Some cool stuff, mostly about Twitter — that might have a chill attached.
The One Million Tweet Map (from Maptimize) shows you where in the world the birdsong is coming from. This has a fascination all its own, a kind of gossip about gossiping writ large, perhaps. And then there’s what it reveals about Canada. When you zero in on this, the second largest country in the world, you see (once again) how much we huddle in the south. And, apologies all round, how much of the action is centred in the GTA.
The map can show you recent tweets as clusters labelled with the count or as a heat map. I prefer the former because it spots out the rare and isolated tweet, of which there are a number in Canada. Click on those 1s or 2s and you’ll be able to see the tweets themselves, something the tweeters don’t realize, I’m sure. (Where there are larger clusters, clicking on these will zoom you in, a process you can repeat until you can isolate individual tweets.) There’s apparently an ability to filter for keywords or for hashtags, but I haven’t found it to be accurate, and the ability to filter for recency seems not to be working at the moment, which is too bad. I hope they fix it.
The bigger point, apart from the map’s voyeuristic value, is the power that aggregation gives you — or, rather, some corporation. Trends emerge, big data is sifted into sense. Take for instance NoHomophobes.com. This site filters the Twitter stream, isolating the offensive expressions “so gay,” “faggot,” and “dyke.” The main page gives you a running count on how often each has been used in the day. And down below that there’s a carousel of tweets that use these words. As well, in other tabs you can see in graph form how the usage has varied over a month or “all time.”
You don’t have to be a whole lot of paranoid to see that these abilities — geolocation, filtering — have the makings of tools of control. And if you couple them with a cunning algorithm, you might not even have to wait for the actual expression of what it is you’re concerned about. I learned yesterday that MIT has developed just such an analytical tool, enabling them, in the words of the article:
with 95 percent accuracy, [to] predict which topics will trend an average of an hour and a half before Twitter’s algorithm puts them on the list — and sometimes as much as four or five hours before
I’ve a cool new app that predicts with stunning accuracy whether and for how long it will rain in my immediate neighbourhood. Soon I’ll be able to know with the same precision what we’re going to be thinking in my neighbourhood. Weather reporters and pollsters, meet buggy whip makers. Sigh. Perhaps presciently, Ray Kurzweil spoke in my neighbourhood a couple of weeks ago; sorry I missed him; he’s making ever more sense, I’m afraid.