Analyzing Icelandic conviction rates with PYBOSSA
Crowdcrafting.org hosts a wide variety of applications that range from science to humanities. Since the official launch of Crowdcrafting.org, lots of applications have been created , but one of them has done a really impressive job: Héraðsdómar - sýknað eða sakfellt.
Héraðsdómar - sýknað eða sakfellt is an application developed by Páll Hilmarsson (@pallih, Github). The application was one of the most popular and active applications in Crowdcrafting.org when it was published (300 volunteers helped!), so I wanted to interview the author and ask some questions about it: why he created the application, what was the result, etc.
Páll told me that he created the application after reading the an article published in an Icelandic news web site.
The article analyzed the conviction rates of a named judged in the Reykjavik district court, stating that the conviction rates for cases where he presided as a judge was 99%. Páll found it interesting, but also “biased” as the reporter only analyzed one judge.
After the publication of the story, some bloggers and readers of the post, discussed
about why analyzing only one judge, reporting it back to the author. The journalist
addressed all the questions and comments answering that
calculating all the conviction rates for every case would take too long.
Páll was not happy with this answer, so he decided to show him, and other reporters, that this could be easily done by crowdsourcing the job, and that it would not take too long.
Páll uploaded around 4,700 rulings as tasks, and the volunteers analyze them in 7 days! Each ruling went to at least three different users, totaling 14,208 assessments. In the end more than 17,000 assessments were made by over 300 users! (you can check the stats here).
But here it comes the best part, Páll only spent 10 hours in this project (including the time to scrape the rulings, set up the tasks on Crowdcrafting.org and displaying the results on his blog. Amazing!