They’re not very sophisticated, but they’re there!
Day-by-day reports of essentially the same data aren’t terribly useful, especially when they only cover seven days’ worth of history — it’s the changes over time that count, and textual lists of SELECT COUNT(*) … GROUP BY WHATEVER-WE’RE-TALKING-ABOUT aren’t the best way to find them. You could at least sort by the differences from the day before — usually not very interesting, unless some shit really hit the fan overnight, and that’s really only a big danger immediately after patches — or the differences from the week before, or the differences from the month before, or the differences from the last patch. Change is usually slow, as explorers and exploiters’ new discoveries are generally kept secret for a while. (This is why I normally advise fellow developers to not bother with a fancy metrics system that reports up-to-the-minute data — as a dev, you rarely care what people are doing right now.)
Now, metrics for player usage — the most auctioned items, for example — yes, I can see players being interested in what’s going on right now. But I imagine those who really care already subscribe to the previously mentioned wowecon.com and have a much better grasp on current auction house events.
Releasing any metrics to players is dangerous, even if they’re not very complex; I imagine the WoW boards are already full of players delighted to find that the quest completion stats prove some vendetta. I commend the effort and hope it doesn’t backfire too much.
Microtargeting the vote — directing propaganda towards voters inclined to vote for certain parties, identified through their consumer choices.
I drink microbrews and drive a Jetta, and I’m a lifelong Democrat. Except for that Libertarian phase in high school when I really liked Ayn Rand — back then, I drank St. Ides fruit-flavored malt liquor and drove a $500 particolored ‘78 Nova. Mine THAT.
The Economist has started a new blog! I, for one, am really looking forward to new posts in the Lies, Damned Lies, and Statistics category.
The most recent post extrapolates future Iraqi death counts, and finds that — shocking — a lot of people are going to continue to die.
This is appalling news for Iraq, but could actually be good news for Mr Bush, who may have been too quick to dismiss the numbers out of hand. With that many fighting-age men out of commission, it seems unlikely that the insurgency could sustain its current operational tempo. This suggests that the violence should have died down considerably by the time the next presidential election rolls around in 2008.
This would not, of course, validate the Iraq war; a million or so dead young men is a human cost too terrible for the mind to encompass, especially if all humanity gets in return is another four years of Republican rule.
How do they find out how many people have died? The first few pages of the new Lancet study (PDF) go into some detail. An earlier version of the study was covered by This American Life last year, and if you’re at all interested in the mechanics of gathering data in a war zone, I can’t recommend it highly enough.
This doesn’t really have anything to do with games, and I don’t want to argue about politics. The Economist blog reminded me of that This American Life episode, and I wanted an excuse to post it, because you should listen to it.
Subscribe