Bruji

I just started trying to do some moderation on the available data and have some initial feedback. For clarity, this is only related to book data.

I've got an issue with how the Doghouse moderation is supposed to work; I'm seeing a flaw in the plan. In short: if I edit, I approve. But there are times when I should be able to edit and not approve.

My understanding of work flow from the the help + Moderator Guide is thus:
* A moderator will only be provided entries that match his client's Language Locale (in my case, 'en')
* If you see bad data, correct it.
* If you aren't sure if the data is correct, leave it alone for somebody else.

I'm noticing an interesting trend in the data: a lot of entries listed as LanguageLocale of 'en', aren't correct. In fact I'm looking at a lot of them right now that are French. Now, I'm smart enough to identify a few languages by sight and I'm pretty sure I can change the language settings to the correct language code, however, I'm hesitant to do that. I don't want to 'approve' the entry, just correct the Lang code. Going back to my French example, sure the fields look like they are filled in correctly, but since I don't speak French I have no idea what the data actually says!

So, now I'm in a quandary: If I don't make this change, then this book entry will keep popping up to be Moderated by the wrong people (people who don't speak what ever language), and by the nature of, 'if you aren't sure, leave it for somebody else', potentially this means the entry could float around for a long time before somebody out there speaks whatever language of the entry and is able to properly Moderate it. But, if I was able to change the entry from 'en' to, let's say, 'fr', and NOT approve it then the entry will start to appear under those Bookpedia users who have 'fr' as their native language settings and then they can easily moderate the data correctly.

Let's throw another twist in here. If Moderators are able to make edits that do NOT result in a approval, now entries can float around and get edited many times and possibly never get approved. That doesn't help either. Here's my recommendation. Give Moderators the ability to make non-approving edits to the data. At the same time, keep track of the number of non-approval edits have been made to a given record. Then do one, or both, of the following:

* First when feeding unapproved data to moderators weight your data selection such that the Moderator's data set includes at least a few of the records with higher edited content. This insures that these 'edited, but not approved' records are seen by more eyeballs than a regular, yet-to-be approved record. (I'm assuming you're doing some type of random distribution right now, but really don't know.) The idea is that if people make edits but not approval for a given record, there's obviously an issue with it and we should pay a bit more attention to it.

* Add an other Moderator collection to the pedia clients, one that just highlights the 'edited but not approved' items. This would let us to quickly identify those records which may need a closer scrutiny.

And one final comment. I think there might be times Moderators might want to discuss what to do with a given record or make comments on it. It might not be a bad idea to think about that. Maybe we need a Moderator forum section where we can discuss the gorey details of some record. Or possibly the ability to add a note or two to a record, (i.e. 'Changed language code from 'en' to 'fr' - please verify').

Thx,
Gopher.

(Yeah, I'm probably thinking about this stuff too much. ;)

Thank you for the thoughtful analysis. It made so much sense that I was pretty sure that we had implemented that on a locale language change the entry would remain as unapproved, so that it still gets seen by two native speakers of the language. But you are correct that at the moment it's moving into the second approval collection. I have let Alex know about the issue and he will update it for the next version of the Doghouse so that localeLanguage changes are free of automatic approvals. This way you can still make edits even in French, but because the locale language changes it will be seen by two other french users as well.

I like the weighted idea, but the two pass approval process has been working extremely well. Entries that have passed through the full moderator queue have been consistently in good condition. The discussion section is also a great idea and something we had considered as an extension for a future version. We are currently working on a more comprehensive moderator options to be available online and the discussion section could be included here for users to check. The Pedias interface have their limitations when it comes to moderating, but also the advantage of laser like focus on viewing the essential and simplifying the moderating task.

I have updated the Doghouse so that only entries that match languages will get the moderated token. Thanks for your suggestion.