Great stuff.
The biggest challenge I see is ensuring quality. Even if there's funding for research, what good if the model leads to bad research being done? Not that most papers being published currently are all that useful... but still, we would want to increase the overall quality of the science being done, not decrease it.
A possible model could be something like this: a researcher submits their proposed experimental design prior to conducting any experiment. Fellow researchers (which will act as peer review AND curation groups) tear apart the experimental design and suggest changes until everything they can think of looks alright. Then they upvote, which provides the funding for conducting the experiment or study. After the experiment is completed, the final paper with the results is published. As a bonus, the data could be uploaded to some open data database where other researchers can access it.
Sorry for not trashing your idea as you requested. Try harder to come up with a worse idea next time and I'll do my best.
For quality, the ideal concept would simply be the curators, the self policing from a democratically voted group of quality checkers - an advanced steemSTEM. The problem then however would be more emphasis on objectivity, since we currently vote on objectivity first, then subjectivity based on creativity, style etc. A blog could succumb to subjectivity but the paper itself should have some objective minimum payment. I guess...
I would put a stepp between this end and include a post about a progress report to show they're adhering to the methodology agreed upon. This is a good adaptation of my idea, making it more present tense, whereas mine was more trying to reward already completed products.
It brings to light a whole new wave of ideas!
Now this I can do =P
There is a required minimum budget for each study, right? It doesn't help to have only half of it, or only 80%. 100% of the needed budget has to be there in order for the study to even begin, otherwise there probably won't be a point in starting it and not being able to finish it. Unless there is some other funding source, that is. If there is, the researcher can apply to the STEEM curation group and ask for, say, 50% of the needed funding, and specify where the other 50% will come from. That's a totally viable model, too. The point is, though, that the curators either have to fund the entire amount asked for, or nothing. The research cannot proceed if there's only a little bit of the needed money. That's how I'm thinking of it. Although I'm sure more flexible/sophisticated funding models can be developed, but probably at a later stage.
For the funding criteria, an approach can be developed where the proposed studies that have the highest benefit to society (or to the research field) and the lowest budget cost will get the highest funding priority. To put it more simply, the greatest benefit and lowest cost is what you probably want to fund first.
For this to work, the estimated benefit of a proposed piece of research has to be quantified. The way you do it can even be something totally stupid as asking the researchers in the field to vote from 1 to 10 on how useful this research will be in their view. So you get some estimate of the benefit of the study - of course, better ways to do it can be developed.
The other part you need is the budget, which is already quantified, so no trouble there.
Then you prioritize using a formula, for example: estimated benefit / budget. Through this, you will get a single number for each proposed study, and you prioritize by that number. The available funding will go to the proposed studies at the top of the prioritized list who have passed through the peer review process.
Would be happy to hear your thoughts. And yeah, having a progress report definitely sounds like a useful additional step.