A character count connoisseur? An algorithm aficionado? As it turns out, the big bad internet has got you beat. (Okay, at least it had me beat).
Yesterday afternoon The Upshot, the data-driven section of New York Times, put tweeters to the test. A new algorithm developed by three Cornell University computer scientists’ claims to outperform the average person in telling which of two similar tweets will be retweeted more. As I made my way through the quiz, I was pretty impressed with myself (though I tended to predict Diddy’s tweets better than President Obama’s so I’m not exactly sure what that says about my journalistic credibility), only to find that the algorithm beat me by 4 points.
This corresponding article by Sendhil Mullainathan, Professor of Economics at Harvard University, goes on to explain why social media-ites and the rest of the digi-sphere must remain calm. While this algorithm is tremendously impressive, it doesn’t mean that we should all go pulling out our resumes just yet, and here’s why.
Correlation does not equal causation
You thought I was done making Psych 101 course references in my blog, didn’t you? #Nope.
“We care about predicting retweets mainly because we want to write better tweets. And we assume these two tasks are related. If Netflix can predict which movies I like, surely they can use the same analytics to create better TV shows. But it doesn’t work that way,” write Mullainathan.
Basically, you could change your tweets to mimic those being retweeted more, but it won’t promise any change in your followers’ behavior. If I’m a bartender and 5 dudes order Michelob Ultras consecutively, it still wouldn’t make sense for me to keep one on deck for the next dude (because hopefully, he would NEVER order a Michelob Ultra and then we would date).
Quality vs. Quantity
The algorithm found that longer tweets were more likely to be retweeted…but that’s only because the lengthier tweets contained more content. Your best bet is to stick with the “less is more principle” when it comes to character count, but to pack as much content into your tweet as possible
Another fault (IN OUR STARS! LOL!) of the algorithm is its inability to predicting what’s interesting…which is good, because otherwise I (along with other content creators and trend forecasters) would be out of a job.
Take viral videos for example. How much did you laugh at that baby that falls down when somebody sneezes? How hard did you laugh the 5th time you saw it? The same goes for celebrity/entertainment news. It’s novel to see how Justin Bieber is effing his life up when one source breaks the story, but when your entire feed consists of different sources relaying the same information you read before lunch, you’re more likely to skip over it. Therefore, while the tweet predictor can pick up on something that is drawing attention, it’s more likely to exploit it than anything else.
So go forth, my fellow tweeters! Tell me about your annoying co-workers or how you’re going DTS this weekend and I will proceed to retweet you if and when I feel like it.