Shared publicly  - 
 
'Micro-Play' -- Getting the Crowd To Have Fun While It Earns You Money.

In this blog, I interview Luis von Ahn, a brilliant entrepreneur and game-design strategist and continue my exploration of how online games can help the crowd sort through data in a fashion that is both Fun and Free.

When you log into certain websites, sometimes the site makes you copy a bunch of blurry, squiggly, "drunken" letters to prove you're a human instead of a bot. Did you know that when you're typing in those letters, you're actually helping to digitize old books and newspapers through a service called reCAPTCHA? In this interview I sat down with its creator, a visionary entrepreneur named Luis von Ahn. (For more details, check out von Ahn's 2008 Science Magazine article called "reCAPTCHA: Human-Based Character Recognition via Web Security Measures.")

I began the interview by getting Luis's background. A computer scientist at Carnegie Mellon, Luis was pondering how humans interact with computers, and what people could do better than computers. He remembers thinking, "There are a bunch of problems that computers cannot do:  Why can't we just get people to solve them for us?" Back at the turn of the millennium, Luis said, "nobody had realized this power of the Internet. The word 'crowdsourcing' didn't exist." So Luis came up with this idea of a game to see for himself how he might tap into the power of the crowd through its game-playing.

One thing that humans still do better than computers is recognize images. In one of Luis's earlier games originally called ESP Game, he gamified the process of labeling images. "The way the ESP Game worked was this: two random people were paired with each other and they got the same image on their screen and were told to type whatever the other guy was typing," he said. "People typed a lot of words related to the common image. I realized that if we could get two people to agree on the same word, that would be a really good signal it's a good tag for that image," he said. But there was a caveat: "Turns out that wasn't quite enough in it to make it fun," Luis said.

"The way to make it fun was by adding a timer." Both players, randomized from different parts of the globe, entered possible words until they had a match. The game then showed them the next image. They had two and a half minutes to label 15 images. "It turns out that the time component for this really made it a lot more fun," he said.

"In fact, that was the entire motivation: to enjoy oneself for a few minutes online." Luis's estimate was that over the course of four years, some 10 million people played the ESP Game and labeled some 100 million images.

Even though the game was later acquired by Google and operated under the name Google Image Labeler to improve image search, the game is no longer available for play. "There's a good reason for that," Luis said. "It lasted about five or six years as a popular game, and then it started losing popularity. That's the thing with games. You've got to keep on. Almost no game lasts."

This idea of doing two things at once -- playing and contributing to research or analysis -- eventually led to the creation of reCAPTCHA, "where we started getting people to do useful work while they're typing the CAPTCHA," Luis said. "We are getting people to digitize books while they're doing this," he said. The number is colossal: All told, about 200 million CAPTCHA squiggles are typed in a day. Why not tap into that?"

"The thing is, each time you type one of those you waste about 10 seconds of your time," Luis said. "If you multiply that by 200 million, you get back that humanity as a whole is wasting around 500,000 hours every day." The inspiration was this: Why not use the phrases from books to be digitized as CAPTCHA squiggles and have people do two things at once?

As the company's website explains, "reCAPTCHA improves the process of digitizing books by sending words that cannot be read by computers to the Web in the form of CAPTCHAs for humans to decipher. More specifically, each word that cannot be read correctly by OCR is placed on an image and used as a CAPTCHA."

I asked Luis to pick out three lessons he'd learned from creating his games. Here's what he selected:

1. Online games for data-mining have a short virtual shelf life. People get bored, especially if the game seems stagnant.

2. The games must be fun AND the data must be worthwhile. Luis created a game called Verbosity, "but the data I collected was not that useful," he admitted. "The idea was to collect what are common-sense facts, such as milk is white and water is wet. The idea is, if only computers had common sense they could be as smart as us. We were able to collect a lot of common-sense facts with this game. The problem is, the facts that we were collecting were just not very structured so we had a bunch of facts but we didn't take that a step further to make them useful to some sort of reasoning engine."

