bjerk, a Norwegian accountant who made a game like this with AI, spent two months stuck on errors of this kind. His advice for beginners is to first make the game work well for a single person[9].
- **A big world to explore.** Every place, character, and task is more content to make and test. Stardew Valley, a farm game with animals, a town, fishing, and caves, took four years of one person's work, and he taught himself to make the music and the art[10].
- **3D games.** A character, backgrounds, and a camera in 3D require much more art and much more tuning than a screen seen from the side or from above.

None of these is forbidden. They just work better as a second or third game, when you already know what it is like to reach the end of one.

## Is cheap to make easy to sell?

No. The cheapest type to make has the same chance as any other game, and some simple types have a lower chance.

The most common yardstick for saying a game "took off" on Steam, the biggest computer game store, is reaching 1,000 player reviews. The study itself calls this yardstick the least bad one available, not a perfect one[8].

| Type of game on Steam | Released | Reached 1,000 reviews | Rate | Source |
|---|---|---|---|---|
| Idle, the game that plays itself (2025) | 965 | 27 | 2.8% | PokeIndie[8] |
| Card games where you build a deck and start over each run (2025) | 212 | 11 | 5.1% | PokeIndie[8] |
| Average of all games in the store (2025) | | | 2.99% | PokeIndie[8] |
| 2D platform (2024) | | | 0.25% | PokeIndie[8] |

The study comes from a studio's blog, with data gathered by Chris Zukowski, who has tracked this yardstick for years[8]. The figures are for Steam; on the mobile stores, the math is different.

**The other side.** A simple type guarantees nothing. The same simplicity that makes a game cheap to make also makes it easy to copy, and the store is full of similar games. Flappy Bird and 2048 are exceptions, made by people who already programmed, and without AI. Your first game is for learning how to finish and publish; expecting a big result from it is betting high on a small game.

## How do you choose your first game?

By the type you have played the most. That is where you can tell, without thinking, what is good and what is dull, and that counts for a lot when you are the one testing (our take).

Then run the idea through the three questions from the start: a single action, a single screen, rounds of a few minutes. If it doesn't pass, cut until it does. A few more tips, also our take:

- **Choose a single-player game.** A game for several people online is much harder to get working.
- **Go for simple art.** Shapes, colors, and few characters. Geometry Dash started as a square[3].
- **Think about where it will run.** A one-tap game suits phones; card games and puzzles work well on both.
- **Accept that it will be small.** A small, finished game teaches more than a big, half-done one.

How long each size usually takes is in [how long it takes to make your first game](/en/start-here/how-long-does-it-take-to-make-a-game).


The first step fits on a sheet of paper: [write your game idea in three sentences](/en/start-here/how-to-start-making-a-game), covering what the player does, how they win, and why they would play again.

## Your action plan

- [ ] Write down the names of three simple games you have played a lot.
- [ ] For each one, note the single action the player does and how they lose.
- [ ] Choose one of the types on this page for your first game.
- [ ] Run your idea through the three questions: a single action, a single screen, rounds of a few minutes.
- [ ] Cut from the idea everything that does not pass, and keep it on a "later" list.
- [ ] Write your game in three sentences: what the player does, how they win, and why they would play again.
- [ ] Keep the dream game on a separate sheet, for when you have finished one or two small ones.

## Related questions

### What is the simplest type of game to make?

In our view, the one-tap game, where the player only taps the screen, as in Flappy Bird. It has a single rule and little art. The work is in getting the difficulty right.

### Can my dream game be my first game?

You can try, but the dream game is usually too big for a first one. A small game is one you finish, and finishing teaches things no half-done project can. The dream game can wait until after one or two small ones.

### Is a platform game good for starting out?

It is easy to understand, but each level takes work to design and test. And, in a study with Steam data, 2D platform games had the worst chance of reaching 1,000 reviews: 0.25% in 2024.

### Is it worth making a game that plays itself, the so-called idle game?

For learning, it can be: a 2026 study points to this type as the cheapest to make. For selling, the chance is the same as for any game: of the 965 released on Steam in 2025, 27 reached 1,000 reviews.

### Was Flappy Bird made with AI?

No. It was made in 2013 by Dong Nguyen, who worked as a programmer. None of the famous games on this page was made with AI.

### Are card games simple to make?

The rules can be simple, but balancing the cards takes a lot of work. Balatro, a game built on poker cards, took more than two and a half years to finish, made by one person.

### How do I choose the type of my first game?

In our view, by the type you have played the most, because that is where you can tell what is good. Then check the three rules: a single action, a single screen, rounds of a few minutes.

## Sources

1. Rolling Stone, "The Flight of the Birdman: Flappy Bird Creator Dong Nguyen Speaks Out" (March 11, 2014, Internet Archive copy): https://web.archive.org/web/20150101112504/http://www.rollingstone.com/culture/news/the-flight-of-the-birdman-flappy-bird-creator-dong-nguyen-speaks-out-20140311 (accessed October 14, 2026)
2. Wikipedia, "Flappy Bird": https://en.wikipedia.org/wiki/Flappy_Bird (accessed October 14, 2026)
3. Cult of Mac, interview with Robert Topala about Geometry Dash (June 27, 2014, Internet Archive copy): https://web.archive.org/web/20250719143527/https://www.cultofmac.com/news/geometry-dash-robert-topala (accessed October 14, 2026)
4. CNBC, "19-year-old makes viral game hit in a weekend" (March 26, 2014, Internet Archive copy): https://web.archive.org/web/20150929225056/http://www.cnbc.com/2014/03/26/19-year-old-makes-viral-game-hit-in-a-weekend.html (accessed October 14, 2026)
5. PocketGamer.biz, "Mobile Masterworks: Subway Surfers" (February 7, 2023): https://www.pocketgamer.biz/mobile-masterworks-subway-surfer/ (accessed October 14, 2026)
6. Wikipedia, "Balatro": https://en.wikipedia.org/wiki/Balatro (accessed October 14, 2026)
7. Wikipedia, "Vampire Survivors": https://en.wikipedia.org/wiki/Vampire_Survivors (accessed October 14, 2026)
8. PokeIndie, "Game Genre Study 2026", with data from Chris Zukowski: https://pokeindie.com/blog/game-genre-study-2026-which-genres-are-worth-building (accessed October 14, 2026)
9. AI Gamechangers, interview with Jonathan Bjerk (April 6, 2026): https://aigamechangers.substack.com/p/ai-was-my-engineering-team-not-my (accessed October 14, 2026)
10. Stardew Valley, official press page: https://www.stardewvalley.net/press/ (accessed October 14, 2026)