What's New

Build 1 was submitted to the App Store and rejected under Guideline 4.3(a) — Design: Spam: the app shared a concept with apps from other developers.

That is a judgement about the catalogue rather than a claim about the code, and read that way it was fair. Build 1's entire content was the Stroop task and the Schulte table — two public-domain psychology instruments the store already carries in quantity. From a reviewer's screen, four seconds in, it looked like thirty other apps.

Build 2 adds two games that are our own.

The rule we picked them by

The obvious next games were the trap. N-back, Simon, Corsi blocks, Trail Making, Tower of Hanoi, mental rotation — every one is another public-domain classic with exactly the saturation that produced the rejection. Adding two of those turns "two classic tests" into "four classic tests", which is a worse answer, not a better one.

So the rule was: the mechanic has to be something only this app can ship. That left the two things eleven epics had already paid for.

Digit Bridge

The same number, written in two different numbering systems. One side shows it in the digits you read; the other shows it in the digits you don't. Find the match among six candidates.

The wrong candidates are near-misses drawn from a declared taxonomy — a transposition, a substituted digit, a reordering, a length change — never random numbers. Random numbers would make it a spot-the-shape exercise winnable without reading either script.

It exists because MindForge already had to pin a numbering system per locale to ship Persian and Kurdish Sorani at all. That machinery is what the game is made of.

False Light

One imaginary light falls on the whole board. Most tiles stand raised under it; a few are pressed flat. Sweep the pressed ones, and clearing a field deals the next.

There is no text and no colour information on that board at all. A raised tile and a pressed one differ in three channels — the shadow is there or it isn't, the tile sits at rest or in the space its shadow occupied, and the fills are far apart in luminance. So it plays identically for a colour-blind player with nothing switched on, and its content is identical in all four languages.

Also in this build

• The home hub now carries four games, and Stats and personal bests follow automatically.
• Both new games ship in English, German, Persian and Kurdish Sorani.
• Two new accent colours. Only one unused colour in the design system could carry a legible label on its play band, so the second is a deliberate addition to the palette rather than a spare one picked off the shelf.

Three things the simulator caught that the tests did not

Written down because they are the interesting part.

• A four-digit Persian target numeral was losing its last digit — 326 points of glyphs inside a 326-point card. The Latin version fitted at 256. That is the exact shape of a bug that reaches a release, because the developer's own language is the one that works.
• The test written to catch it passed against the build that had it. Asking a laid-out widget for its size returns the size its parent squeezed it into, so an overflowing run reports the box width back and the check is true by construction. It now measures the text itself, unconstrained.
• Tapping a wrong tile in False Light was quietly un-sweeping every tile already cleared, which let the same tile be scored twice.

Known gaps

• Uploaded to App Store Connect and awaiting resubmission. Not yet approved, and 4.3(a) is a reviewer judgement — shipping differentiation improves the argument without settling it.
• No Android build yet.
• The sound toggle is still not wired to any audio, and stays hidden until it is.
• Persian and Kurdish strings — including both new games' — are still machine drafts awaiting a native speaker.
• Still measured on a simulator only, not on real iPhone hardware.