3. Choose "Cooperation" or "Competition." "In your game design, you can either have your players competing or cooperating. That's kind of a big, big difference," Luis said. "For example in the ESP Game you were not competing with the other guy you were cooperating with them. You both got points together. That, I think, attracts a different type of player. You can experiment with both types to see whether to make the game cooperative or competitive."

With gamification, a company can actually do something that had not been doable, Luis said. "Sure we can do all kinds of things we weren't doing before [with computers], but I don't think we've covered 50% of what we need to cover. Anything having to do with language and visual recognition is very hard." And for the foreseeable future it will require both humans and computers -- and games.

In my next blog, I'm going to look at Duolingo, another company Luis co-created, and I'll explore his 10 steps a company can take to ensure the successful gamification of a project.

NOTE: As always, I would love your help in co-creating BOLD, and will happily acknowledge you as a "contributing author" for your input. Please share with me (and the community) in the comments below what you specifically found most interesting, what you disagree with and any similar stories or examples that reinforce this blog that I might use as examples in writing BOLD. Thank you!
24
8
Joel Bomane's profile photoSergio VM's profile photoCarlos Bisbal's profile photomarika amvrosiadis's profile photo
27 comments
 
I'm so glad to learn this! I've been hating those annoying time wasting Captcha things for years. Now that I know what their alternate function is I think they're genius. Looking forward to my next opportunity to help digitize books!
 
Luis is a master of cleverly disguising the useful side-effect, and wherever we can leverage human needs, or desires fulfilling them while obtaining additional information that computers can use, it is a further piece in the co-evolution of our human-technology complex.

Foursquare does this well, with location based information.

+dotSUB applies it to make videos smart, and to make sure that computers can understand videos at the depth that we humans relate to them, emotionally, and almost viscerally. (Full disclosure, I am CEO of dotSUB.)

My guess is that the one of the next groups of applications that are going to be built will gamify the management role, and disrupt or even eliminate middle manager roles which today still encumber and slow down organizations!
 
I really like this blog. Congratulations Luis!!!

The universe learns every day of what happens in it and applies it in a logical and orderly manner always to achieve improvements in what it is creating.
  
What distinguishes human beings from other species is the speed humans have to link ideas and different concepts within their mind, what Luis is doing will  help us, in the future, to understand the mental processes of humans, process that will take us to be even more intelligent (similar to the infinity process of the universe)
 
Love this. I'm a big fan of citizen science—crowd data collection and analysis projects (think "SETI screensaver)—and didn't know Captcha was pulling double duty. Citizen science: http://scistarter.com/index.html A few real-world data science games to play:

FoldIt (protein folding) http://fold.it/portal/
EteRNA (DNA molecules) http://eterna.cmu.edu/web/
Galaxy Zoo (scan the universe) http://www.galaxyzoo.org/
And many others

Minor: On the issue of cooperation and competition, I think a mixture of both can lead to some terrific game dynamics, a balanced push-pull of motivations and rewards. Doesn't have to be an either-or, and in fact, some of the best cooperative games introduce an element of competition against a common foe. REALLY interesting games entice the player to choose how they want to play—as competitors or cooperators and offer obstacles and rewards for each.
 
Interesting, but surely a human has already deciphered the reCaptcha because there is a correct and incorrect answer?
 
The study of the human mind by the EEG technique is truly interesting, this video shows what Dr. Aditi Shankardass has achieved in this regard:  http://www.ted.com/talks/aditi_shankardass_a_second_opinion_on_learning_disorders.html

Other renowned scientists are also making great progress in this regard. 

As Dr. Shankardass said what we do or answer (the result of the processing in our mind) gives us clues of how we process the information in our mind, this result is influenced by the neural connections that we have and the culture (country) in which we live.
 
The Oracle of Delphi said: “Know yourself”... This is really difficult!!!
 
