The good and bad of crowd-sourcing
When Google unveiled their latest project, Street View, recently, I doubt anyone over there expected such a collective back lash from privacy advocates. I mean, the service is really cool. Available in select cities, Google now allows you to zoom in on their maps and see what things look like at street level. This isn't CG. I once rented Spiderman 2 for the Xbox just so I could take a virtual tour of New York and climb to the top of the Empire State Building like King Kong. I was a bit disappointed because it didn't feel "real" enough. Google's Street View doesn't disappoint. It's actual photos spliced together and is very immersive.
Perhaps that's the problem. It's a little too real. They are actual photos taken by Google crew driving around in this van, which was caught in a reflection as it was driving by a window. That's not the only thing caught on camera. I can imagine the guys in the van thinking it was funny when they also snapped photos of girls sunbathing, some dude going into an adult bookstore, some dude coming out of a strip club, and a guy breaking into a house (among others). Some photos have since been removed like the infamous guy peeing on the side of the road and the poor girl whose thong was showing as she was getting out of a truck. The privacy advocates raised the red flag when it was revealed that in some of the photos you could see into people's houses. I'm sure Google's photo-taking policies are in the process of being revised.
So how did all these photos in a haystack surface? That's the power of crowds. Google serves millions of users per day. Whole groups of users, right now as you are reading this, are scrutinizing every inch of Street View looking for interesting photos. In this particular case, that's come back to bite Google in the arse. Millions of judging eyeballs is the downside of having a huge user base. But when it works to their advantage, it's called crowd-sourcing. Huge tasks, like maintaining articles at Wikipedia and protecting them from vandals is turned over to the crowd itself. This greatly reduces the amount of time and money spent on paid moderators by replacing them with volunteers. Yahoo bought Flickr as a replacement for its photo service exactly because Flickr has self-maintaining crowd-sourcing features built-in. Though the deal cost them millions, they need less of a work force to work it, and end up saving much more in the short long-run.
It sounds like a lazy way of getting something for nothing, and it sort of is. Still, crowd-sourcing shows promise in solving real problems. Computers are great because they can automate tasks. Unfortunately computers aren't human, and there are some things only humans (currently) can do. For example, "Optical Character Recognition" (OCR) can be used to scan books and turn them into digitized text that can be searched. It works great on clearly printed books, but when dealing with old worn manuscripts and handwritten journals, OCR starts encountering problems. Is that a lower-case "a" or an "o"? Only humans can currently tell the difference. This very same idea is used in CAPTCHA, or "word verification" spam deterrent systems. Spam bots can't tell if that's an "a" or an "o" either. It just so happens that about 60 million CAPTCHAs are solved by humans around the world every day. This few seconds of individual human deciphering translates to more than 150,000 hours of work each day. Imagine if that could be put to use somehow, like maybe translating characters OCR can't recognize. Well guess what, it is. Check out reCAPTCHA, a service doing exactly that.
Who knows what future problems crowd-sourcing will solve? Reading books while fighting spam is new way of harnessing its power, but crowd sourcing isn't exactly new. SETI years ago started using volunteer's excess computer processing power to scan the skies for extraterrestrials. Similar projects likewise borrow processing speed to crunch data on AIDS research. The "some dudes" of the world are probably planning really complex uses we can't even imagine yet. At the very least crowd-sourcing keeps us honest, as in pointing out Google's unintentional transition to Big Brother. I'm still trying to figure out how to crowd-source my finances. If only one million people would send me a buck...
Perhaps that's the problem. It's a little too real. They are actual photos taken by Google crew driving around in this van, which was caught in a reflection as it was driving by a window. That's not the only thing caught on camera. I can imagine the guys in the van thinking it was funny when they also snapped photos of girls sunbathing, some dude going into an adult bookstore, some dude coming out of a strip club, and a guy breaking into a house (among others). Some photos have since been removed like the infamous guy peeing on the side of the road and the poor girl whose thong was showing as she was getting out of a truck. The privacy advocates raised the red flag when it was revealed that in some of the photos you could see into people's houses. I'm sure Google's photo-taking policies are in the process of being revised.
So how did all these photos in a haystack surface? That's the power of crowds. Google serves millions of users per day. Whole groups of users, right now as you are reading this, are scrutinizing every inch of Street View looking for interesting photos. In this particular case, that's come back to bite Google in the arse. Millions of judging eyeballs is the downside of having a huge user base. But when it works to their advantage, it's called crowd-sourcing. Huge tasks, like maintaining articles at Wikipedia and protecting them from vandals is turned over to the crowd itself. This greatly reduces the amount of time and money spent on paid moderators by replacing them with volunteers. Yahoo bought Flickr as a replacement for its photo service exactly because Flickr has self-maintaining crowd-sourcing features built-in. Though the deal cost them millions, they need less of a work force to work it, and end up saving much more in the short long-run.
It sounds like a lazy way of getting something for nothing, and it sort of is. Still, crowd-sourcing shows promise in solving real problems. Computers are great because they can automate tasks. Unfortunately computers aren't human, and there are some things only humans (currently) can do. For example, "Optical Character Recognition" (OCR) can be used to scan books and turn them into digitized text that can be searched. It works great on clearly printed books, but when dealing with old worn manuscripts and handwritten journals, OCR starts encountering problems. Is that a lower-case "a" or an "o"? Only humans can currently tell the difference. This very same idea is used in CAPTCHA, or "word verification" spam deterrent systems. Spam bots can't tell if that's an "a" or an "o" either. It just so happens that about 60 million CAPTCHAs are solved by humans around the world every day. This few seconds of individual human deciphering translates to more than 150,000 hours of work each day. Imagine if that could be put to use somehow, like maybe translating characters OCR can't recognize. Well guess what, it is. Check out reCAPTCHA, a service doing exactly that.
Who knows what future problems crowd-sourcing will solve? Reading books while fighting spam is new way of harnessing its power, but crowd sourcing isn't exactly new. SETI years ago started using volunteer's excess computer processing power to scan the skies for extraterrestrials. Similar projects likewise borrow processing speed to crunch data on AIDS research. The "some dudes" of the world are probably planning really complex uses we can't even imagine yet. At the very least crowd-sourcing keeps us honest, as in pointing out Google's unintentional transition to Big Brother. I'm still trying to figure out how to crowd-source my finances. If only one million people would send me a buck...
Labels: business, crowd-sourcing, Web 2.0
posted by Jeremy at
12:14 AM
2 Comments:
A Great selection of Street View Finds:
http://www.mapmole.com
Feel free to view, vote and submit your own.
That's a pretty good directory J. I've seen a few out there, but I like Map Mole because they show a thumbnail. Many of them don't. Thanks for the link.
Post a Comment
Links to this post:
Create a Link
« Home