Four crowdsourcing lessons from the Guardian’s expenses-scandal experiment

As a programmer who works at a newspaper, I must say that this is totally wicked.

For minimal investment you get a lot of people doing work for you - for free - and a huge PR boost. Plus a bunch of nerd cred.

This is one of the more brilliant examples of a company getting their shit together to do something awesome on VERY short notice.

The Guardian did an awesome job, and kudos to the developer, Simon Willison.

Interesting article (for an American who didn't even know about the UK scandal).

How much of a coincidence is it that the main Guardian developer is also one of the co-authors of Django? Article reads as a very nice endorsement of Django.

I'm impressed by the speed and frugal resources with which they implemented the site and the project has obviously been a huge success, but it is not true that it all had to be completed on that timescale. The release of the MP's expenses has been expected for months, if not years. A delay to the launch of the site would have invoked the 'isearch' law of project management:

"The worst delay to the completion of any project is the management decision to start the project."

Impressive story, despite the Django and EC2 plugs.

Of course this suggests a start-up project: a business service that does crowdsourcing. The service would be customized for one task and would have various forms of cheat detection (e.g., CAPTCHAs, user profiling, meta-moderation and maybe some secret sauce).

Mechanical Turk is fine in many cases, but this story shows sometimes it's better to roll your own (or have somebody else roll it for you).