I'm running two courses this term, and in both, I'm asking students to make up their own grading schemes.
The first course (CSC494 for undergrads, CSC2125 for grad students) is the consulting course I've run for the past few years. Its aim is to teach students how to work as consultants for real clients, and a big part of that kind of work is scoping out what the client actually wants, and how much of it can be built with the time and resources available. I found in the past that if I asked students to give me an estimate, what I got was mostly science fiction :-) If I turn it around and ask them to tell me, "How do you want me to grade your work during the term?", they put a lot more effort into figuring out what they're going to deliver when.
There's a lot of flexibility in what students can submit as a grading scheme. In some cases, there's a fixed endpoint and very clearly defined subgoals, so they make up a schedule that spaces those subgoals out at two or three week intervals, and we're done. In other cases (particularly with grad students), the work is much more exploratory, so what I get in the first instance says, "Here are the first two deliverables, worth X% of the course, and here's the date by which I'll be able to tell you what else I'm going to do and how I want it graded."
In both cases, we'll revisit the grading scheme in early March (when students are more than halfway through the project), and give them a chance to re-evaluate. In most cases, "re-revaluate" means "scale down expectations", but again, the real point is to figure out why the January plan and the March plan are different: what questions could the students have asked in January to create a more realistic plan, or what information did they need that they couldn't possibly have had, and how should they have gone about getting it.
Long story short, students are responsible for putting this together, and doing so forces them to have some real conversations with their clients early in the project. If in Week 2 a student is asking lots of specific questions, we're probably OK; if a client hasn't heard from him or her yet, or if the questions are still, "Um, where do I find an installer for this?" then I'll probably bounce their grading scheme back to them with some sour comments and a low grade :-).
My other course is "CSC301: Introduction to Software Engineering". This term, instead of posting lecture slides online, I'm picking one of the four-person teams in the course at the start of each lecture, and having them write a wiki page about the lecture as an assignment. This is worth 5% of their course grade, and they have to do it twice during the term; once they post it, other students are encouraged to ask questions, make comments, etc., up until one week after the lecture (which is when I give the page a grade---setting a deadline encourages the team to make their page earlier rather than later).
In previous terms, I asked students to make up a grading scheme for Assignment N+1 as part of Assignment N; the idea was to teach them how to write specs (and to give them some appreciation for how hard it is to do that well). This term, I'm having teams write assignment specs instead of doing a lecture page (also for 5% of their course grade). We talk through the core ideas in class, and after that, it's down to a bulletin board discussion thread and a wiki page. It seems to be working pretty well so far: students are asking more specific questions earlier than they ever have in any of my previous courses, and (I hope) are feeling more engaged in their education. I'll be very interested to see what they say in the end-of-course feedback.