In the year between March 2011 and 2012, a gang of hackers collected and published more precise and useful data about radiation contamination in Japan than any other organization, including TEPCO and the Japanese government combined. And we did it almost entirely with devices we soldered, bolted, and duct taped together at the Tokyo Hackerspace. We call ourselves Safecast.
Let’s back up for a second, to the days following the Tohoku earthquake and result-ing tsunami, when it was becoming clear that things were going … well, less than awesomely at the Fukushima Daiichi Nuclear Plant. Everyone was looking for details about what was actually happening, or what had actually happened, or what might happen. And they weren’t having much luck. I know because I was one of those people, and I spent countless hours trying to find any useful information, to no avail. I was asking friends, friends were asking me, and before too long, we had a whole group of people asking the same questions and trying to find answers. Finally it became clear that the only way we were going to get any useful data was to go out and collect it ourselves.
So what’s the best way for a handful of people to collect data about nuclear fallout around Japan? Huddled in an office in Tokyo, we wrestled with the question, and then Ray Ozzie (former chief technical officer at Microsoft) delivered the answer at a brain-storming session with other Safecast members: strap a Geiger counter to a car and go for a drive, obviously.
Our first shot at mobile radiation logging was as hands-on and basic as you could possibly get. A minimum viable product if there ever was one. Safecast volunteer Dave Kell had the (mis)fortune of doing the very first measurement drive with a system we’d rigged up for him. Actually, the word “system” is a bit generous — we gave him a Geiger counter and an iPhone with a preinstalled Flickr account and instructions to “drive through Fukushima and every once in a while stop and take a photo of the reading and upload the image to Flickr.”
The photo would be geotagged to give us a location, and the reading shown on the display in the photo provided the data. These of course had to be entered by hand. We also had to note variables such as whether the reading was taken inside the car or out, in the air or on the ground, etc. Despite all this, it actually worked and gave us our first real data set to analyze.
But if you’re thinking the system could use some refinement, you’re right. A few others used this method for some drives in the following days, but we knew it needed an iteration, and quick.
The first major improvement came from Kei Uehara and some of his students from Keio University. Like all first major improvements, this was enabled with duct tape. The “upload-a-geotagged-image-to-Flickr-and-input-data-by-hand” method was still in effect, but variables were reduced by duct taping the Geiger counter to the window of the car and taking a photo every five minutes, on schedule. This gave us a more consistent measurement and a clearer route, but was still a pain in the ass. And a pain in the ass drives innovation as well as anything — sometimes better.
While those “by hand” drives were happening, a core group at Tokyo Hackerspace had busted out the soldering irons to work out a more automated solution. Robin Scheibler, Steve Christie, Mauricio Cordero, Akiba, and Pieter Franken spent most waking hours of a solid week drilling, bolting, inhaling solder fumes, and staring at lines of code to give birth to the bGeigie. The name was chosen because the little Pelican case everything is crammed into looks like a bento box — a traditional Japanese lunchbox. This would become a line of devices that are the epicenter of Safecast data collection. This first bGeigie Classic (as we now refer to it) was a Geiger counter (we used the Inspector Alert by Medcom which has a 2″ pancake sensor) inside a modified Pelican case (modified to let radiation in, but keep weather out) along with an Arduino Uno and a GPS module at-tached to the outside of a car. A line out from the case ran inside the car to a netbook which logged a data point every five seconds, with the added benefit of informing the people inside what the radiation levels were outside the car. Now this was some serious progress!
The next two devices were variations on this Classic design, but used a Freakduino instead of an Arduino because we thought the onboard wireless might be useful and because Akiba had a bunch of those lying around. One of these was hooked up to a MacBook Air (making it the most expensive bGeigie version to date) and the other to a Dell laptop. The Dell version is now permanently installed in the Safecast car in Japan. While this was a major step forward from the “by hand” method, it was still rather bulky and cumbersome. That was the next issue we tried to solve.
The fourth device we made, which we called the bGeigie Mini, has become the base model for all future devices. It had the advantage of scrapping the laptop altogether. Akiba design-ed an SD card shield for his Freakduino boards, and Robin wrote the software to power it. With the addition of two AA batteries for power, this became the first self-contained version of the bGeigie. No external power, no cables, no laptops. This was a giant leap for us, but not even a step forward for mankind. We think of this one as something of a working prototype, since we changed a few things right away.
