Crowdsourcing the Search for Malaysia Airlines flight 370

MALAYSIA AIRLINES flight 370 disappeared in mysterious circumstances. Malaysian authorities are presiding over a spiralling public relations disaster. The tragic loss of passengers and crew is exacerbated by seemingly inaccurate and untimely information. As the search widens, could Crowdsourcing using the internet help solve the mystery of the disappearance of flight 370?

Crowdsourcing the search for MH370

Crowdsourcing the search for Malaysia Airlines flight 370, Tomnod and DigitalGlobe

Crowdsourcing – as old as Civilisation Itself

Crowdsourcing is using the wisdom, experience and enthusiasm of a crowd to solve problems. It is as old as civilisation itself. In the digital world, Wikipedia is Crowdsourcing, the internet itself is Crowdsourcing – information provided and shaped by the crowd at unprecedented scale.

The potential of “the Wisdom of Crowds” has long been recognised by problem solvers, product designers, businesses and governments. In the search for Malaysia Airlines flight 370, an ingenious solution from Tomnod (a subsidiary of DigitalGlobe) is attracting significant attention.

Crowdsourcing (MIT Press Essential Knowledge)

Tomnod, Crowdsourcing the Search for Malaysia Airlines Flight MH370

DigitalGlobe has positioned satellites over the primary search areas. It provides map data via the Tomnod website for volunteers to analyse. Tomnod subdivides the satellite image data into sets of ‘map tiles’ and displays them to volunteers to search for signs of wreckage, life rafts or oil slicks. Millions of volunteers are using Tomnod and DigitalGlobe’s technologies to search for the missing flight.

When a volunteer identifies an area warranting investigation, they mark the map tile. Each map tile is studied by successive volunteers and if others mark the same areas, expert resources use more detailed satellite images to confirm or rule out any ‘sighting’.

malaysian airlines flight 370 - crowdsourcing the search

The search for flight 370

Challenges in Crowdsourcing this Type of Problem

The main challenges are the experience of the crowd, the quality of the images provided to them and the scalability of the web platform. As millions participate, it becomes hard to provide high-resolution images to huge numbers of participants. Bandwidth and server performance drop and outages occur.

tomnod-crowdsourcing-scalability

Technical issue on the Tomnod site (18th March 2014)

The experience and ability of the crowd matters. As flight MH370 has yet to be found, ‘all’ items marked by Tomnod volunteers are ‘false positives’. Looking at more detailed image data, the false positives come from volunteers identifying (what turns out to be) commercial shipping.

We talk about ‘searching for a needle in a haystack’. I think of the Tomnod Crowdsourcing approach as looking for the haystacks in which to look for the needles. Volunteers with low-resolution satellite images may find haystacks. Specialists with high-resolution images can look within those haystacks.

A Guide to Open Innovation and Crowdsourcing: Advice from Leading Experts in the Field

Is Crowdsourcing Better for Approximation?

The search for flight MH370 is binary, it is either found or it is not. Tomnod used its technology to good effect in Somalia when they mobilised a large volunteer network to search images for evidence of displaced populations. They tasked volunteers to look for temporary dwellings, which would indicate displacement of refugees fleeing conflict. The project was successful, but the problem itself was tolerant of ‘approximation error’.

The Somalia problem was one of ‘size and scale’, and Crowdsourcing the estimation of ‘size and scale’ is relatively simple. The search for Malaysia Airlines flight 370 is harder, as the outcome is not an approximation and therefore has no tolerance of error.

Keeping the Crowd Engaged and not Enraged

The size of Tomnod’s volunteer network is impressive, but keeping 3 million volunteers engaged is challenging. I mentioned in the opening that Crowdsourcing is as ‘old as civilisation itself’, but we have only started to ‘scratch the surface’ in terms of understanding how to best use (and motivate) massive crowds in online problem solving.

Gamification shows enormous potential. Recent projects from Cancer Research UK, such as the ‘play to cure’ games are particularly interesting.

Crowdsourcing can also be dangerous if not adequately monitored and moderated. Crowds can turn into mobs and a mob with poor information, prejudice and social media can quickly become hysterical and uncontrollable. The search for the Boston Marathon bombers gave us insight into the potential dark side of Crowdsourcing.

It is uncertain whether the Tomnod project will help locate flight MH370. What is clear is that a massive number of volunteers want to help. Crowdsourcing on a massive scale has high potential but arguably needs more sophisticated models and tools.

As anguish for relatives and friends continues, it can only be hoped that the mystery of flight MH370 is resolved quickly. The ‘Power of the Crowd’ might prove to be a critical factor.

Further Reading on Crowdsourcing and Collaboration