(ebooks) OWNI shop

Best-OfScience

The Collective Intelligence Revolution

Academics & researchers are harnessing our human love of games to crowdsource vast and complex tasks. And when the capacity of the human brain is combined with digital computing power, the possibilities appear truly staggering.

by Anaïs Richardin On May 16, 2012

79 Reactions
facebook share mail email A+ A-

Related posts

Gaming is an important part of society today, and many see the evolution towards homo ludens (playing man) as its major characteristic. Now researchers from around the world are organising the first international conference dedicated to “Harnessing collective intelligence with games“, to be held in September this year in Germany. The conference is aimed at deepening our understanding of harnessing large groups of participants to perform certain tasks through games, an emerging catalyst of collective intelligence.

On the agenda will be crowdsourcing, crowdsolving and human computation, among many other phenomena. They’re all variants on the same increasingly prevalent theme: the combination of human intelligence capabilities and digital computing power. Human beings still have a superior capacity for certain faculties such as aesthetic judgment, intuitive decision making or critical thinking. But unlike computers that require only a little electricity to operate, humans need to be constantly motivated in order to continue contributing. Motivation they could find in games and their mechanics.

For Markus Krause, a PhD student at the University of Bremen working on collective intelligence and its public uses, human and machine must work closely together.

The combination of human intellectual capabilities, and the power of computers to store and disseminate data, can be a very effective model for helping out humanity.

The term crowdsourcing (the use of many people to accomplish a task) was born in an article by Jeff Howe in Wired in 2006. While the word itself may have been new, the phenomenon it described had existed for a long time.

It was crowdsourcing that got Luis von Ahn, now an Associate Professor of Computer Science at Carnegie Mellon University, thinking about CAPTCHAs in the early 2000’s. If you’ve ever opened an email account or filled out a form online, then it’s likely that you know about CAPTCHAs. They are quick tests which allow systems to differentiate a human user from a computer, preventing robots from sending out automated responses, polls, phishing and any other malicious spam or data retrieval activity.

The test is based on the analytical capacity of the human brain. Two sets of letters are presented, often distorted to make their analysis possible for humans but difficult for robots. At a TED conference in 2011, Luis von Ahn explained how he had reflected on the possibility of using the time spent deciphering CAPTCHAs for a specific purpose, something that would be useful for everybody.

200 million CAPTCHAs are typed every day worldwide. With an average of ten seconds per CAPTCHA, that’s 555,000 hours per day. During those ten seconds, your brain does something extraordinary. It realises something that computers are incapable of realising. I then asked myself if we could do something useful for those ten seconds. There are problems that computers can not solve. But, somehow, we could divide that problem into segments of ten seconds, and every time someone types a CAPTCHA, it solves part of the problem. From now on, when you type a CAPTCHA, not only are you identifying yourself as a human, but you are also helping us to digitally scan books.

Thus was born reCAPTCHA, acquired by Google in 2009. Each time Google Books digitally scans a book, a certain portion of the text (usually around 20%) can’t be decrypted by the Optical Character Recognition (OCR) software Google uses. With the two sets of letters proposed in each reCAPTCHA, there is a CAPTCHA already verified by OCR which identifies you as a human, and some text from a book that you will decrypt. For Luis von Ahn, the software is a real success.

We averaged 100 million words digitised per day, which gives us 2.5 million books per day. Nearly 10% of the world’s population, some 750 million people, have helped to digitise books so far.

His ambition has led him far beyond the digitisation of books. In an interview with Wired, von Ahn revealed:

In fact, I would like to make mankind more efficient, by taking advantage of wasted time.

In 2008, he created Games with a Purpose (GWAP), a collection of short online games that are more than simple entertainment. By playing GWAPs, users compensate for the incompetence of computers in performing certain tasks. Each player is randomly assigned a partner, and the team of two has a limited time to earn as many points as possible. In the ESP game, for example, participants must add captions to images. The same image is presented to both players, who must suggest specific keywords. When both players offer the same caption, it’s recorded and the next image appears. The interaction between the players drives them to return to these simple games, and in the process they help decrypt and complete the web.

Many companies and public institutions have followed suit and opted to take advantage of the information processing capacity of human brains rather than simply relying on computers. NASA decided to appeal to people to help it manage the flow of information and data it receives, and to create a database. In the game ZOOniverse, users are encouraged to analyse the size and depth of craters on the Moon, allowing NASA to catalogue them. It’s an example of significant, free and effective sub-contracting of tasks where even the most powerful algorithms had failed.

But crowdsourcing does not stop with simple image analysis, explained Adrien Treuille, creator of Fold.it, at the Solve for X conference:

When we think of crowdsourcing, we think of the carrying out of simple tasks such as image and pattern recognition. But with a large group of people working with data and artificial intelligence, it is possible to solve problems which are found at the limit of human knowledge.

Fold.it is a game developed by US academics which helped discover the structure of an enzyme that plays a key role in the spread of HIV. Something which had confounded scientists for 10 years was resolved in just over 10 days by 40,000 people playing an online game. The results obtained by the players could constitute a major breakthrough in treatment of the virus. Each participant was credited alongside the scientists for this significant breakthrough.

This type of game is also allows academics to discover experts who they were unaware of, as Adrien Treuille explained.

Foldit allowed us to screen hundreds of thousands of Internet users and find people who had real expertise in carrying out certain tasks. I can imagine that in the future, when challenges present themselves to us, we can create games and puzzles based on the aptitudes required, and find people who are experts for these types of problems.

McGill University in Canada has created a game, Phylo, which makes use of the human capacity to arrange DNA sequences and thus participate in genetic research. The sequences are transformed into coloured geometric shapes which must be aligned; similar forms must be combined with each other while avoiding holes that represent mutations. For the designers of the project, the human/computer combination meant better results.

Solutions generated by users can be used to improve the quality of alignments calculated by traditional algorithms. These results do not show that man is better than machine, but rather that we have a symbiotic relationship between human and computer.

In these games which make use of collective intelligence, the vectors of engagement are numerous. Social peer recognition, reputation and human-computer interaction play an important role, as do rewards and gaming mechanics such as tagging, collaboration and missions.

Project Noah is an application that allows users to contribute to documenting their local wildlife. Missions encourage a community of citizens to photograph, tag, identify and learn about the flora and fauna in their area. All this aggregated data provides valuable assistance to researchers and has facilitated the establishment of a database that can be updated in real time and is accessible to all.

When Japan’s tourism board sought to improve its image and boost visitor numbers, devastated after the tsunami of 2011, they motivated participants with rewards.

The images of devastation that were appearing prominently in Google searches for Japan were doing nothing to ease the tourism problem. Students from the Berghs School of Communication in Sweden came up with an application, “Post from Japan“, to which tourists could upload their pictures of the country. The goal? To encourage the sharing of these souvenir photos on social networks by offering free connection time on government wi-fi. And for each like the photos receives, the user is offered an additional three minutes of connection. Ingenious when you take into account the price of a mobile internet connection abroad. Little by little, each day the application caused the images of the disaster to be pushed further down the Google Image results page.

Through gaming and extrinsic motivations, man is capable, without any specific expertise, to contribute to advances in many areas, whether scientific, social, environmental or others. In the future, new systems could even allow users to take advantage of an experience, while contributing massively to the improvement of the quality of web browsing. With Duolingo, for example, currently available in beta version, users can learn a foreign language, free and according to their level, while helping to translate web pages.


Image Credits: Elirook & Wi_2Photography CC (BY-SA)

Follow us onTwitter and on Facebook.