The US Department of Defense isn't sure where Damascus is: statistical proof • Hacker News Board

Re: The US Department of Defense isn't sure where Damascus is: statistical proof

#11

Post by Tloewald » Mon, Sep 09, 2013, 3:06 PM UTC

curveship wrote: post #9

I knew exactly where Damascus was, but they put me down as barely making it inside their 200 mile cutoff, at 163 miles away. I think this is more about bad UI design. I got a hand cursor rather than a pointer, making it difficult to be precise, and I never noticed the zoom logo in the upper left. Plus, with only 65 DOD responses, how significant is a deviation of only 4.5% from the norm?

I cheated and still got it wrong -- it's too hard to click on their map (I doubt I was any more accurate at clicking on Arlington, VA where I live).

The Linux users doing so well against Mac users thing was interesting. I wonder if it has to do with how it displays in browsers. Also -- did the measure the delay in responses (e.g. how many people went and googled it before clicking?)

Re: The US Department of Defense isn't sure where Damascus is: statistical proof

#12

Post by VLM » Mon, Sep 09, 2013, 3:54 PM UTC

The unanswered meta question is why it should matter.

They went into the game knowing they were doing a meaningless internet quiz, so its going to be hard to abstract how intensely groups goof off when they're goofing off (or propensity to cheat in a meaningless internet quiz using an adjacent google maps tab), vs actual knowledge.

Oh well, result were reported to 3 sig figs using a whopping dozens of samples, so it must be meaningful, LOL. When you know the denominator is extremely small, you can play a math game trying to find small integer fractions that divide out to about .493 and so on.

Re: The US Department of Defense isn't sure where Damascus is: statistical proof

#13

Post by lhc- » Mon, Sep 09, 2013, 6:22 PM UTC

Tloewald wrote: post #11

curveship wrote: post #9

I knew exactly where Damascus was, but they put me down as barely making it inside their 200 mile cutoff, at 163 miles away. I think this is more about bad UI design. I got a hand cursor rather than a pointer, making it difficult to be precise, and I never noticed the zoom logo in the upper left. Plus, with only 65 DOD responses, how significant is a deviation of only 4.5% from the norm?

I cheated and still got it wrong -- it's too hard to click on their map (I doubt I was any more accurate at clicking on Arlington, VA where I live). The Linux users doing so well against Mac users thing was interesting. I wonder if it has to do with how it displays in browsers. Also -- did the measure the delay in responses (e.g. how many people went and googled it before clicking?)

I used the mousewheel to zoom in, making it much easier. I still was 140miles away though.