It should be noted that the software running Data.gov itself has not been made open source. This is a project of the American and Indian governments to produce an open-source version of Data.gov to be shared with other governments.
Here's an open-source government data catalog originally made for the City of Philadelphia, built on Django:
If the data is available for download (preferably via ftp) what needs to be "open source"?
Personally I do not care what software they use to run the site so long as the data is available for download in a open format like CSV. Compare this with offering the data in little bits via some silly JSON API or offering it in some format that requires some addtional closed source software to process.
To me, the data is what is important. What software they choose to use is not important, as long as the data is easily accesible (ftp is my preference). What's important is that I can use whatever software I want to process the data.
Yep. They open sourced the equivalent of their marketing site, not an actual data platform. Very disappointing. As another poster mentioned, CKAN is open source, and does include the data hosting platform (based on elasticsearch). Max Ogden has also published two projects that seem relevant: datacouch and (I think) the pdx API for Portland, which are both couchdb based. All three are worth checking out.
It's been awhile since I've read through Drupal projects...but how much of this is usuable, extendable modules and how much of it is "Create a database and run the Drupal init script on it"?
I'm really excited about the possibilities of this. The exact implementation might have it's various flaws but it's an exciting move in the right direction.
Here's an open-source government data catalog originally made for the City of Philadelphia, built on Django:
http://civiccommons.org/apps/open-data-catalog