What if you could instrument activities in your life, analyse them, and learn from them? What if you socialised that data, or produced analysis centres which found patterns? I spent 30 minutes thinking about this after my internet connection went down, and came up with a bunch of questions.
The Instrumented Self
Instrument your heart. Go grab a heart rate monitor and wear it for a while. I wear one at the gym, but I’d like one I can wear all the time. What could I tell about myself? I strongly suspect there’s a good correlation between raised heart rate (at least over a few days) and my happiness.
Instrument your well being. Okay, happiness is a little more difficult to measure – so how about stress levels. Well, measure the the cortisol in your saliva (spitting every few hours) and you have a reasonably good indicator of stress. And why is that useful? Well, go read Robert Sapolsky’s excellent Why Zebras Don’t Get Ulcers, or better yet, listen to a great lecture of his on the subject (#94 on iTunes here), and learn what stress does to you.
Instrument your calendar. Don’t just attend meetings. Measure the impact of those meetings. What’s the impact on your heart rate, or your stress? Wouldn’t that be interesting to know?
Instrument your walking. Okay, we all know how to do this with a nice little gadget, like the Fitbit say.
Instrument your bed. Your sleep, more specifically. Even an iPhone with Sleep Cycle can do that.
What else could we instrument?
- How much garbage is generated in your house every day
- How much petrol you use every month
- How many emails you send every day
- How many times you pick up your phone
- How many times you switch between windows on your computer
- Your credit card (see blippy)
- Your speech
- Tons of stuff of course – the list is endless.
There’s so much in our lives that can now be easily instrumented if we had the right tooling. But it strikes me that the instrumentation part is relatively easy. It’s easy to record our speech. It’s difficult to analyse it.
Ideally you want instrumentation to be easy and implicit. Using devices like a heart rate monitor, Fitbit and Sleep Cycle are mostly implicit. I don’t have to “do anything” for something to be measured.
The stress measurement is a little more difficult – I have no idea if someone can measure something like that a little less explicitly than hawking. It would be amazing if we could instrument eating. I suspect that’s a little difficult, but there already gadgets to instrument your faeces…
Change – Graphical, Daily, In Your Face
But we know we’re stressed – do we do anything about it? What if we knew it graphically, unambiguously? What if we instrumented ourselves and analysed that data? What if we got an email every day with interesting results? Would we act on that data?
I think some people will be encouraged to act – to change. Not everyone for sure – but if I knew that eating beans on Tuesday makes me feel good on Wednesday, well, I’ll eat more beans. Of course, people know that smoking is bad for them. What if the “bad” was measured? What if they saw a graph of continual degradation in lung capacity, distance walked and so on – and made “bad” more “real”?
What if that data was socialised? Perhaps between my close friends, or work colleagues? Would we act on the data then?
There’s a large psychological aspect to this – which would be fascinating. What’s the impact of all of this on behaviour, or how could we create a positive impact?
What if we had the instrumented-self-as-a-service? Imagine vast numbers of instrumented humans feeding their data to a central service that drew even more interesting results.
Perhaps there’s a collective increase in well-being near the end of the month after the pay check arrives. Or perhaps the local store shutting down means that everyone in my neighbourhood now walks a little further to the nearest store?
We’ll need a little more than simple sentiment analysis to do this kind of analysis – and it sounds to me like this is one of the greatest problems with this idea.
I think this is tremendously exciting and interesting – yet I can already hear the moaning naysayers. I had some of my DNA tested (using 23andMe) and tell people about it (usually quite excitedly) – yet I’m always surprised by the number of people who I meet who are totally and utterly aghast that I would do this.
After privacy, there major concern appears to be “what if I learned something bad”. They’re worried about learning something.
Side note: 23andMe learn things – they ask their users to answer small questionnaires. They’re learning more about correlations between genes and diseases & behaviour. Getting better the more people use them. There’s no reason something analysing our walking, eating, spending and defecating patterns couldn’t do the same.
How Are You?
We get asked this a lot. How are you? Well, what if I answered with a few graphs? What part of me are you really asking about? Sure, it’s usually phatic, but sometimes there’s real interest behind that question.
We lead these private lives (yet buy a tremendous amount on a credit card that is pretty well instrumented by financial institutions). Why is these data in their hands instead of ours? Why are large corporations changing the way shops sell based on our data – yet we don’t have access to it ourselves?
How are you? Imagine being able to answer this holistically. The health you – the mind you – the financial you – the …
The Quantified Self
I started thinking about this after knowledgeable friend Andy Hyde put me onto an interesting NPR podcast On the Media which had a pointer to a blog on The Quantified Self . “Self knowledge through numbers”. That’s a little cold – but that’s the heart of it.
Also check out this interesting TED talk by Deb Roy who analysed 90K hours of video to learn stuff about himself and his family. The techniques he used are powerful and implicit. (I had 4 tabs open to this stuff before my internet connection went down, so most of this has been written offline (a slow 3G) without referencing other material).
The Instrumented Self
Knowing more about my genotype has told me a little more about my phenotype. Sure, I am not just the result of my genes – there’s context too and more besides. That’s how I see the Instrumented Self. These data won’t tell us everything – but correlating multiple metrics will go a little way towards telling us something meaningful – and the more metrics we put in there, the more of us that we instrument – the more context and insight we could draw from this.