One day, the animals will run the zoo

Grading must really be one of the worst parts of teaching. Although we judge each others' work, and perhaps even each other, almost constantly, voicing these judgments is uncomfortable. It's especially uncomfortable when we're asked to reduce the whole of our interactions to a number score or letter.
The common thread among these approaches to evaluation is that they are ways of informing students about the expectations for their coursework. For example, community grading seems to not only depend entirely on the pupils for the criteria of grading, but it actually encourages them to challenge these self-made criteria. For a traditionalist, this might seem like letting the animals run the zoo, but the fact of our mortality means that these animals will be the zookeepers whether we like it or not, so we might as well show them how to work the gates. In another sense, this approach threatens to seem like the instructor is passing the buck--taking one of the most difficult parts of teaching and handing it over as the students' responsibility. But as long as the students are being controlled on some level, then they will be more interested in surpassing their own criteria.
Similarly, contract grading can give students a better idea of the process that often goes on inside the instructor's head when giving grades. In this way, they have a sense of power that, while is somewhat overblown, can pave the way to success.
Finally, the idea of a good-enough evaluation system can help students understand how their work will be evaluated. Students should not feel unnecessary stress about a one-page homework assignment. That only makes them learn to the grade. When they are given a better idea of what's important, they better understand the other side of the desk where the grader sits.
My primary criticism of these grading approaches is that they tend to focus on mid-level students more than those at the highest or lowest levels. For example, some students might be interested in being graded on their instructor's expertise more than on the half-learned approaches of their classmates. And while failing students might not so easily demonize their instructors in a community grading system, they now face even further separation from their peers. We cannot pretend that classrooms will ever exist without excellent or subpar students, and our grading systems must account for these differences.
Similarly, a contract set out for a B sends a confusing message about an A, and a failing student can easily question the legitimacy of a contract. When it manages too much on process, students are asked to perform in a uniform manner. It's critical that contracts are only tied to what the final grade would suggest--that is, a B in English 1000 ought to accurately suggest proficiency in writing, not an ability to adhere to contract clauses involving attendance, conferencing and peer evaluation.
Most importantly, students must learn through grading what constitutes 'goodness.' Perhaps they write it themselves, perhaps they sign on to a managed process. But what's clear is that the process of grading should be a more explicitly discussed process in schools from an early age.