For those new to the discussion, here is a summary of what we know, compiled from the various threads that address this issue:
Every sign points to a software problem with Mavericks, not hardware.
- It didn't happen before Mavericks (people may have experienced crashes for several other reasons, but from experience this specific crash started with Mavericks).
- It is not specific to one chipset - it is happening on many different MacBook models.
- Replacing hardware makes no difference. Not one person has come back after a hardware replace and said this has fully fixed the problem.
- Other software/perhipherals do not affect this problem. It's an OS issue.
The one counter indicator is that some people have reported that a small number of machines in Apple stores appear to be immune to this issue. However I personally believe there's something else happening in the background on those machines. I do not believe there is such a thing as an "immune" computer, just computers that for some reason don't trigger at the time.
There are multiple threads dealing with this problem now, each with over 10,000 views: e.g. https://discussions.apple.com/thread/5548226
If someone does end up speaking to an Apple software engineer, the following may be of use.
There are two software "bugs" that are related to this issue:
1) There is a race condition in the shutdown/power up routine.
That's the technical programming term for these types of situations. Two or more bits of code are running and working off the same data. The race condition happens when either bit of code assumes that the data isn't going to change before the code is finished, but then the other code changes it anyhow. With low-level code race conditions usually lead to fatal crashes as the instruction table gets corrupted.
2) There is no end-user workaround as we've lost the ability to re-map the power button, or activate the "confirm" dialog that existed before Mavericks.
I've tried various software packages to do remapping - they don't work. They worked pre-Mavericks. I *think* the reason is that they might have changed the kernel code handling interrupts from the keyboard and other sensors. Meaning when you hit the button, very low level functions in the OS decide what to do - not the higher level user environment. They can't just "change it back" without potentially breaking other things. May have been an early design decision with Mavericks...
I came across this article recently: http://appleinsider.com/articles/13/11/01/mavericks-uses-ambient-light-sensor-da ta-to-detect-movement-delay-sleep-mode My hypothesis: they rewrote the way the kernel interacts with the SMC to take advantage of more sensor data to better manage battery life. Because there are so many sensors now (Ambient Light, Accelerometer, battery charge, etc.) the code has probably gotten complex.
My temporary solution is to tape a stiff piece of plastic (cut down from a SIM card holder or old credit card) over the button so that I can "flap it up" and hit it if I need to, but otherwise it mostly stops me from accidentally hitting the button instead of Delete. Obviously, this is not a long-term solution but might be a good stop-gap for some people.
My 2013 MBA is by far the best computer I've owned - hardware and software - in the 35 years I've been buying computers. I understand this issue frustrates people but I've had FAR worse problems with other machines from other vendors and even DIY machines. So don't throw the baby out with the bathwater if you've bought a new Apple and you otherwise like the machine!
I hope people find this information useful. The more informed we all can be when we do talk to apple the greater the chance of an eventual fix.
