Google, take this idea please.
So I'm on a little bit of a Google kick today. ;) I don't think anyone from the big G reads this blog, but maybe that will change.
The web can sometimes be a cyclical beast. An image or video gets posted, does its little viral thing, makes the rounds, then disappears. As viewers become less passive and graduate into participators, we also often see remixes of said images and videos that can extend the life of this little slice of entertainment. Sometimes they disappear for good -- othertimes they get reposted a week/month/year later and get a second wind.
Each image/video on the net could theoretically have a unique fingerprint calculated and assigned to it that would differentiate it from all the other images out there. Something as simple as a hash of the pixel colors/ratios (over time if its a movie), perhaps (although I'm sure a real implementation would be much more complicated then that).
If Google were to run its spiders through its massive database of images and videos to create these fingerprints, it would be in a place to mine some interesting data out of the life-cycle of objects on the web.
As a google user, we could find out when an image was first posted on the web, and where. Who _was_ the first person to post that bouncing-breast gif animation you see in a thousand forum sigs?
As a marketer, you could track the spread of your viral image/video campaign. Search according to the above mentioned fingerprint, find the first instance (the 'seeding' that would have been done by your own organization) and then look through all subsequent hits with their time/location stamps to track how the campaign spread.
Finally, in line with my comment above re: user participation, a smart algorithm to calculate the fingerprint could also identify inherited images. If 85% of the image/movie is the same but the other 15% had been modified, google could say with some level of certainty that the second hit was a child of the first -- perhaps someone edited it, 'remixed' it, added some text in a word bubble to add a funny caption, etc. Again, probably something researches and marketers alike would find valuable in tracking how participation helped extend (or not) the life of a viral campaign.
I know I would find this an interesting tool. Hopefully one of the big brains at Google has already thought of this and we'll all have a new toy to play with soon.
The web can sometimes be a cyclical beast. An image or video gets posted, does its little viral thing, makes the rounds, then disappears. As viewers become less passive and graduate into participators, we also often see remixes of said images and videos that can extend the life of this little slice of entertainment. Sometimes they disappear for good -- othertimes they get reposted a week/month/year later and get a second wind.
Each image/video on the net could theoretically have a unique fingerprint calculated and assigned to it that would differentiate it from all the other images out there. Something as simple as a hash of the pixel colors/ratios (over time if its a movie), perhaps (although I'm sure a real implementation would be much more complicated then that).
If Google were to run its spiders through its massive database of images and videos to create these fingerprints, it would be in a place to mine some interesting data out of the life-cycle of objects on the web.
As a google user, we could find out when an image was first posted on the web, and where. Who _was_ the first person to post that bouncing-breast gif animation you see in a thousand forum sigs?
As a marketer, you could track the spread of your viral image/video campaign. Search according to the above mentioned fingerprint, find the first instance (the 'seeding' that would have been done by your own organization) and then look through all subsequent hits with their time/location stamps to track how the campaign spread.
Finally, in line with my comment above re: user participation, a smart algorithm to calculate the fingerprint could also identify inherited images. If 85% of the image/movie is the same but the other 15% had been modified, google could say with some level of certainty that the second hit was a child of the first -- perhaps someone edited it, 'remixed' it, added some text in a word bubble to add a funny caption, etc. Again, probably something researches and marketers alike would find valuable in tracking how participation helped extend (or not) the life of a viral campaign.
I know I would find this an interesting tool. Hopefully one of the big brains at Google has already thought of this and we'll all have a new toy to play with soon.
posted by Ben Mattes at 2:27 PM
0 Comments:
Post a Comment
<< Home