Recently, Inside Activity Tracking posted reports of inaccuracies in the Apple 5S’s accelerometer. The web is littered with them. While there are many theories that might account for the inaccuracies, we had a fascinating discussion with what we believe is a very reliable source that called into question whether this was a business-driven decision that was financially motivated, whether this was a motion sensor competency or incompetency issue or something else.
Realize that this is all speculation on my part, fueled by conversations with much smarter people than I. It is probably the purview of professional journalists to carry the ball forward on this one, as this is just one person’s opinion. And I’m a venture capitalist, not even remotely close to a professional journalist. But it is important, especially for someone like me trying to figure this out in the context of should I fund new health, wellness and medical applications that take advantage of M7 and depend on accuracy.
As we covered, and to recap, teardowns of units show the 5S shipping with a 6 bit accelerometer that has the same footprint as the Bosch identical sister part but is slightly cheaper on a per unit basis. Though these teardowns can certainly be wrong, if correct, the problem could be that this part is less accurate than the Bosch 10 bit part. What that means is that iPhone 5S with the M7 Motion Co-Processor shipped arguably with potentially significantly less accuracy than what was and is available on the market. Now, this isn’t exactly a complete answer, but it is could explain a lot of the inaccuracies being reported.
To add to that, no explanation for the inaccuracy could be complete without also pointing out the possibility that the iPhone 5s’s motion-sensing firmware could be at fault as well. Realize that this is not a simple task for Apple or any other software company. It is made even more challenging by the various form factors and processors it must support at the software/firmware level to continue to drive accurate readings across all phones. It wouldn’t be the first time that Apple executed poorly on the software side in a major release. Anyone that tried to set a route on Apple Maps that navigated them to the same address 2 counties away remembers that not so fondly!
As I’ve been intrigued by this issue for the last week, over the weekend I continued to talk to a few additional people seemingly in the know. And I’m left wondering why, if true, a company, that is so publicly driven by the perception of driving towards perfection, ships something that it certainly should know is less accurate when it comes to motion sensing, than it could have or should have been.
The obvious answer of course, could be that it was much cheaper to do so, and most applications in the activity tracking space have user bases that are used to inaccuracies. Given the piss poor state of accuracy, as we wrote about here, who really would know. The trade off is that they are at least “directionally” accurate, and enough to get people off their butts and barstools. But what’s the business value of that accuracy tradeoff? If we assume that Apple could save, say, $.40 per unit, over 100M units, that is $40M to the bottom line. Would Apple do such a thing? We don’t know, but given the trade off and usual “it will get better over time” attitude of most consumers of hardware and software, we are certainly wondering.
The other theories could also point to a lack of competence, that motion-sensing and accuracy are problems that Apple really doesn’t understand well enough to get that extra accuracy from software. And I’m sure there are a few others.
But, I’m also intrigued by the following and it is worth monitoring over time: The Bosch sister part, which is more expensive and available, is also pin compatible, meaning that it can easily be substituted in future units and potentially deliver more accuracy, all without anyone knowing that the part was changed out. Is this just a $40M play, or is there more to this? The conspiracy theorist in me, the one that worshipped at the alter of Gillian Anderson (Scully on X Files), is wondering.
I believe the net results are this: For consumers, the iPhone 5 or previous iPhones, for applications that are using the accelerometer, are often being found to be more accurate than the iPhone 5S using the 6 bit accelerometer, though of course the price for that accuracy is incomprehensible battery drain. For a venture capitalist that is looking closely at wearables, it presents strong opportunities to fund companies that can further draw distinctions between noise and signals, with other approaches to accuracy. But for Apple, for a company that just spent ridiculous amounts of money marketing their M7 motion co-processor, all of this is a bit baffling and troubling, though certainly M7, as we covered previously, has huge potential. Perhaps just not out of the gate.