Tips for Snuggle improvement
I have the following
two five suggestions for snuggle -
- Articles and article talk are in green. i suggest changing the colour so that there is no confusion between article green and good faith green.
- There ought to be a colour code/legend symbolizing whats what, possibly at top of the page.
- A quick one-line description along with the prototype will do good. We need to know WHAT exactly I can do, or else I might find myself clicking on lots of buttons only to later find out it was only for classifying editors
- the current version shows only the changes done by the editor, which may be confusing. Having a line before and after that line under question will do good to provide a quick context as to whether adding a link to Osama bin Laden is actually justified, as might be for an Al Qaeda article, or not, for one on a school in UK.
- There must be a quick way of sending in those welcome templates etc from the interface. I would want to be able to send the template too, rather than only classify the editor as good or bad.
- Thanks! I just saw these pushed into the issue tracker. I've split them up into individual issues. Note that one is on hold since I'm not sure what you need.
- --EpochFail(talk|work) 15:47, 18 December 2012 (UTC)
Just had a quick go on the prototype. I think its awesome. My thoughts on TheOriginalSoni's suggestions
1. Happy to see greater differntiation between colours but I do like the fact that the colours are currently quite subdued.
4. I think Soni chose a bad example. However, the line above and below the edit made would provide some useful context.
- The screenshot helps a lot. I struggled a bit with figuring out the right way to show context. The default diff viewer in MediaWiki shows an awful lot of context and I don't have much space in the little diff-viewer widget of snuggle. I could make some changes to see more context though. I'm in the middle of a big code-rewrite to start testing the authentication integration with MediaWiki. As soon as I've gotten that in a good state, I'll update the diff viewer too and ping you with a link so you can give it a test run. Given how close we are to the winter holidays, it might take a week or so for me to get there. --EpochFail(talk|work) 14:32, 20 December 2012 (UTC)
There are some alternative ways of showing diffs at:
I have been using the prototype and have found it mostly easy to use and very impressive that you have managed to extract and present so much information in a fairly easy-to-understand way.
I'm sure I will have more questions at some point, but here are four:
- What is the intention with classifying the edits into good and bad faith?
- Do you think it would be more helpful to allow users to give, say, a 5-star rating, 0 being bad-faith and 5 being good-faith? For example, most vandals would be 0, someone who is pushing a POV too much may be 1 or 2, depending on how spammy they get.
- Do you think it would also help to have a separate rating on constructiveness? Sometimes someone seems to be well intentioned but manages to mess things up, as illustrated by the phenomenon of the good-faith revert. These are the sort of people that might need a bit the most coaching.
- If people are going to be rating other users, you may need to make it only accessible to, say, reviewers. We certainly don't want vandals rating themselves as good-faith!
- I'm hoping that, by providing a manually curated list of good-faith newcomers, mentors at the Teahouse and elsewhere will be able to more efficiently direct their efforts toward newcomers who are worth their time. Using snuggle, I could construct a notification system that sends out a ping when a "good-faith" newcomer is sent a new warning template.
- That's a good question. One of the reasons I'm reaching out for feedback is requests like this. Do you have a specific use in mind for a more nuanced quality evaluation -- or maybe a set of cases where the boolean (good/bad) classification is insufficient.
- See point above. This is a good point though. What sort of constructiveness scale would you imagine. I'm thinking something like:
- No productive contributions
- At least some contributions would be productive with some editing
- Mostly productive contribution
- I think this is a good point. I'm building the authentication system that will tie Snuggle to Wikipedia user accounts right now. I'm hoping to use this to allow Snuggle users to interact with Wikipedia (e.g. post Welcome messages) and track each others ratings of newcomers. Using this system, I could also limit ratings to users on a whitelist or with a specific wiki right enabled.
- Thanks. This helped me to think about how you want Snuggle to be used. I have had lots of ideas which I have set out in sections below. You may find some more useful than others and some more urgent than others but I thought I should get it all out there.
- I am now inclined to thinking that we really just need one scale with three categories: good-faith, ambiguous and bad-faith.
- This should be combined with a system that makes it fairly easy for Snuggle users to spot when newcomers have been misclassified and make the change in classification.
- Ambiguous would be different from unsorted in that someone has made the positive statement that the newcomer is ambiguous. Also a Snuggle user can reclassify the newcomer as ambiguous – they can't currently reclassify a newcomer as unsorted.
- I am currently thinking that the constructiveness would be hard to separate in most cases. Maybe people who get a lot of good-faith reverts would end up in the ambiguous category... That may be all that can usefully be achieved with a big-data approach like this. Snuggle users may decide to offer lots of help to particular newcomers but maybe a constructiveness scale doesn't improve the ability to do that.
- I think the most important thing is what you do in terms of filtering and ordering the newcomers for display in Snuggle, which I'll discuss in a new section because it is a massive topic.