| |
The next level of Social Networking
Social networking is the current big internet trend. Every brand wants to create a network where people can collect and manage "friends", in order to own an ever-expanding network of customers.
While creating a social network collects an audience for a brand, any social network that is newly created has to become popular enough for people to be willing to sign up for "yet-another-network". In order to get people coming back and participating, there needs to be some sort of recurring motivation. Simply letting people post comments on other friends' pages is not enough to encourage usage and retention.
The next level in the social network equation is a concept called crowdsourcing. Basically, crowdsourcing (or community-based design) is the outsourcing of a task to a large group of virtual strangers. Similar to a flash mob (which is a gathering of a group of strangers to perform certain tasks in a randomly defined location), crowdsourcing asks large groups of people to perform discrete tasks, in order to complete a larger project.
SETI @HOME
The Search for Extraterrestrial Life had a program that anyone could download, which would run on their computer and analyze radio data recorded from outer space (to try to find signals that could only be created by another intelligent species). Each computer would analyze tiny snippets of a signal, and send information back and forth to SETI's main computers. In this way, programs that would take decades to run on their computers, could be run in weeks or months on thousand of volunteer's computers around the world. However, the SETI program was a fully-automated process, with no "human" interaction – which is where crowdsourcing comes in.
reCAPTCHA
A recent example of crowdsourcing is a service called reCAPTCHA. CAPTCHAs are those little squiggly codes (five to ten letters) that you have to type in, in order to prove that you are a human (and therefore not an email spam program). reCAPTCHA takes it to the crowdsourcing level, by using text from books that could not be digitally "read" by software algorithms (because they were so blurry, or distorted), in order to have real humans type in what the words say. They have been able to create text versions of archives from the New York Times and digitized text from the Internet Archives. According to an article in The Register (13 December 2007), reCAPTCHA delivered about 30 million images a day...which results in about 3,000 man-hours of free labor (according to Wikipedia).
Back in 2007, I was working with a team of people on a sports trivia content-based website. Unfortunately, we didn't have any initial content, and the client didn't have the budget to license the vast universe of data that we needed (the equivalent of every sports statistic possible). None of us were sports trivia experts (those that were specialized in a sport or two) and this was simply too much information for a small group to consider managing.
Case Study: Trivinger
To solve this, we came up with a crowdsourcing microsite called Trivinger (after a "trivia-scavenger hunt"). The idea was to randomly generate a trivia question (based on some predetermined data that we had generated), which users could directly answer (or search for the answer on google and then enter). For example, we knew the names of all of the teams for baseball, football, etc. and we knew possible positions for each sport – which meant we could start by asking "who is the quarterback for the Dallas Cowboys?" Once we had multiple answers that matched (for us to consider an answer "correct", we looked for at least 3 matching answers, or a minimum of 80% matching if there were more than three answers). Once we knew that, we could start asking more personal details such as "what year did Tony Romo join the Dallas Cowboys?"...
This system allowed us to let users populate our data, but also protected us from bad answers. (For example, once a question was answered, we did not allow the same question to be answered again in the same state/zip code). Because we had millions of possible questions, it was unlikely people would be able to gang up and put invalid data in the system.
Then, in order to encourage people to answer these trivia questions, we ran a prize program. For every correct answer, the user would win a series of points (scaled to be higher the more finely detailed the questions got – for example, more points if people knew what high school a certain person attended, especially if they had more of a background team position – basically, the more obscure the question, the higher the points). We already had a cost for what an accepted answer should cost us, which gave us the cost per points, which allowed us to give away a flatscreen TV (and know that it was worth a specific number of points). Users could save up their points (waiting for a larger prize) or use their points on a series of smaller prizes (which changed depending on what we wanted to buy). Our client was also looking into getting sponsors for prizes (making the cost negligible).
One of our favorite parts of this system was that it encouraged cheating – we even had a search bar (with dropdowns for Google, MSN, and Yahoo!) directly integrated into the form, so users wouldn't have to leave the page. Spidering and scraping websites could only give us so much information – there was a human element to this that we needed, and we appealed to people's gambling and prize-winning nature in order to populate our database. From all perspectives, this was a highly successful venture.
Conclusion
When designing "viral" or "social network" programs, opportunities for crowdsourcing should be considered as ways to get users more fully engaged (and continually re-engaged) as well as an opportunity to collect a database of content that would otherwise be impossible to collect, or at the very least cost-prohibitive.
0 Comments
Submit a Comment:
|
