Subject: Re: Bug Bounties. Making $ from bugzilla.
From: Brian Behlendorf <brian@collab.net>
Date: Mon, 26 Nov 2001 20:47:00 -0800 (PST)

On 25 Nov 2001, Kevin A. Burton wrote:
> Yes.  I talked to Brian (Behlendorf) about this.  He was very down on the
> idea.  This was basically what I expected from someone who just had his dream
> crushed.

Well, SourceXchange wasn't so much a "dream crushed" as an experiment that
didn't quite turn out, but seems in retrospect like it was worth trying.

The main difficulties were:

a) Being a "trusted third party", especially one that tries to keep the
participants honest by sending the bill and cutting the checks, is a lot
more work than it was worth; even with a 20% fee.

b) We did have a system with lots of checks and balances to try and
address concerns about the process - having a peer reviewer on every
project, for example.  That added lots of overhead and delay to the
process.

c) Coming up with a proposal, with milestones and schedule, is very
difficult unless that's your core business and area of expertise.  Cygnus
was such an expert with GNU compiler tools they could get this down to a
science, but most developers just aren't there.  Instead, the usual is to
work on a contract basis for a certain $$/hr with a rough guess to the
amount of work to be done, and occasional review.  SXC, by contrast, was
pretty much all about fixed-fee contracts; that was a risk the developers
would have to take, and worked against them.

d) Selling the concept was hard - we tried to describe it as Ebay for open
source software development, but then get dinged for such a high fee.
Selling a process is hard to do to those who have never been bitten by the
lack of one.  Those who were willing to shell out a bunch of money for
software development would rather have worked with a domain-specific
outfit, like Cygnus, or wanted to think we were since we played such a
strong trusted third-party role.

There's no one to blame for this - all the developers and sponsors were
great participants and willing to give it an honest try, and under ideal
conditions the process worked.  We did distribute a couple hundred
thousand dollars to open source developers to write true open source
software.  All of the above issues could probably have been addressed with
sufficient time, funding, and market opportinity.  We looked at this after
being up for a year and saw several other sites doing similar things,
without the restriction of being only about open source software
development (interest in proprietary development made up the majority of
inquiries to us), in many cases better funded, usually with less process,
and there just wasn't the amount of interest we thought would be needed to
make it worthwhile.  To get $10M in revenue a year (what would be needed
to break-even a company of 50 or so, within a factor of 2) we'd have to
broker $50M in jobs, and even getting $20K jobs was a couple weeks' work
for a couple people.  Especially since we were constantly trying to
explain to people why it was worth spending their money to create software
their competitors could also use for free (sigh).  We probably could have
downshifted to being only a matching service, but even asking for a 1% fee
at that point would be awkward and wouldn't have supported more than one
or two people.  Our other interests were picking up at that time, so we
dropped it instead of exploring that option.

If someone wants to give this a try again, start by being just a matching
service, something that *can* be done by one or two people and doesn't
have high overhead or complexity.  Let the developer and sponsor work
out terms between each other, perhaps provide them some templates but
don't mandate anything except your listing fee.  Then go from there.

	Brian