Anyone who works with data knows what a messy situation it always is. All data - at least all useful data - has issues, ranging from how it’s recorded to how it should be processed. This is particularly the case with systems that rely on some human input. Self-reporting is notoriously hopeless, with research suggesting, for example, that the self reporting of ‘calories eaten’ is under reported by around 30 percent. Even using a device like my Fitbit Flex, you have to remember to switch it into sleep mode for it to automatically measure sleep (yes, I know you can manually add it later, but who does?). Similarly my Nike+ GPS watch can take anything up to a minute to lock onto the satellites it requires. Not to mention, randomly resetting itself to 2010 if it runs out of power, and the five or so times it never got a satellite signal during a 30 minute run. Margin of error These are slight errors of self and quantification, of course. More worrying - given we appear to be ready to hand over aspect of personal medicine to such devices (c.f. Apple’s HealthKit) - is the basic accuracy of the devices we’re using. A recent example. I went for a run using a Beurer fitness watch with heart monitor, and my Nike+ GPS watch. Both were started at the same time, but at the end of the run, the Nike+ watch listed 403 calories burnt, while the Beurer said 320. Assuming the Beurer+heart rate monitor is the more accurate of the two, that’s a 26 percent difference. Combining my two examples then, I could be over-exaggerating my calories burnt by 26 percent and under-exaggerating my calories in by 30 percent. That’s massive inaccuracy. Life and death? Yet, again, it could be said this is a trivial case. Where it does get concerning is when devices say they can accurately measure proper medical metrics such as cholesterol or blood sugar, or for some people even heart rate and calories (in and out). Not to mention when doctors are relying on people using the devices properly,: and that could be whether legitimate or illegitimate misuse c.f. the statistics about how many people complete their set treatment of antibiotics (22 percent of 16-24 year olds) But maybe I’m being too pessimistic. The Quantified Self can clearly produce great results, even with the current generation of devices. In most of these cases, though, it works because the motivation behind the Quantified Self is the self. Initial self awareness and self motivation enables quantification to positively feedback into a better self. Where we need to be concerned is in the cases of those where we expect quantification alone to drive self and change behaviour.