Characterizing crowd participation and productivity of foldit through web scraping

Abstract

Citizen science, scientific work done by non-experts, is an emerging method of continuing scientific investigation. In recent years, Crowdsourced Science Games (CSSGs) have become a particular area of research. In this model, citizen scientists play a video game in order to help solve scientifically hard problem sets. Recent work has shown CSSGs are severely affected by low engagement rates (ER) and a disproportionate amount of work done by a small subset of the entire player base. In this thesis, we will examine Foldit, a seemingly successful CSSG. In the absence of publicly available data, we used web scraping to obtain data on a daily basis from a player scoreboard from June 1, 2015, to February 15, 2016, and from an accumulated puzzle database encompassing the lifetime of Foldit. Utilizing previous methodology quantifying the productivity of CSSGs, we show that Foldit continues to draw players despite a gradually declining number of active users. Furthermore, a core base of experienced players contributes the most to the game. With these two factors, Foldit’s game design and emphasis toward creating a small but highly trained player subset provide a strong argument for a more productive CSSG over a more entertainment-focused, casual style of game.