Is it just me, or is social media getting a little creepy? This week I came across a couple of Web sites that, as well as being cute, funny, and a little bit different from the usual Facebook and Twitter offshoots. The first, SleepingTime, lets you find out when individual Twitterers sleep–and it’s even compiled a list of the world’s foremost tech experts’ sleep/tweet times. So, if you want to know when you can rob Kara Swisher’s house, or even shin up their drainpipe just to stand over her gently sibilating body, now you know when to do it.
It’s not exactly accurate, however. Apparently, I sleep between 1pm and 9pm–utter tosh, of course, as my editor will vouch. Tyler? Tyler?–as that’s when I’m hardest at work. What it really means is that’s when I tweet least. Actually, as a somewhat second-rate Twitter user, I completely conform to analytics-firm Sysomos’s recent data, which claims that it takes around nine months for Twitter users to get really adept at using the 140-character service.
For the first three months, Twitter newbies are diligent little posters, updating their tweets regularly, and generally being good little birdies. Then, in the second trimester, they get a touch of what is known in the record biz as “Difficult Second Album-itis.” Productivity drops. Either they drop out or, three months later, they’ve got the tweeting thing down pat. Veterans’ tweets account for over 40% of the network’s tweets–and Twitter is currently churning out some 53 million tweets per day.
However, I digress. Our second slightly creepy (but still quite fun) site is from Hunch, the guys who showed us just how politically stereotypical eating habits can be. Its Twitter Predictor uses algorithms to work out who you are, and, with the help of your Twitter name and the people you follow, asks a bunch of very esoteric questions that most people wouldn’t even think of asking. (Sample: Are you more likely to spoon or be spooned?) Its average accuracy is, apparently 85%, although I scored 69%.
CNET’s Caroline McCarthy tried it out and said that it wasn’t until question 39 (Do you ride a Segway?) that the predictor got it wrong. I, on the other hand, outfoxed it on question 4, when it seemed to assume I was in possession of a pair of testicles. This is, however probably something to do with the fact that most geeks are male, rather than female, and Hunch’s co-founder Chris Dixon admits that the site is biased towards early adopters, rather than an older demographic.
Hunch is on the cusp of releasing an API of its algorithm to third-party sites, which bodes far better for them in the long run, as its original guise of using crowd sourcing to help people make decisions is a bit of a one-trick pony. It goes without saying that those who stand to benefit most from Hunch’s cleverness will be social networking marketeers and advertisers. No wonder it just banked another $10 million in venture capital.
As with so much of the social media playground, there could be privacy issues, although the firm says it has no plans to sell the data on to the marketing men. Dixon does play down Hunch’s accuracy, claiming human nature. “People in our studies are only consistent with themselves about 90% of the time,” he told McCarthy.
Playing around with the Twitter Predictor, and writing this post rather reminded me of a very strange feature i wrote on another site. A couple of years back, I interviewed a very interesting guy who described himself as a technosexual. Zoltan (not his real name) had a robot called Alice which ran on software, also called Alice (aka Artificial Linguistic Internet Computer Entity) created by Dr Richard Wallace, with which he was having a relationship. A.L.I.C.E. works by applying heuristical pattern-matching rules to the human’s input, but falls way short of the Turing Test–although Zoltan explained to me that Alice dumped him when he went too fast too soon, and he had to wipe her memory and start over–so maybe she wasn’t that dumb after all.