Rethinking the Freemedia problem

[Max Spevack]

Greg,

I think this a great idea, and a perfect project for some folks who are 
interested in contributing engineering talent to Fedora in an area that 
it's sorely needed.

> 3. THE REQUEST ENGINE.
>
> Basically, it goes through and matches volunteers to requests -- at a rate
> that precisely matches the volunteer's desire to help.  If someone can
> send one DVD / set of CDs a week, that's exactly what we ask them to send.
>
> 4. THE BACKLOG.
>
> We show the backlog on a webpage, and given the number of requests we
> have, we estimate a delivery time for every single CD/DVD request we have
> in the queue.

The selection algorithm actually allows us to make some very interesting 
choices.  You want to optimize the throughput based on the number of 
people available to burn media, but we could also choose whose requests 
are processed not only based on order in which they were requested, but 
based on minimizing geographical distance between the requestor and the 
providor, or frequency of request, etc.

Should someone who's requesting media all of the time should be thottled 
compared to someone who's never requested anything and is now making an 
initial request?  I think so.

There's tons of room in this design space for clever thinking.

> Might also be nice to donate such an application to the community at
> large...

For sure.

-- 
Max Spevack <mspevack{at}redhat.com>                 gpg key: mspevack.asc
http://people.redhat.com/mspevack/                  fingerprint: 5A17 CF21