Comments on: Hacking the Scantron http://ascii.textfiles.com/archives/931 Jason Scott's Weblog Thu, 09 Feb 2012 03:22:38 +0000 hourly 1 http://wordpress.org/?v= By: Mungojelly http://ascii.textfiles.com/archives/931/comment-page-1#comment-3352 Mungojelly Sat, 21 Jan 2006 22:41:08 +0000 http://ascii.textfiles.com/?p=931#comment-3352 It's Digg's fault. The decision whether to put something on the front page of Digg is made by a whole bunch of users-- based on their momentary first impression upon seeing the link. It should therefore be considered only a rumor mill. Anyone who takes things there as sovereign facts based on how many times they've been Dugg is a fool and deserves their ignorance. I've been thinking about strategies for verifying information online. The old school media wants us to believe that the way to get accurate information is to have a Very Smart Person sitting between you and the facts-- thank you but no. That's obviously no recipe for infallibility. Wikipedia (or rather, the strategy "put up a sketch version of the facts and then argue about it") is proving to be just slightly closer to the mark than the old system, but it's still pretty far off. I have a hunch that we could learn a lot about this task from the intelligence community (spies). Basically what they do is to distill what-is-true from tremendous amounts of imperfect data, through large numbers of imperfect (sometimes even treacherous!) agents. They know how to split the job down into small parts (compartmentalizing). They know how to analyze raw documents in ways that not only extract meaning, but extract meaning in ways that more meaning can easily be extracted from the composite of the extractions. I think we need to work on systems like that for citizen journalism. <3 It’s Digg’s fault. The decision whether to put something on the front page of Digg is made by a whole bunch of users– based on their momentary first impression upon seeing the link. It should therefore be considered only a rumor mill. Anyone who takes things there as sovereign facts based on how many times they’ve been Dugg is a fool and deserves their ignorance.

I’ve been thinking about strategies for verifying information online. The old school media wants us to believe that the way to get accurate information is to have a Very Smart Person sitting between you and the facts– thank you but no. That’s obviously no recipe for infallibility. Wikipedia (or rather, the strategy “put up a sketch version of the facts and then argue about it”) is proving to be just slightly closer to the mark than the old system, but it’s still pretty far off.

I have a hunch that we could learn a lot about this task from the intelligence community (spies). Basically what they do is to distill what-is-true from tremendous amounts of imperfect data, through large numbers of imperfect (sometimes even treacherous!) agents. They know how to split the job down into small parts (compartmentalizing). They know how to analyze raw documents in ways that not only extract meaning, but extract meaning in ways that more meaning can easily be extracted from the composite of the extractions. I think we need to work on systems like that for citizen journalism.

<3

]]> By: Masked and Anonymous http://ascii.textfiles.com/archives/931/comment-page-1#comment-3351 Masked and Anonymous Sat, 21 Jan 2006 07:29:56 +0000 http://ascii.textfiles.com/?p=931#comment-3351 Actually, you can undigg things, but the link is hidden in an obscure part of the site. Just one of many reasons why Reddit is better than Digg. Actually, you can undigg things, but the link is hidden in an obscure part of the site. Just one of many reasons why Reddit is better than Digg.

]]>