Isles of Stability and the Perception of Apple's Software Getting Worse

On the most recent episode of the Accidental Tech Podcast (#99), the fine hosts discussed a number of things related to whether Apple's operating systems have become less stable or well executed. They made a few points that I think emphasize the sensation that things are worse, even if one could argue that this is part of a routine cycle.

Isles of stability. New OS releases always have teething pains, and if we're lucky, a rocky 10.x.0 is fully stable and good by 10.x.2 within about two to three months. (Dirty secret of all software: when version x.0 ships, they're already working on x.1, because they ship with known problems, but the damned thing has to get out the door. ) However, when OS releases were more than a year apart, we might have a few months of pain, during which period we would either delay or deal with the problems, and then exit what John Siracusa called the regions of pain. We would have sometimes a year or much longer in an island of stability. Marco Arment said, "Now it seems like we're always using a 1.0 or a 1.1…because the updates are moving so quickly."

Joran Elias compiled this chart of major and minor releases. The X axis is time; the Y axis breaks the operating system versions apart. The dots represent each minor release for a given cat/California place. The critical number are the number of days between major releases. If you pair Leopard/Snow Leopard and Lion/Mountain Lion together, you can see how little time Mavericks had.

Cloud dependencies. Because so much of iOS and OS X has a cloud component, any failure in syncing or availability makes little problems seem far worse, and provides an overall sense of gloom that may not be backed up by the actual experience and uptime. John recounted his wife having an iCloud Contacts syncing problem which I did as well: without any notification of failure, syncing stopped, and it required mucking about to restart it, instead of it auto-healing (or even warning me or his wife).

They should do better by now. Even if, as John argues, Apple is no worse at software now than in the past, at this point of maturity and sophistication, it's simply unacceptable to be as good as they were in the past. (This says nothing of breaking features that work.) Apple can and should be held to higher standards of software development quality than a decade ago. And many of its problems can be tracked in some measure to quality assurance (QA), which involves rigorous testing of changes or additions to make sure nothing breaks. Either QA remains weak at Apple, which is bad because developers shouldn't be relied upon to test all the iterations of their own work against real-world scenarios; or QA teams aren't provided the tools to send work back to developers to be fully fixed.

Together, these provide the feeling of things being worse, even if they aren't. I'd argue that the sheer quantity of stuff that doesn't work the way it should is huge, and has gotten worse, but the sensation is probably what drove me and others to write about it. In the past, we expected stuff would eventually get fixed; now, it just feels like things are broken all the time, and don't improve.