App Rejections

This is beautiful, 3D Camera []:

1. Apple creates new API, tells everyone to use it
2. Apple advises everyone to check that features are usable before using them
3. A developer uses the feature, but checks the iPhone version, and provides a graceful fallback
4. Apple rejects the app … for using Apple’s “recommended” API’s. Why? Because old iPhone’s don’t support them

The key thing about Juicy Bits experience is that both apps eventually got accepted, which strongly supports their interpretation of why they were REALLY rejected.

This seems to be a case of the reviewer(s) failing: they’ve been issued a new too, but they’re not using it correctly (maybe the tool itself hasn’t been properly explained to them). We’ve noticed that since summer 2009, Apple has starting rejecting a *lot* of apps incorrectly – failing to follow their own stated policies. I’d speculated that they’d switched from “trusing their own staff to make decisions based on guidelines” to “requiring that staff follow a prescribed checklist”.

For 3D Camera and 3D Camera Lite, Apple asked that we set the minimum OS to 3.1 and re-submit. Presumably, this is because the reviewer saw the camera overlay and also noticed that the app would run on a 3.0.x device. Our guess is that the reviewer never tried the apps on a 3.0.x device to see that the camera overlay feature is disabled. Because so many of our users are still running a 3.0.x device, we chose to ignore this advice and instead tried to explain the situation to Apple. After a very slow and non-interactive e-mail exchange, 3D Camera and 3D Camera Lite were both approved.

We know (from decades of academic research) that people “think less” when following pre-written lists and make more mistakes-of-judgement, but fewer mistakes-of-process. I’ve heard of a lot of rejections recently where it seems that the reviewer (human) made a mistake and ran the tool wrongly, and instead of noticing the strange result, just took it literally. Didn’t use their common-sense / judgement at all…

(or maybe … the programmers who wrote the tool distinguished between “recommendations” and “requirements” – i.e. they knew when the tool’s output is “probably, but not definitely, correct” – but some of the human reviewers, now wanting to “let the computer do the work for me”, are reading every recommendation as a dictat)