Games with a Purpose (GWAPS)

Mathieu Lafourcade, Alain Joubert, Nathalie Le Brun

• Publisher: Wiley-ISTE
• Year: 2015
• Print ISBN: 9781848218031
• eBook ISBN: 9781119136316
• Edition: 1
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

Human brains can be seen as knowledge processors in a distributed system. Each of them can achieve, conscious or not, a small part of a treatment too important to be done by one. These are also "hunter / gatherers" of knowledge.  Provided that the number of contributors is large enough, the results are usually better quality than if they were the result of the activity of a single person, even if it is a domain expert.  This type of activity is done via online games.

1
Biological Games

In this chapter, we present the main GWAPs currently available in biology and biochemistry.

In the highly specialized field of molecular biology, the results of FOLDIT¹ have been so positive (three articles published in the journal Nature – [COO 10, MAR 12] and [HAN 10] – and another in the Proceedings of the National Academy of Sciences, or PNAS [KHA 11]) and media coverage so important (a large number of articles in the non-specialized and popular science press) that the concept of GWAPs has quickly spread to biology. Many researchers try, with varying degrees of success, to use an entertaining-contributory approach to solve complex problems.

Zoran Popovic, one of the creators of the game, and director of the Center for Game Science at the University of Washington, believes that his laboratory is responsible for the emergence of a new path of scientific discovery. This path takes advantage of the creative capabilities of a wide population, once it is possible to involve it in research mechanisms through an effective initiation.

Adam Gazzaley, professor of neurology, physiology and psychiatry at the University of California and creator of NEURORACER (a game for improving the cognitive abilities of older people), being more restrained, thinks that the field of serious games is still in its infancy and that there should be a selection between that which claims to be based on science and what has actually been validated by scientific methods. According to him, approaches and tools should be multiplied and the results should be verified before making strong recommendations.

1.1. Foldit

Type of task – Macrotask.

Preliminary remark: FOLDIT is certainly the ancestor of biological GWAPs dedicated to the solution of complex problems. One of its designers, Adrien Treuille, has also participated in the creation of ETERNA² (see section 1.2), and the FOLDIT team is also at the origin of the Center for Game Science, an organization of the University of Washington dedicated to the universe of gaming for scientific purposes. The Center for Game Science hosted a very recent biological GWAP: NANOCRAFTER² (see section 1.3), a construction game of molecular nanostructures.

Launch – May 2008.

Audience/popularity – 57,000 players 2 years after launching.

Goal – Create predictive models of three-dimensional (3D) structures of proteins from their amino acid composition.

Significance of crowdsourcing for the problem to be solved – Predicting the 3D structure of a protein by its amino acid composition is a fundamental issue: a large number of diseases are caused by mutations, which by affecting the 3D structure of a protein, alter their normal operation. If the normal and the stable structure of a protein can be predicted, it can be understood how a mutation occurs at the level of the spatial conformation, and appropriate adapted therapies can be developed (vaccines, proteins for therapeutic use, etc.).

At the origin of FOLDIT, there is the distributed computation software program Rosetta@home: this is a program that determines the possible conformations that may adopt a protein by calculating the probability of interactions between segments of polypeptide chains according to the amino acids that compose them; interaction chains being all the more probable that the level of energy required for them to take place is minimal. Therefore, the most probable 3D structure is the one that will require the least amount of energy to develop. However, for a given amino acid sequence, there exist so many possible spatial conformations that simulation with algorithms is an extremely long and complex process. Rosetta@home uses home personal computers and runs when they are hardly used or not, thus increasing the computing power available to the project known as distributed. FOLDIT has been designed as an interactive and recreational extension of Rosetta@home.

According to David Baker [HAN 10], one of the researchers at the origin of the project, FOLDIT players are not required to have simple skills of visual recognition, of classification; they are asked to utilize their logic, intuition, ability to see in 3D, initiative and creativity in solving complex optimization problems. Baker says to use three skills of the human brain, three typically and exclusively human talents: of classification; they are asked to utilize their logic, intuition, ability to see in 3D, initiative and creativity in solving complex optimization problems. Baker says to use three skills of the human brain, three typically and exclusively human talents:

1. – the ability to know how to take a risk in the short term for a long-term advantage and vice versa;
2. – the ability to see in 3D;
3. – the ability to realize that a dead end has been found and to stop just in time (that is before too much time has been wasted).

Zoran Popovic, one of the main stakeholders of the project (of its software/computing aspect), claims that for the time being, the human brain remains better than the machine as soon as the creativity parameter becomes essential and unavoidable. In addition, the game dimension creates the ideal conditions for the emergence of a collective expertise: the designers of FOLDIT have observed that not only do the players improve over time, but also that they organize themselves to add their respective competences. Finally, the game is designed in such a way that the players’ findings are reinjected, recycled in order to be integrated in the manipulation mechanisms of the virtual protein, such that new players become increasingly faster experts.

Thus, a typical case (illustrating the first of the three abilities mentioned above) where humans are performing better than machines is that where a structure must be profoundly reshaped to optimize its configuration, that is to reach a stable form with a minimal energy cost. This implies too many sacrifices in terms of computation time for the machine, since this requires, as a first step, disorganizing an existing layout.

The most passionate gamers showed such skill in manipulating the protein chains, that designers have rapidly updated the game so as to provide the ability to create ex-nihilo proteins (potential applications: finding molecules likely to catalyze certain metabolic functions, manufacturing vaccine candidates, designing molecules that meet specific biological constraints for biotechnological or medical applications, etc.).

Moreover, David Baker uses the players’ findings, that is to say specific folding strategies, to automate them in order to improve the prediction algorithms of protein structures. The community of players is regarded as a scientific partner on its own, and is associated as a coauthor in scientific publications [KHA 11].

Qualities/required human skills – Intuition, logic, spatial vision (“humans have 3D understanding that computers just cannot handle yet, and computers have the number crunching capabilities tha...

Table of contents

• Cover
• Table of Contents
• Title
• Copyright
• List of Figures
• Introduction
• 1: Biological Games
• 2: Games with a Medical Purpose
• 3: GWAPs for Natural Language Processing
• 4: Unclassifiable GWAPs
• 5: The JEUXDEMOTS Project – GWAPs and Words
• Conclusion
• Bibliography
• Index
• End User License Agreement