Last entry I talked about the evolution of “crowdthings” – such as crowdsourcing and crowdwisdom – which bring together the vast amounts of people connected by the Internet to achieve a complex task, or even just to carry out a simple task in an unforeseen way. Now we’re seeing the power of crowdsourcing in a crisis as the online public helps out in the search for missing Malaysia Airlines flight MH370.
How are they able to do that? The answer lies in a website called Tomnod, which allows its users to scour footage captured by satellite imaging company DigitalGlobe in a bid to locate anything that might be of interest to the search parties. Users helping with the search for flight MH370, which vanished without a trace four days ago, have the option of tagging what they think could be wreckage, life rafts, an oil slick or “anything interesting or suspicious” in 3,200 km² of satellite imagery, in which each pixel represents 50cm of space. The site gives visual examples of the items in question, in order to clue users in about what to look for.
You might be wondering, as I initially did, how on earth ordinary web users could possibly be qualified to carry out such a delicate task with such high stakes, in which a wrong lead could cost the search teams time and resources they cannot afford to waste. But the whole point of crowdsourcing is that it isn’t about handing over a task to a single person, qualified or unqualified; it’s about using the collective output of hundreds, even thousands of people to accomplish a task. Each of those people will have their individual strengths, weaknesses and margin of error, but they will be backed up and balanced out by hundreds and thousands of other contributions. Those contributions are analysed and the unreliable data is sifted out until only the most reliable answer or set of answers remain. In Tomnod’s case, it uses an algorithm called CrowdRank to determine which tags have the most consensus amongst users, and then submits those areas of the map to further, expert analysis. Finally, a top ten list of noteworthy locations will be shared with the authorities.
So by narrowing down the general area in which rescue teams need to conduct their search, crowdsourcing can save hours of time that would otherwise be wasted trawling empty ocean – in theory. The crowdsourced search is inevitably biased by the particular area of satellite imagery uploaded to Tomnod. Developments about the missing flight are coming in faster than DigitalGlobe can image the relevant area and upload it, and the site is already suffering from an overload of traffic, resulting in frequent downtime. Its Facebook and Twitter feeds are promising improved servers and an expanded search area to come very soon. Meanwhile, discussions have sprung up on sites like Reddit and again, Facebook, where searchers can help one another with the site and compare findings.
What kind of track record, then, does Tomnod have for crowdsourcing crises? In the aftermath of Typhoon Haiyan last year, the website put users to work tagging sites of destruction in order to help direct relief efforts, using “before” and “after” images to compare the region pre- and post-tycoon and identify damaged structures. Within hours of the images becoming available, Tomnod users had identified more than 10,000 points of interest covering over 100,000 square kilometres. DigitalGlobe’s senior management and research developer Luke Barrington believed it to be one of the first times volunteers from around the world had been able to contribute time and knowledge towards the response to a disaster of this scale.
Tomnod and DigitalGlobe have also crowdsourced hunts for missing aircraft before MH370. Over 14,000 taggers contributed to the search for a light aircraft which went missing en route from Oregon to Montana last December, and was eventually located nearly a month and a half later near an airport in Idaho. And only last month Tomnod assisted in efforts to find another aircraft which went missing in Arkansas, and which was found after eleven days of searching. It’s unclear whether or not the crowdsourcing actually influenced the success of the search efforts, since obsessing over those kinds of details would have seemed irrelevant in the wake of such tragic outcomes. Needless to point out, the disappearance of MH370 far surpasses the two incidents in scale and severity – necessitating a correspondingly huger search effort, but also promising far more serious consequences if the search efforts are misdirected.
By and large the global public seems to have faith in the potential of a crowdsourced search, but there is still scepticism about whether or not Tomnod is putting its resources to the best possible use in the way the search is being conducted. This looks to be the most severe crisis involving a vanished craft that Tomnod and DigitalGlobe have turned their technology to, and it’s already putting the concept of crowdsourced searching – as well as the site itself – to the test. MH370 could be a landmark incident in the history of crowdsourcing crises, ultimately proving whether or not the act of crowdsourcing a search has any merit, or else helping to improve upon the methods employed in crowdsourcing a search of this scale. Its results could mean the difference between life and death in crises other than just this one.