|This weekend isn't getting any cheaper
||[May. 28th, 2011|11:45 pm]
One day and half-way through the first Quantified Self conference, my mind is afire with thoughts of EEG machines.
After yet another delayed and circuitous flight that arrived over an hour late, and a night in a borrowed bed, the day began early with a blood draw for a free Scanadu analysis service before the conference proper started. I've spoken at QS meetups before (old face!), but this is the first time that something this big has happened. The program was a little nebulous until the last minute, but the results are solid and much better than I'd dared hope.
An early session concerned sleep tracking. Thanks to 2muchexposition's wise melatonin suggestion I'm sleeping better than I have in months, if not years, and, between that and modafinil, feel largely immune to jetlag. But I do wish I could prove those results, quantify them objectively, or monitor them for regression.
The Zeo is a wireless headset paired with an overgrown clock radio that between them have enough EEG cleverness to reliably distinguish between different sleep states. More importantly, the data from it, by default, gets shipped up to the Zeo service which provides a bunch of value-adds.
Christine Peterson had a poster up demonstrating some of the basic stats that you can correlate against sleep performance, which quantify just how bad caffeine and alcohol were for her (reduced quality sleep by something like 30 and 60 minutes, from memory).
Zeo also provide a deidentified aggregate dataset to researchers, which is the largest single sleep history database by two orders of magnitude. That's kinda great.
Some of the observations made by the Zeo staff suggest interesting experiments - for example, most of the deep sleep happens in the first third of the night. So despite the fact that most people seem to have a natural sleep duration that they require, it could be interesting to try to reduce that by trimming the time spent asleep after you've extracted all the early benefits. They only knew of one or two people who were trying such things, but it would require being able to measure the impact to waking productivity and performance.
That segues nicely into the second session, which was on the subject of attention and focus. I have a huge variance in productivity day to day, and always have. There are times when I focus monomaniacally on a task and make great strides in hours that seem to pass like days. Other times days will pass when I can't stay concentrated for more than a few minutes at a time and feel distaste towards starting on any meaningful projects. I have no idea why, and, until I started experimenting with Adderall, had no luck in forcing the productive mode. Even now, the drugs aren't particularly reliable.
Neurosky sell EEG headsets that are designed to distinguish between waking brain states, and one company rep spoke about some recent experiments that showed success in distinguishing between degrees of focused and distracted states. There's other approaches to this that are potentially interesting, like logging the title of the active window on the computer every few seconds, or monitoring idle time on mouse and keyboard. These, however, suffer from the dual flaws of being unreliable proxies for mental state and not involving high-tech wireless toys.
It'd probably be more useful to measure high-level task productivity, but that seems like a much harder problem with a great many confounding factors. I'd gladly settle for being able to more reliably force a state of flow; something that I find elusive all too often. Establishing a solid baseline of work focus prevalence would be a good start to this.
(There was an interesting side discussion about what one should generally chase as a result in personal experimentation: whether to seek the dramatic change or the continuous slow improvement. My feelings are that dramatic improvements - much like my melatonin epiphany - are a better goal. In theory, you don't need instruments capable of subtle measurements to observe these results; they're blindingly obvious. But between concern over tracking regressions and fallibility of self-assessment, having some objective metrics does have clear benefits. Particularly when I've already tried a lot of things to drive productivity, with frustrating inconsistency in results.)
These probably sound like the sessions were commercial product pitches; that really isn't the case. There were a lot of people talking about their experiences with various approaches, some of which involved commercial products. My personal biases bubbled those two up to the top. It does, however, seem a little rash to go directly from owning zero EEG units to two.
The Zeo appears to be the better integrated unit end to end. It has well defined goals - improve your sleep - and comes with tools, workflow and community that's all aimed at the same goal. The Neurosky, on the other hand, seems to still be groping for purpose. They're pitched as components to be used in various brain-related enterprises, and come with only trivial games and some product partnerships. Anything I'd want to do with them would likely have to be built from the ground up.
On the other hand, my sleep is actually pretty good now. Improving my daytime productivity is definitely the higher priority, and the first step to improving it is to be able to measure it. But I'm also relying on the Neurosky being capable of delivering the results, and need to buttonhole the rep to find out how to replicate the attention experiments that she described. More when I learn it.
One surprise booth was GreenGoose, who have built tiny, disposable, wireless accelerometers. They have a small-puffy-sticker form-factor and three years of life off their built-in 100mAh battery, which boggles my mind. The general idea is to stick them on... well, everything, really (a frisbee was one demo) to automatically build acceleration timeseries, then ..., and then - profit! They're totally cute.
And then another surprise was a gent carrying around a miniature makerbot-y looking box (it's the laser-cut-wood look) that was an OpenPCR thermocycler! For $512! I so regret missing the kickstarter for this.
But I digress (as usual).
In general, one of the critical observations that came out of today's panels was that there's a high variability in people's responses to any particular approach. The effects of pharmaceuticals aren't currently predictable (I get no concentration boost from Modafinil, unlike everyone else I know), nor those of dietary, behavioural or lifestyle changes. The only real way to improve is to measure your own response to these experiments - which is, of course, the heart of this entire movement. Maybe with more data we'll glean the factors which govern the responses and come full circle to being able to reliably predict effective treatments. But for now, the basic principle remains suck it and see.
I'm now at Hacker Dojo. My initial impression is that, compared to Noisebridge, the plant is better but the people are worse (harsh, but pithy). I've begged a couch for the night in South Bay (well on track to my expected result of spending my five nights in the bay area in five different beds; though I only manage four different cities) and tomorrow's conference program looks equally fun.