Most of our apps started as a copy of another one of ours. That’s a fast way to get a new app with ads, analytics, support and settings already working, and it’s exactly what App Store guideline 4.3 is suspicious of. So on October 15, 2026 we measured how much of each fork is still its parent.
How it was measured
Each of the 29 forks was compared with the one app it was cut from, using our app-similarity script on the main branch of each repo. A source file counts as shared when 90% or more of its lines match the parent’s after the app’s own name is normalized away, so doughball_theme and quilt_calc_theme are compared as the same word. The percentage is shared files over all source files, tests included.
The same run compared the store listing text (fastlane metadata) and the bundled assets: sounds, fonts, images. It did not compare screenshots, and it compared each app with its parent only, not with every sibling.
The numbers
The share runs from 7% (Bramble Knight, against Taco Rain) to 41% (Doughball, against Quilt Calc). The median is 22%.
| App | Forked from | Source files | Shared |
|---|---|---|---|
| Doughball | Quilt Calc | 225 | 41% |
| Practice Proof | Househand | 188 | 40% |
| Sizzle Swap | Asteroid Duel | 235 | 37% |
| Weekly Plate | School Log | 181 | 35% |
| Pinfall | Chess Clock | 245 | 34% |
| Tidepail | Park Ledger | 227 | 34% |
| Notebird | Practice Proof | 227 | 30% |
| Chalkline | Notebird | 215 | 30% |
| Puddle Jumper | Wordasaurus | 268 | 28% |
| Stubbook | our app template | 155 | 27% |
| Slap Pile | Sizzle Swap | 204 | 26% |
| Park Ledger | Househand | 281 | 25% |
| Covert Mail | Notebird | 308 | 25% |
| Wobble Wheelie | Taco Rain | 318 | 23% |
| Quilt Calc | Handyman Calc | 355 | 22% |
| Chess Clock | Initiative Tracker | 374 | 19% |
| Tilt Duo | Asteroid Duel | 315 | 18% |
| Pixel Pond | Broadside | 489 | 18% |
| Peel Frenzy | Taco Rain | 368 | 17% |
| Card Lab | Mix-a-Beast | 439 | 15% |
| Broadside | Bramble Knight | 534 | 14% |
| Acorn Trail | Tilt Duo | 458 | 12% |
| Asteroid Duel | Taco Rain | 478 | 11% |
| Pie Cannon | Taco Rain | 485 | 11% |
| Handyman Calc | our app template | 446 | 11% |
| Stitch Keeper | our app template | 494 | 11% |
| Hoverline | Bramble Knight | 363 | 10% |
| Mix-a-Beast | Taco Rain | 380 | 10% |
| Bramble Knight | Taco Rain | 656 | 7% |
Listing text and assets barely registered. 27 of the 29 share no listing text with their parent at all. The other two are Weekly Plate at 14% and Sizzle Swap at 12%. Only Wobble Wheelie shares any bundled assets, 6%, with Taco Rain.
What the shared files are
Doughball is the top of the table, so it’s the one to look inside. It shares 92 of its 225 source files with Quilt Calc. The list is the support form, the delete-my-data sheet, the banner and interstitial ad services, the analytics service, the logger, the crash reporter, the splash screen, the settings tiles, the review prompt and the router. None of the 92 has anything to do with dough. The calculator, the yeast model and every screen a user would recognize are new.
That holds across the table. What a fork keeps is the part of an app nobody downloads it for.
Games share less than utilities
The fourteen games in the table have a median share of 15.5%. The fifteen utilities have 27%.
We think the reason is size. A game’s loop, physics, levels and art replace most of what it was cut from, so the plumbing ends up a small slice. A utility’s feature is often a few screens and a model, sitting on the same plumbing as before. Doughball is smaller than Quilt Calc, which is smaller than Handyman Calc, and the share climbs down that line: 11%, then 22%, then 41%.
Sizzle Swap is the exception among the games at 37%. It was cut from Asteroid Duel for its nearby two-phone link, so it kept the networking code on purpose.
Did the number predict App Review?
Two of these forks have been through App Review with a decision, and both came from Taco Rain.
- Asteroid Duel, 11% shared with Taco Rain: approved.
- Pie Cannon, 11% shared with Taco Rain today: rejected under 4.3(a). We didn’t measure it before that rejection, and it has changed since.
Same number, opposite outcomes. Two cases don’t make a rule, but they’re enough to say source share didn’t decide ours. Our reading, and it is only a reading, is that 4.3 looks at the account: many apps, close together, from one developer.
Six forks sit above the 30% line our script fails at. None of the six has been submitted for App Store review yet, so we can’t tell you what Apple makes of them.
Why we still measure it
A reviewer can’t read your Dart. They can read your listing, look at your screenshots and hear your sounds. The cheap catches are there: a description paragraph copied from the parent, or an icon or sound file that is byte for byte the same. The script fails on those no matter what the percentages say, and this sweep found none on a fork’s main branch.
The source number is still useful as a trend. If a fork’s share is climbing between releases, the new app isn’t growing into its own thing.