+Mark Hidden
 Mark, my advice here is to remove those videos from public view immediately. Only show them to people you are seriously considering partnering with. The game idea is a good one. You need to partner with a game developer and you need funding to make it real. If you set up a kickstarter account, you can get funding with which to hire developers. You'll gain an initial audience for the game from those who choose to fund it. Some of your funding should also go to advertising once the game is ready to go. From what I saw in the videos, doing this as a digital game would allow you create infinite variations of play, as well as adding layers for more complex play. I would also suggest you add a very basic layer for people who don't already play chess, so they can learn.
Visit kickstarter.com, look at how the proposals are made. Create a polished proposal - use a script and stop apologizing for "flaws!", explain the game in an elevator speech, not in extreme detail. Hire a copywriter if you don't feel skilled enough to present it in a compelling 3 minute synopsis. Set your fundraising goal at a rational place - don't ask for $50,000 if you can get a reasonable game online for $5,000. Ask for the $5,000. If you surpass the goal, all the money raised goes to making it happen. If you fall short of the goal, none of the money comes through, so make the goal rational.If you need more than you end up with in pledges, you can trade a percentage of the game sales to the programmer(s). Either way, make sure you have good contracts in place with whoever you hire. That was important, so I'm going to repeat it: Have solid contracts in place with anyone who works on developing the game. Use some of the raised funds to hire a lawyer for the contract process!
It's a good game. Make it happen.
 
+Rikki Ansell First that is a very nice thing to say.

but why do you think the videos should be taken down?  
To me I am not concerned about someone stealing the idea, because I have documented it is my idea.  I will even given anyone the source code I have already written.  And if you want to do a KickStart I would be glad to help you. And if you get rich off it great!  my true interest is in seeing the idea catch hold, because there's depth in the game I want to explore and I need a lot of people to make that happen. So my interest is  in promoting the idea, and myself as an idea guy.  So I know what I am getting by giving the idea away. And it really is a crowd sourcing problem how do I get as many brains working on this that I can.
 
If you want it to be open source, post it to linux. If gamers are interested in it, they'll work on it.
 
Peter, Not being of a science background, I had difficulty understanding this. Christian Brugger's link to the TED video was helpful. I also like the generosity between Mark Hidden and Rikki Ansell, Mark looking at the issue from an "abundance" point of view and Rikki from a "scarcity" standpoint. I do not intend to pass a judgement on Rikki (whom I admire for her generosity) by writing this but it is worth noting an illustration of "Abundance" vs. "Scarcity" approach. Thank you for the educational experience on "gamification"!
 
+Mark Hidden +Rikki Ansell Love the communication and idea sharing guys! The idea of gamification for decoding data is fascinating to me. Similar to the Fold It challenge which involved the gamification process of protein folding, the reCAPTCHA process beautifully intertwines a rewarding factor to a monotonous everyday process. This type of gamification is essential to the booming world of technology.
 
What I like in the reCAPTCHA initiative is the motivation behind it. I like the idea to stop wasting human brain work and human potential. The Duolingo project helped to integrate human power and intelligence to complete parallel tasks (the translation) in addition to the main task (verifying of being human).  Such additional tasks which can be done as by-products with the same original expense in human capacity I would characterise as economic externalities.   Completing work and harvesting benefits during playing is also a great idea.
 
Very cool as usual Peter, you certainly have a nose for talking to the right people! It's excellent to learn that Luis was also part of Duolingo, I didn't know that, but it makes sense knowing what I do. IMO the entire idea behind Recaptcha is one of the most brilliant that I've seen in the last decade or so... right down to the name etc. In fact, in some of my own work we use Recaptcha to point to a major distinction in our theoretical research about dispersed knowledge, the Crowd, and IT. It's called "The Theory of Crowd Capital" and you can download it here if you're interested: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2193115

In this work, we make the distinction between Crowd Capital derived from "episodic" and/or "continuing" means, where Recaptcha is an example of the episodic variety...some Citizen Science apps and Prediction markets could also fall into this category. Whereas on the other hand "continuing" means could be exemplified by something like wikis or wikipedia, among others. In other words, for the former you do not need community, collaboration or relationships for the Crowd Capital resource to be generated, whereas with the latter, you do.

