When we originally added bounties, there wasn’t much of a backer income stream, and I wasn’t drawing much from the Open Collective. Bounties were a bit of a bonus for me to get a bit of extra income while working on users’ various pet projects. It also had the advantage of showing interest in a particular feature.
Once I started getting paid by the hour, it made working on bounties far more murky. I wasn’t sure if I should be charging for work done on bounties and award myself the bounty when the work was done. It felt like double-dipping, especially if it was something that didn’t benefit a plurality of users. It also meant that I felt compelled to work on these issues because I didn’t want the liability of unfinished bounty issues plaguing the finances. It meant that I would feel pressure to work on these issues to the detriment of what I felt the project needed.
I began to feel that as more bounty requests came in, I was having to spend non-trivial amounts of time design solutions for things I hadn’t previously prioritised, so even if I wasn’t necessarily working on the bounties themselves, I was still doing stuff that didn’t feel as important as what I wanted to work on.
In the midst of all this, we had people using the Bounty tier in Open Collective, then not indicating what issue to assign it to, or people putting money into things that weren’t even discussed as being eligible. I also had to track the status of the bounties to ensure we had the funds reserved for potential payouts (this is been less of an issue as the balance has grown). It was a headache to maintain and manage, and it was contributing to the overall pressure I was feeling. I made the decision to suspend the bounty system until I could rethink how it could be done.
I think the main problem to solve when it comes to reintroducing a bounty system is handling cases where a bountied issue isn’t or can’t be resolved in a reasonable time. Refunding users for such a bounty is not an easy problem to solve, and I’d rather avoid it altogether.
To that end, I’d prefer a system similar to @Ronnie711 described, where we can allocate a set of bounties for the duration of a release, funded from the general funding pool. This mitigates the above issue since if the issue isn’t successfully resolved, the bounty just goes back into the pool - or can be rolled over to the following release. I imagined getting the backers involved in nominating and voting on features to assign bounties to.
Whatever we do, the process needs to be clear and transparent. Ideally the management of it could be delegated to someone, rather than being one more thing I need to juggle.