How to Craft the Perfect Instagram
By Natalie Matthews
Sometimes, when I find only a fraction of the "likes" that I expected in that crucial first hour following an Instagram upload, I start walking towards my Internet box to reset it, because that must be the problem. My vacation shot of myself and my European model boyfriend driving an exotic vintage car through the gardens of an Indian palace? How could that not explode the Internet?!
Of course, I don't actually have that photo, because that's Olivia Palermo's life, not mine. But I do know that sort of Insta-confusion when you think a photo's THE BEST THING EVER and the Internet doesn't seem to agree. (Weirdly, that Indian palace photo got Olivia 21,000 likes, but this one of her just standing in a basic NYC office building got 42,000.)
If you've been struck by it, too, you might want to check out this new website, courtesy of MITs Computer Science and Artificial Intelligence Lab, eBay Research Labs, and DigitalGlobe. Led by MIT doctoral candidate Aditya Khosla, they wrote an algorithm to help predict how popular photos will be on social mediabefore you even post them. And all you have to do to test it yourself is upload your own photo and press the "Run" button to see its "popularity API."
To calculate their formula and final scoreswhich can be as low as 0.8 or as high as sevenKhosla and his team sourced 2.3 million photos from Flickr and gathered data points on each of them. They considered the photos' colors, textures, gradients, objects present, tags, and more, then measured what gets the most reach. They parsed out some overarching trends, like this one:
Some of it is pretty obvious (scantily clad ladies = likes?!), and some of it seems like it only makes sense in certain situations (a gun-holding, miniskirt-wearing woman drinking tea and looking at maillots would probably just be...weird).
But still, that doesn't mean browsing their research or playing with their image scoring widget is any less fun. Plus, it's basically the only Instagram tool that can help you predict if a post is a good idea/bad idea/medium idea before you put it up. (Hefe filter, thanks for the perma-tan, but even you can't do that.)
Yet, much like those guys who calculated how to make the perfect online dating profile, the science seems like it would zap some of the fun out of things. If you always know how well something will do and how many people will like it (or roughly) on Instagram, there's no heart-thumping, moment-of-truth when you check itgood or bad.
More on Social Media:
Photos: Courtesy of Instagram