The Scientist in Us All: How crowdsourcing in science is changing the world

Let us imagine John, 30 years old and a lawyer in London, at home after work. Before he goes to bed, he switches on his laptop, and plays a simple puzzle game for leisure. John just helped to nudge humanity a tiny bit closer to solving another scientific puzzle that could cure malaria better.

If John were a real person, he would be just like one of the hundreds of thousands of players in the many crowdsourcing-science online games today. Together, these laymen online gamers are battling alongside scientists against some of the newest and most difficult scientific puzzles. In the past dozen years, crowdsourcing and “gamification” of scientific conundrums have become a notable part of scientific research.

The study of science has often been pictured as an abstruse discipline with complicated experiments and concepts. The system of scientific research was honed over centuries to maintain its rigor and efficiency. Successful scientific studies require astute insight, based on intuition and deduction built from experience and knowledge. However, in the past decades, the puzzles in science have grown dauntingly difficult; the more we are able to see and understand, the more complex the problems become. For example, the variety of proteins in humans is on the scale of millions, and to decipher all the structures and possible interactions would be extremely difficult. No matter how intelligent and hardworking the scientists are, even with our best computers and algorithms, some of these questions cannot be satisfactorily solved on a realistic timescale based on the resources available. This is particularly discouraging when these questions directly link to the life or death of people. How do we solve them?

This was the question in the minds of the scientists at the University of Washington in 2008, when they worked with Seth Cooper, Adrien Treuille, and other computer scientists, to develop a new computer game for predicting protein structures. The idea of crowdsourcing-science games, or citizen science games, is to tap into the immense amount of time and energy invested by people into computer games, and use their efforts for serious purposes. Video games, much like science, are about overcoming puzzles and obstacles to reach specific targets, in the most optimal way possible, while under the constraints of certain rules and restrictions. Successful gaming takes trials, experience, and intuitive thinking. These are, in fact, some of the same skills needed for solving scientific problems, too.

This was the beginning of Foldit, one of the first crowdsourcing-science games to gain worldwide attention. Proteins, one of the most important groups of molecules in all living forms, fold up into specific 3-D conformations in order to carry out their functions. These conformations are usually the most stable shapes based on the chemical structures of the molecules, much like how machines need to be precisely assembled according to the shape of each piece. However, some of these protein conformations are difficult to predict or measure.

In Foldit, players are required to turn, twist, bend, and move around pieces of proteins in 3-D, to create the most stable folding shape possible. The process is like solving a 3-D jigsaw puzzle, or folding an origami, or even playing with Lego. Players are ranked by the scores calculated from the stability of their protein designs. While computers are good at doing massive calculations and repetitive tasks, other tasks such as visual thinking and intuitive insight can be difficult for computers. Players in Foldit solve the structures of and interactions between real, named proteins, with the help of real scientific data, such as electron density maps, evolutionary sequence data, known parts of the structure etc. Their solutions, as well as the ways they reached their solutions, are used by labs to ultimately solve these structures as well as improve protein modelling algorithms.

Fig 1. Screenshot from Foldit. Image courtesy of Wikimedia.
Fig 1. Screenshot from Foldit. Image courtesy of Wikimedia.

The results of Foldit help solve the mechanism of key proteins, find new pathways, or even design new drugs for diseases. Players discuss in the online forum, compare their solutions and strategies, and collaborate in teams to compete. Players are also able to write down pre-set sequences of actions that they find most effective, and apply them on different puzzles. These programs can sometimes be integrated into actual modelling systems.

Very soon after the start of Foldit, Foldit players were generating algorithms that defeat every existing protein modelling algorithm. Once, Foldit players were able to help scientists solve the structure of M-PMV retroviral protease, a crucial protein in HIV, within three weeks; the same puzzle stumped scientists with traditional approaches for more than a decade. Over the past eight years, Foldit results have led to six publications in journals, including top-notch ones like Nature and PNAS, on both specific notable results and the overall methodology.

In the last decade, the world of citizen science computer games has expanded dramatically to let people solve other science problems, too. We’ve seen the rise of Phylo (2010; on aligning sequences to solve evolutionary relationships), Nanocrafter (2014; on designing DNA nano-machines), EyeWire (2012; on solving neuron connections in the retina), and dozens of games in Zooniverse by the Citizen Science Alliance, covering fields from astronomy to evolution. All these have contributed to various research that has led to many top journal publications and repeatedly swept across world news media with their incredible results.

