Bounties and the GIMP

An initial approach

In February 2004, the gimp-developer list got an unexpected email from Mark Shuttleworth

Image manipulation is one of the key application areas that needs to be addressed for open source tools to become the mainstream desktop environment. I'm currently funding a number of different open source projects, and am considering funding work on the GIMP or CinePaint.

Among others, Daniel Rogers and I replied off-list to Mark and started a few conversations. A few weeks later, this mail arrived on the list.

More details have come forward about the Mark Shuttleworth offer. Mark Shuttleworth made up his mind and decided to fund myself and Calvin to work on GEGL and GIMP/GEGL integration... Mark has said Calvin and I should divide 30,000 dollars between each other and the milestones.

Our first mistake

We should have said "hold on there" at that stage, but didn't. This sounded very much like any claim to the bounties (which were, at that stage, unspecified) would be encroaching on someone else's territory.

I know that Daniel considered, at that stage, working full-time towards the goal of deep paint in the GIMP, using the bounties as financing.

A month later, Daniel wasn't making any progress, and burned out a bit, and dropped off the GIMP scene. The ideal thing to do would have been to publicise milestones & bounties, and have a free-for-all. Because of the way this was presented to the community via the mailing list, that never happened.

In addition, Daniel became the gatekeeper to Mark - it was his plan, his milestones... we were setting ourselves up to fail at this point. Soemone needed to take charge of the offer of bounties and drive it forward as a community process, open it up, announce it to the world.

But there lay the offer, for about a year.

Failing to pick up the pieces

I contacted Mark at the end of 2004, and again in early 2005, and we talked again at GUADEC in Stuttgart. I also talked about the bounties with Sven Neumann (the project maintainer) several times from mid 2004 (when it became clear that Daniel wasn't going to do the work) through to very recently.

Sven assured me at several stages that he was interested in revising the bounties, was planning on drafting a new set of milestones, based on work which Øyvind Kolås has been doing on a generic graphics library (gegl is dead, long live gggl. Sven clearly didn't have it on the top of his TODO list, and Øyvind was pretty ambivalent about the bounties in any case.

And so we arrive to recently. I got a bit pissed off at being the man in the middle, and told Sven that we needed to announce that there were bounties available, or the offer would disappear. A couple of weeks later, I got a mail from Mark (during a related discussion) where he said (paraphrasal) "The bounties were intended to accelerate progress, and if they're not doing that, they're not working. If there's been no movement in a couple of months, then the bounties will get smaller for every month they're unclaimed".

While Sven was actively working towards the goals of the bounties, and I tried to persuade him that early, incomplete information that shows that things are moving is better than no information, I didn't think the message had been heard.

Mark pulled the bounties at that point, in September 2005, and wrote the following in a mail letting us know.

I am withdrawing the bounty offers. They have not delivered progress in the many months since they were made. Please circulate this to anybody who may have been a contender for those bounties.

I'm open to proposals, but at this stage we have to call the first round "defeat". Alas.

Lessons learned

Øyvind Kolås wrote a good summary of the situation, describing why for certain projects bounties really aren't a good fit.

One thing is certain (the reason I work on gegl and babl) is not to have the opportunity to write milestones and roadmaps for a library I don't intimately know and have to do administrative paperwork or such things, I get enough of those in my day-job.

You can read Øyvind's mail in full here.

I can now see that the GIMP, with its absence of a benevolent dictator for life, and with a pretty ill-defined power structure, is a particularly poor fit for this type of collaboration.

There was no motivation to chase the bounties, and so the bounties were an ineffective way to generate progress and momentum. The bounties were never publicised, and so we never got the young ballsy student with lots of time on his hands coming up with something (probably not very good) that barely worked, but that solved the problem facing the GIMP, so that he could claim his bounty.

I wrote this all up as a warning to other projects considering bounties as a way to get the project moving forward. And to people with some money to invest, wondering how to do it. Perhaps paying for the interested parties to get together for a weekend might be a better investment. Or perhaps you can learn a lesson from our experience, and avoid making the same mistakes we did.


Sven Neumann wrote a response to this as a blog comment:

From the point on that Daniel dissapeared, the bounties have been dead and there was nothing that we could have done about that. Your summary puts it like there would have been an offer on the table all the time. But there wasn't. There was a deal between Mark and Daniel and that deal had failed. It would have been wrong to assume that there was still a valid offer at this point. Without clearly defined milestones, there are no bounties.

We are still in the very same position that we have been over the last years. If we think that bounties are a good idea, we just need to make a detailed plan, set up a number of well-defined goals and proclaim bounties. There is money available that we could use for it and we can still approach Mark or anyone else and ask for financial support. Nothing has changed with respect to that.

I disagree with Sven. Mark never publically said that he had made a deal with Daniel, although certainly people had that impression. Mark's goal was to help the GIMP because it "needs to be addressed for open source tools to become the mainstream desktop environment".

And the offer clearly stayed on the table through to this Summer - over several months, I had kept in touch with Mark, and with Sven, and asked Sven to engage Mark on the issue. A major issue was that that didn't happen - I think an e-mail to Mark back when Daniel dropped out of the project would have saved us a lot of pain, and perhaps moved the GIMP further along in the meantime.

With thanks to Instiki for the stylesheet, and to Mark Shuttleworth and Øyvind Kolås for permission to print private communications