I just submitted an idea to the Wikimedia Idea Lab on [Using Wikipedia paragraph improvements to teach literacy.]1 The basic idea is to take a paragraph in a Wikipedia article, and use the revision history to extract earlier versions of the same paragraph (assuming any exist). We can assume these iterations of the same paragraph are quality improvements over time, or we can build in other measures (e.g. looking at some WikiTrust indicators seems like a good idea here). Students could then be presented with sets of such paragraphs, and asked to a) order them in terms of quality, and b) identify the features which are related to such quality improvements. I like this as a teaching exercise. I used to give students sets of exam responses and ask them to do the same thing (of course controlling so length was not the best predictor of ‘quality’!), and it was a good way to start discussion, and do something a bit active while very practically focussed. It also highlighted particular issues (e.g. thinking that lots of ideas was the most important thing, with no critique). In this Wikipedia example, another possibility is that student feature identification could be used to support our own assessments of article qualities (students might identify particular features which we could then use to automatically assess the quality of other texts). Get involved on the idea lab talk page, I’ll be writing another couple of these soon.