Crowdsourcing - Part 2 (Research & Information)

(This is Part 2 of the notes for my presentation on crowdsourcing for my INFX598 course in social media.)

Information Projects – “Encyclopedia” Type

The most familiar type of crowdsourced project is the “encyclopedia” type, where a project provides a structure and often editorial and other management toward creating a large “anthology” work composed of public contributions. I will touch on some examples, although I’ll discuss them quickly because so many of them are already familiar.

Wikipedia is, of course, the archetypal example of this.  Another example of this type of project is the recipe repository, of which there are many online – RecipeSource is one of the oldest, while AllRecipes is one of the most popular. There are also many sites that collect song lyrics contributed by users, such as Lyrics.com. A crowdsourced project of special interest to librarians is LibraryThing, which was originally designed as a tool for cataloging a personal book collection, but has grown into a more general-purpose database containing book information, demographics, historical data on book collections of famous people, and more. Another important literary project is Project Gutenberg, the oldest repository of e-texts in existence. Since 1975, Project Gutenberg has coordinated public efforts to digitize and archive public domain works and provide them free to the public. Project Gutenberg also uses a sophisticated system of crowdsourced effort to proofread its books, and an affiliated project, LibriVox, is working on creating and archiving free audiobooks of public works produced by crowdsourced effort.

Some projects begin as a different type of crowdsourcing project, but turn into an encyclopedia-type project over time. Ask Metafilter began as a project to create a large community that would answer any question asked; it was an outgrowth of a larger community blog, MetaFilter, where members would post anything they thought might interest other people. Over time, both sites have become a huge searchable repository for answers to a staggering variety of questions ranging from technical problems to personal quandaries to where to find other obscure information. If you need a more focused way to crowdsource personal issues, WotWent Wrong offers a website and app that lets users anonymously upload the details of romantic relationship breakups, so that other users can advise, counsel, critique, and offer closure.

Research Projects – “Mosaic” Type

Much crowdsourced research is based on the idea that lots of people will donate small bits of time or effort to a cause, especially if it’s one they believe in and the task is fun, and those small bits of work add up. Crowdsourcing really excels at doing work that’s easy for humans, but hard for computers to do, such as image recognition and metatagging,  and natural language recognition.

A sample FoldIt puzzle

One very effective crowdsourced research project is FoldIt. Understanding the structure of proteins and how they can fold is a key scientific problem in understanding many diseases and finding cures for them, including HIV, Alzheimer’s Disease, and cancer. Solving this problem has been one of biology’s toughest challenges, difficult and expensive to research using computers. FoldIt enlists the help of users by offering them a puzzle based on a protein folding problem. Humans can see solutions to such a puzzle much more easily and faster than computers can, and have fun doing it. FoldIt players have made great progress in adding to scientific knowledge in this field, including solving the structure of a retrovirus enzyme critical to developing anti-AIDS drugs in a matter of days, where it had previously eluded scientists completely.

 Another game project, Galaxy Zoo, enlists the public in morphological classification of galaxies by asking them to look at telescope photographs and identify galaxy types visually. Volunteers have identified more than 70 million galaxies so far, most of which had never been seen by human eyes before, since the photographs were taken and processed by robotic cameras.

A CAPTCHA is a website security device that makes sure users are human and not virtual, to reduce spamming and other malicious site interference. Almost everyone has seen the box where you’re asked to re-type some letters before you can enter a site. reCAPTCHA is a service that uses this effort, millions of times per day, to help identify digitized text that can’t be read by optical character reader software. Words that cannot be read correctly by OCR are given to users in conjunction with another word for which the answer is already known. The user is then asked to read both words. If they give the right answer for a known word, the system assumes their answer is correct for the new one. The same word is given to many users to verify the answer. Currently, the project is helping create digital editions of older issues of the New York Times and books in the Google Books project.

Another project that uses language recognition is What’s On the Menu, the New York Public Library’s effort to make its historical menu collection into a searchable database. Users who log in can help identify dishes, prices, meal organization, geographic information, and more from scanned selections from the collection’s 45,000 menus. Eventually, this information will be available for historians, cultural researchers, chefs, educators, and anyone else to make new connections and discoveries, and learn more about our culinary past.

Project Implicit provides visual tests that the public can take to measure implicit bias and attitudes. By asking users to choose quickly between images and words onscreen, the Implicit Association Test makes it possible to measure attitudes and beliefs that people are either unwilling or unable to report, such as prejudices that may be unconscious or embarrassing. The information provides interesting individual feedback as well as valuable research information about public attitudes.

PatientsLikeMe was created in 2002 as a way to help accelerate learning on amyotropic lateral sclerosis, aka Lou Gehrig’s disease. Now its 80,000 members share personal details of their medical history with fellow members in a network that not only provides support, but data that is aggregated to track patterns and responses to various treatments. For rare diseases like ALS, many doctors may only encounter one or two patients in their lifetime, but PatientsLikeMe allows them to compare and review treatments with thousands of other patients, helping them quickly understand options and determine an effective course of action.

Often, crowdsourcing can respond quickly in times of crisis or when other infrastructures have broken down. Ushahidi  is a company that provides open-source software to enable collection, visualization, and dynamic mapping of crowdsourced reporting of time-sensitive crisis information. It was initially developed to map reports of violence in Kenya after the post-election fallout at the beginning of 2008. The original website was used to map incidents of violence and peace efforts throughout the country based on reports submitted by 45,000 users via the web and mobile phone. Since then, Ushahidi continues to develop and freely distribute its platform, apps, and services to help coordinate social media and individual in-the-moment reporting of problems and assistance routing for people dealing with situations like bombing in Mumbai and the recent earthquakes in Japan and Haiti. Ushahidi is also being used by human rights organizations to map incidents of violence, corruption, and other issues of note that are often overlooked or suppressed in traditional media reporting. Crowdsourcing can also help in the aftermath of a crisis – MapMill is currently enlisting the public to analyze aerial photos of places affected by Hurricane Sandy, to help quickly identify and assess storm damage.

(Next: art projects, financial applications, and more.  See all the links at this Tumblr, and read more articles about crowdsourcing at this Scoop.It page.)