EVE Community surpasses expectations with Project Discovery

Paramemetic 2018-01-27

EVE Online’s active and engaged community once again rose to and exceeded expectations with the ongoing Project Discovery. In a recent Dev-Blog, CCP’s Team Psycho-Sisters provided an update on the project, as the first batch of data is now undergoing analysis. In a mere 190 days, 77,709 participants from EVE’s community provided an astonishing 44.4 million classifications. As a result of their hard work, many new candidates for potential transits were identified with a high consensus. These new candidates are now being compared to other astronomical databases by researchers at the University of Geneva.

Project Discovery

Launched in July of last year, Project Discovery puts players on the front lines of astronomical research. Using in-client tools and encouraged by in-game rewards, players are first trained using mini-games developed by CCP’s Team Psycho-Sisters, in collaboration with a research team at University of Geneva using the Massively Multiplayer Online Science API. Players look at data relating to light curves for periodic variations that sometimes indicate the presence of planets in distant star systems. This process was previously used with great success in classifying data in the Human Protein Atlas; however, Project Discovery has presented a much greater challenge for developers and players both.

Because of the sheer amount of data, special methods for verifying and checking problematic samples had to be developed, as results could not be hand-checked. On top of this, the difficulty of classifying the CoRoT light curves proved much greater than the Human Protein Atlas. “After launching Exoplanets on Tranquility we quickly encountered an issue where players were getting stuck in this training period because they couldn’t solve the higher difficulty tasks,” the Dev-Blog reports. Additionally, some of the data was difficult enough that it couldn’t be solved by players at all. Working with the research team in Switzerland, CCP ultimately decided to exclude this data.

Project Discovery focuses on classifying data which cannot be analyzed algorithmically because of the huge number of variables that can cause errors when using algorithmic frequency analysis. Systems with multiple nearby planets sometimes demonstrate variations in orbital periods due to each planet’s gravitational pull affecting one another’s orbital period, creating Transit Timing Variations that algorithms cannot account for. Project Discovery is also expected to outperform algorithmic analysis in systems where the stars are highly active, or where planets are orbiting binary stars.

Challenges Met

These initial difficulties were identified in the first release of the project and, through collaboration with players, the developers were able to implement solutions in the second release. Data that had too high a signal-to-noise ratio was omitted, and new tools were added to give players a higher degree of precision in responses. For example, a magnifying tool was added which allows players to accurately mark very high frequency transits. Data that presented frequent problems or which could not be accurately classified by players due to eclipsing binaries were also removed. These adjustments gave players the tools necessary to complete this first data set.

Exceeding Expectations

CCP’s tools are remarkable, but ultimately it’s the EVE Online community which has carried us successfully to this point. 77,709 players working together were able to pore over more than 44.4 million period samples in just over six months. This high level of engagement provides data that otherwise would take years to classify manually.

In their statement on Massively Multiplayer Online Science’s website, researchers emphasized the importance of manual classification of data that computer analysis struggles with. Having access to “capable, motivated” people is an important resource that helps identify data that may otherwise have been overlooked. While the data identified is helpful, researchers caution that a high fraction of the data is likely a result of artefacts, saying “much like the galaxy of New Eden, the work of finding planet transits is fraught with difficulty, even for professionals.” Still, the rapid return on data with a high level of consensus indicates that many EVE pilots are ready to meet those difficulties and rise above them, advancing human expansion not just in New Eden, but in our understanding of our own galaxy.

Let your voice be heard! Submit your own article to Imperium News here!

Would you like to join the Imperium News staff? Find out how!

Comments

  • Ryan

    Awesome article, Para, and awesome work everyone who’s contributed to Project Discovery!

    January 27, 2018 at 9:06 am
  • NotAffiliated

    EVE is such a remarkable game, and the fact that it helps advance real science just makes me love it even more.

    January 29, 2018 at 3:57 am