[Vorbis] Bitrate Peeling (no. really)

[Ralph Giles]

On Thu, Sep 30, 2004 at 07:46:18PM -0400, ben-extra at MIT.EDU wrote:

> 1) What is the project?  Specifically, a list of criteria that must be
>    met before the bounty is claimed.  I personally hope this will
>    include that it works with existing .ogg files.

If you don't think the descriptions at http://wiki.xiph.org/Bounties are 
clear enough, please add to the page.

> 2) Who is the judge(s) that will decide (1)

As I said, the core xiph.org members would decide if the conditions had 
been met.

> 3) An estimated start and finish date.

I was thinking less of a 'street performer protocol' approach than a 
traditional bounty arrangement where the money accumulates until it's 
motivating enough for someone to put forth the work to claim it. This 
discussion of bitrate peeling has sort of assumed there's someone who 
could do it just waiting for funding. That's not really the case (more, 
the numbers we suggested is more what we thought we could convince 
someone to do it for) and definitely not the case for the other 
bounties. They're projects that need doing we've not been successful in 
attracting the volunteer labor to complete.

As such, I don't think timelines and reports are a good idea. That makes 
sense if we grant bounties 'a priori' as in 'we contract with you to do 
this work for this amount' but I think the market effects are better 
with the paid-on-first-delivery approach. Maybe I'm wrong, most of my 
experience has been with bug bounties, where the tasks are a little 
smaller.

> 5) Procedures for the money to be returned if the project isn't
>    completed by a certain date.

As I said, we could do this for people who so request, but I'd rather 
the money just revert to the foundation so it can be used to support 
other work. Keep in mind the contribution won't be tax deductible if 
it's refundable. I think the decision to take down the bounty should 
also be up to the xiph org board, rather than the calendar. They may
stand for years and eventually become irrelevent, but that's not 
something one can determine ahead of time.

 -r