Grants talk:IdeaLab/Improving paragraphs - teaching literacy, improving Wikipedia

From Meta, a Wikimedia project coordination wiki

Ranking the text quality[edit]

How will you tell the difference between improved text, and text that has just had extra content added to it? --EdSaperia (talk) 16:14, 19 April 2014 (UTC)[reply]

That's where something like WikiTrust, or possibly some of the metrics discussion for WMUK WikiRate would come in. WikiTrust would probably be good because timeliness doesn't matter in this instance (we could just pick lots of very old examples in fact). Looking for stable edits would be a good idea, and a minimum diff (so it might be that not every rev is presented). Interesting issue with WikiTrust is how you average 'trustworthiness' of edits across a paragraph to reflect the over quality of that paragraph (I haven't thought about this much yet). Sjgknight (talk) 16:24, 19 April 2014 (UTC)[reply]
Interesting, I hadn't heard of Wikitrust. I still think you'll have difficult differentiating between copyedits and new material, though; and in this mechanism, there's nothing to say that this will fail gracefully, nor do you really need a piece of software that generates more examples than could be done by hand. As currently described, it seems like this task is something that humans will be more effective at than computers. EdSaperia (talk) 17:10, 19 April 2014 (UTC)[reply]
Re human ratings - agreed, absolutely. And that might be a good way to start proof of concept. But, to be clear adding new content to a paragraph isn't a big issue re: task design unless it changes the paragraph so much it's unrecognisable. So the obvious thing to do would be work back from the most recent version (not necessarily the 'best'), but not necessarily back to the first version. It might still not work though, but there's something interesting about thinking about a) teaching possibilities, and b) using the technical element and corpus size to support that (even if it's not on this exact idea!). Sjgknight (talk) 20:37, 19 April 2014 (UTC)[reply]

Related ideas[edit]

If anyone knows any related ideas do add them here by adding a bullet and signing, e.g.

A few thoughts[edit]

Hi Simon! I really like this idea! Getting to the asking-students-to-edit part: Are you thinking the students would be working more at deep copy editing of these paragraphs, or actually finding sources/adding references/adding information? I think certainly some of the copy editing aspects could be good on a paragraph-by-paragraph level, but I'd have concerns about anything more than that.

We've seen assignments that are just focused on small bits of articles tend to be the least successful; when students don't have to read the paragraph in the context of the whole article, they just don't understand what's useful for that paragraph. And honestly, most don't take the time to understand what a reliable source is according to WP:RS policy unless they have to make a significant contribution to the article. What ends up happening for paragraph-level contributions is students google the subject of that paragraph, click the first link that comes up, add a quote from that source, and call it done. Most of this work ends up getting reverted because of sourcing issues or that particular fact is already in a different paragraph of the article the student just didn't bother to read. And then angry editors who spend a few hours reverting all work from that class are (rightly) upset. So we discourage these short assignments now because of that. I think you'd have to invest a lot of time up-front teaching what reliable sources are and how to improve a paragraph within its existing context -- but I'd also say these are clearly a skill students need to have! I'm excited to see where this develops. --LiAnna (Wiki Ed) (talk) 15:55, 21 April 2014 (UTC)[reply]

Thanks LiAnna, so the point here is that people would work primarily with existing paragraphs (and their revision histories). So the aim would be to get them to understand how/why paragraph rev-1 is 'better' than rev-2 or rev-n. Only then might they be asked whether they could improve it. Hopefully, having explored the ways paragraphs are improved, their understanding of how to improve on paragraphs would be better. I also wouldn't make that bit a mandatory part of the 'game' (although you can imagine some sort of points system to gamify based both on successfully spotting features and organising revs, and on actually editing to improve the paragraphs). In the first instance I'd say these wouldn't be inserted directly on the article, the initial element could be used as a way of seeing if this approach looks promising and then looking at how to integrate with actual article editing. Sjgknight (talk) 16:19, 21 April 2014 (UTC)[reply]

Wikipedia adventure[edit]

Obvious thought, this sort of thing could relate to games like The Wikipedia Adventure @Ocaasi: Sjgknight (talk) 14:34, 11 May 2014 (UTC)[reply]