Now that I think about it, if we take our distinction and combine it with Luis's distinction of "cooperative" and "competitive", we may have a nice 2x2 matrix to work with!

Would love hear everyone's feedback in this regard!
 


the web of moments
the basic idea is to create a meeting ground between all the various social networks ... a place for users, admins, brands alike to exchange in a highly councious and respectfull way. also a place to collaborate and use synergies from combination of modules / features of the several services.
there is an open public text pad at http://titanpad.com/RXne8kf5YS where the concept can be evolved collaborativly, accompagnied by an open drawing room at http://cosketch.com/Rooms/nneqvrw

similar is another concept of mine i named
throw it onto the heap
what is about
feedback gamification trough playfull content rearranging
joyfull data exchange furthering research 
articles/comments/advertisements .... being dissembled, trown onto a heap ...for the reader to play with it lego style, rearrange it in any way using snippets of it, doodle and graphic editing tools. the rearranged unheaped played with material might provide clues to comprehension rates and contribute to (marketing) research.  data of both emotional as well as logical nature might be extracted / interpreted / indicated from the playfull remixes.

at the time when i was thinking about how "games for good" can be improved, i have collected some more ideas at
https://plus.google.com/112447599115174765459/posts/13G23EbcxdJ

additional to the software part i have also been thinking about the material gear what the future most of the time mobile self guided studying ( and gaming and enjoying ... ) individual might prefer instead of sitting in classrooms

http://mayloveheal.wordpress.com/2009/01/14/my-wish-of-a-virtual-reality-interactive-game-creation-and-playing-dome-with-energy-harvesting-features-and-also-communicationexchange-evolution-goals/
 
This was posted on my facebook page and I wanted to share it with you:

Kent Kemmish Hi Brad Arnold- These pioneers of gamification are doing awesome things, but it's important that I differentiate what we're doing from everything that's gone before. The Game to Cure All Diseases, the game the The Church of Sequence Space will play, will allow direct remote R&D through access to remote automated facilities. Anyone with access to an internet connection will have de facto access to infrastructure equivalent to that of a facility such as Harvard's Wyss Institute. Nothing like this has been done before- current science gamification projects are essentially used for "human computation", The Church of Sequence Space will be used directly for R&D design and execution.
Saturday at 10:50am

https://www.facebook.com/ChurchOfSequenceSpace/posts/4900167108703?comment_id=61492805&offset=0&total_comments=1&ref=notif&notif_t=share_comment

This guy (Kent Kemmish) is a genius, and I strongly suggest someone utilize him.  If Mr Diamandis reads this posting: if what I am perceiving is accurate, Mr Kemmish is doing something paradigm changing.  Although I'm an alpha thinker, I am still trying to grok it though.
 
Cancer, unfortunately. I was listening to a story on NPR while driving up the coast about an area that had a strong concentration of cancer in little kids. The problem was caught way later than it should have been simply because the data wasn't being collected. It was actually collected eventually by a journalist, not doctors. CANCER PATIENTS: This is an area where instead of doctors cataloging information, why not patients?
 
Peter, 
Thanks for interviewing Luis. I'm hooked on Duolingo and hooked on his story. What a fine example of Conscious Capitalism (I'm on the board) and an Exponential Organization (Salim's upcoming book). I'll look forward to your next post, and to SU in September. All best, S
 
LOVE and PLANETARY HEALING is so much FUN
 
How about we work on a profoundly challenging idea? Why not develop a game that demonstrates a new economic system? The game would assume that there are only two resources, mankind and earth. The game would show how each of us would have all of our necessary human needs met in a system that is none competitive by it's very design. In this system each of our personal needs to create, whether great or small, would add to the collective pool. The game would have to have some incentive to reward those that contribute more of their time to the process but at the same time limit over payment to any particular individuals that are driven to extraordinary feats of contribution. Not a game to be played in virtual reality but to be applied to our life here on earth. This project needs doing for all the obvious reasons.
Add a comment...