The fifth device swapped out the two AA batteries for two C batteries and modified the Freakduino to run on 3 volts. This was finally a very solid and repeatable design, and the next 20 or so devices we made were modeled on this one. We made minor revisions along the way, trying a different kind of foam to hold the Geiger counter in place, different positioning of the GPS unit, arguing over the merits of using a toggle or rocker switch for the power, different color cases — but essentially the guts of the device stayed the same. The onboard SD card logged the data, but still required someone to manually pull the card out and get the log file, then upload it to the Safecast server. Until early 2012, the vast majority of the Safecast data was collected with one of these models.
Somewhere along the way we realized two things: 1) While having a laptop in the car was annoying from a logistical and power supply standpoint, it had been useful for the team in the car to have the information it displayed; and 2) The Freakduino boards we were using had built-in wireless connectivity we could be using. This led Robin to build a companion device we lovingly refer to as the bGeigie Ninja. It’s a small wireless display monitor that sits inside the car and networks with the bGeigie outside of the car, to give immediate feedback on the outside readings.
We rocked this setup for months before the next major revision. In November of 2011, Joe Moross debuted the bGeigie Pro, which was immediately renamed the bGeigie Plus for reasons I’m forgetting at the moment.
Its awesomeness came from being a truly self-contained device that never needed to be opened, and in fact had a sealed case to keep out water and volunteers’ fingers. The added USB port powered rechargeable batteries and made data output a snap. After considerable deliberation, the signature duct tape window was removed (much to Joe’s delight and Pieter’s disapproval).
With an inventory of bGiegies, combined with an ever-growing roster of people jumping in to lend a hand, as of March 2012 we’ve collected and published close to 3 million data points. Compare that against the few thousand data points provided by official sources, who I’m going to assume have more resources at their disposal than some guys at a hackerspace.
I really love this story and am so psyched to have played a part in it, because in many ways it’s the quintessential example of hacker/punk motivation and DIY execution. Permission wasn’t asked. Rules weren’t considered. A need was determined and immediately acted on. No one cared that the devices weren’t perfect on day one, and they didn’t care that they’d be revised on day two. There are any number of schools of thought that would have dictated consulting local authorities, beta testing hardware, and refining and fine-tuning everything before ever heading out into the world. There’s certainly a place for that way of thinking, but it wouldn’t have produced anything close to the results we got in such a short time frame, and getting data fast was the problem that needed to be solved.
While year one of Safecast clearly was short on sleep, this isn’t the end. There’s more data to collect, and with the combo of our open designs and committed hardware partners, we hope to be able to reach corners of the Earth we haven’t even considered yet.
If this sounds interesting, please check out safecast.org. We’d love to have you get involved too!
From Safecast’s FAQ page, based on the 2 million radiation readings collected in Japan and around the world.
Question: Is it safe for me to visit Japan?
Short answer: Most likely.
Long answer: Depending on where in Japan you are planning to visit, radiation levels are no higher and sometimes lower than any in other major city around the world.
Question: Did the contamination from Fukushima Daiichi spread outside of Japan?
Short answer: Yes.
Long answer: Traces of particles released by the event at Fukushima Daiichi have been found in every corner of the Earth where they have been looked for.
Question: Are the radiation levels changing?
Short answer: No.
Long answer: While occasionally news of a “new hotspot” is being reported in the media, the reality is that these hotspots have existed (at least) since the 3/11 event, but just haven’t been measured.
Question: Is Fukushima Daiichi stable?
Short answer: No.
Long answer: All reports suggest that efforts to contain the contamination from the plant are ongoing, however access to the 20K exclusion zone around the plant is restricted and very difficult for anyone to research, so independent confirmation of published reports is difficult to obtain.
Safecast’s Data Is for Everyone
At its core, Safecast is all about the data, which is why we’ve chosen to publish every bit of data we collect under a CCO designation (creativecommons.org/publicdomain/zero/1.0).
CCO is similar to public domain and waives all rights so there’s completely unrestricted access and open use of the data allowed to anyone. Data is only as valuable as its usage, and we didn’t want any limitations on who could use it or how it was used. We don’t even require people to attribute anything to us.
This allows our data to be included in other studies, analysis, services, and large data aggregation projects that would fail if they had to document the hundreds of thousands of sources for each and every data point. Attribution can sometimes be more work than the study itself, and we didn’t want that to stand in the way of anyone doing good things with our data. We already know of scientists contrasting our data against a study of stress levels in certain areas,
medical professionals looking for similarities with their own symptom maps, and mega-trend research which may help us better understand what the future holds. We think the more data humanity as a whole has access to, the better off we’ll all be.
A caption in the print version of this article incorrectly identified Dr. Akira Sugiyama’s affiliation; he’s with Tokyo University.