EteRNA, released in 2011, is another game that garnered the attention of scientists and the public alike. A joint-project between Stanford, Carnegie Mellon, and other scientists and developers including Rhiju Das and Adrien Treuille, EteRNA is an online game that aims to solve and predict RNA folding structures. RNAs are a massive and centerpiece family of molecules that carry out many biological functions. RNAs also fold up based on their particular sequences of “A”s, “U”s, “G”s, and “C”s, in order to take on specific shapes. Again, existing algorithms then were far from satisfactory in predicting folding shapes from sequence, especially when the molecules were larger.

Fig 2. Screenshot from EteRNA. Image by Peter Wang.

In EteRNA, players change and design sequences to fold molecules into target shapes in the most stable way possible. Players, like in Foldit, cross-check and compare each other’s results, and vote for the best ones. These sequence designs are then sent to scientists, who would actually synthesize these designed molecules in the lab. Real experimental data are then given back to the players. Throughout the process, there is active and direct engagement between the scientists in their labs and the players around the world in their homes and offices. Players and scientists discuss regularly in the forum, and even hold conferences to evaluate results.

According to Treuille in an interview by Nova ScienceNOW, the first attempts of EteRNA designs came out disastrously, making the developers themselves doubt about EteRNA. However, the surprising enthusiasm and determination of players pushed the game forward. Players worked together to learn from the experimental data. Within six months, players became even better than the best computer algorithms available. In merely three years after its release, EteRNA had gathered over 150,000 players, with 2 million total played hours and 13,000 synthesized designs.

The effort by the players is extremely valuable, not only because it generated an unparalleled amount of data, but also because it gave rise to many rules and strategies never known before. Just like in normal gaming, players in EteRNA come up with unwritten “rules” and special strategies to maximize success in the final lab results, such as the direction of GC-pairs, symmetry of sequences, number of consecutive repeated nucleotides, GC-pair necks, etc. (These are documented and managed by players in the Wikia and a curated list of public Google Docs.) Some of these may seem almost arbitrary or don’t seem to make sense, but work well in the lab; these often then lead to new discoveries of the subtle molecular interactions behind folding. Such “higher level” rules are characteristically human, due to the creativity involved. They are impossible for scientists to devise without extensive testing, and certainly cannot be automated by computers.

The game eventually reached the point where the questions, too, came from the players. We all know how we sometimes wonder about the best ways to play a video game. “Should I invest more points in defense or healing potions?” “What if I try to beat this game using nothing but a shield?” “Why is it better to reach this checkpoint first?” Some of these “what if,” “how,” and “why” questions are the same kinds of questions asked in science. Based on previous results and literature, EteRNA players started raising questions, some of which carrying direct biological significance without the players even realizing it. These include questions about the interactions between neighboring structures, effects of pseudo-knots and kissing loops (two special types of tertiary structures), or competition between two similarly favorable possible folds.

To answer these questions, the player community would then design custom puzzles for fellow players to solve. Scientists sometimes pick from these player-made puzzles to develop into investigations in the lab, giving rise to new hypotheses and findings. Two years ago, as part of this effort, I designed a series of puzzles called “Reverse EteRNA,” which prompts players to place mutations that cause structures to unfold instead. It was part of a fascinating line of questions about strategies to block or prevent unwanted structures.

In little more than five years, EteRNA has contributed to four notable journal papers, starting with one in PNAS. Two of these listed EteRNA players as co-authors. The players even had the experience of writing and publishing a journal paper themselves. Earlier in February 2016, the first ever research paper with non-expert citizens as dominant and co-leading authors was published in the Journal of Molecular Biology. Non-experts, presented with an approachable and comprehensible puzzle game, had eventually learned enough science to write a complete journal article to be peer-reviewed and published.

The impact of these games lies not only in how they accelerate research in an unprecedented manner, but also in how they introduce non-professional citizens to the work of science in a controlled, interactive, approachable, and provoking manner. Through such crowdsourcing laboratories, citizens are introduced to and involved in more than the answers. They are able to learn what questions are asked, why they are asked, how they can be answered, and how we can infer the conclusions from data. They are able to understand and apply the concepts of uncertainty, incremental problem solving, data interpretation etc., which used to be difficult to explain to laymen. These games have simultaneously become a massive educational tool for young students and adults alike.

This new approach of experiments, termed “Massive Open Laboratory”, holds immense potential to transform and further catalyze future research endeavors. This does not mean that we should break down the framework of research that humankind honed over centuries of work. However, if we can involve the greater human population in these efforts, we can tap into a great pool of human intelligence to solve some of the most important questions we have today.

Peter Wang is a sophomore in Timothy Dwight College. Contact him at

Feature Image by Peter Wang, using EteRNA.


Sources, Links, and Extra Reading:

You, too, can help to solve cutting edge scientific puzzles. Websites:

Crowdsourcing science computer games in the news:

Notable Foldit publications

Notable EteRNA publications

TED talks and documentaries: