Here is an idea (just an idea). Let says there is a matrix of what needs to be implemented to federate features between forges. A cell of this matrix could be, for instance, Gitea issues can be followed via ActivityPub and it would be linked to a bounty where people can add pledges. Something similar to bountysource (but I would not recommend them right now because they changed the way they handle the funds in a confusing way in the not so distant past).
A test would be associated with the issue to verify it actually works. Before its implementation the test would fail. And after the implementation the test would succeed. The author(s) of the commits that made it work are then eligible to collect the pledged funds, via liberapay or something similar.
The tricky part is: would that work in a horizontal way? It might since the individual people who made the pledge on a given feature (cell in the matrix) will presumably verify that it indeed works before releasing the money. The key to a successful experiment probably is to gracefully handle fraud, when and if it happens.
To be continued (maybe)