Page MenuHomePhabricator

Article Feedback - Default Warning Message for Abuse Filter
Closed, ResolvedPublic

Description

This ticket is dependent on Abuse Filter Bug 37561 - Make it possible to have differing default warning messages.

Andrew Garrett posted a revision for that feature on Gerrit:
https://gerrit.wikimedia.org/r/#/c/10394/

But it didn't pass code review, there are 2 minor issues that need fixing.

Once that ticket has been fixed and deployed, we will need to test that it works for Article Feedback, and make any modifications that may be needed to make sure the proper message is displayed for feedback warnings.


Version: unspecified
Severity: major
See Also:
https://bugzilla.wikimedia.org/show_bug.cgi?id=37561

Details

Reference
bz37579

Event Timeline

bzimport raised the priority of this task from to Unbreak Now!.Nov 22 2014, 12:22 AM
bzimport set Reference to bz37579.

Thanks, Matthias, well done!

Andrew, would you like to confirm that this revision works for you, before we deploy it to production?

We pushed this feature to the English Wikipedia in July, but it was not used for several months, because no default warning message was created for it.

To address this issue, I posted a new warning message (with the language originally approved by WMF's legal team), at this address:
http://en.wikipedia.org/wiki/MediaWiki:Abusefilter-warning-feedback

We briefly enabled it on this filter to warn users about posting email addresses:
http://en.wikipedia.org/wiki/Special:AbuseFilter/463

But the action for that filter was later changed to disallow.

You can read more about abuse filter for article feedback on our feature requirements page: http://www.mediawiki.org/wiki/Article_feedback/Version_5/Feature_Requirements#Abuse.2FSpam_Filters

As we approach full deployment for AFT5, we would like to soften the warning message that is now displayed to users if their feedback is rejected by the abuse filter on AFT.

Some team members have expressed concerns that it is a bit too harsh in its current form, and may be a turn-off for the very users we are seeking to engage.

Others have also pointed out that it is often not clear why your feedback is being disallowed, suggesting that we give users a clue by citing the name of the filter which disallowed the post.

To that end, we propose this kinder warning message, with the name of the filter in parenthesis:

"Your post has been rejected by a software filter that suggests it may not meet Wikipedia's <feedback guidelines>. Please revise your post and try again. (Filter: Common vandalism 1)"

Instead of:

"Your post has been rejected by a software filter that suggests it may have violated Wikipedia's feedback guidelines. Please revise your post and try again."

The reason we are advocating this gentler language is that sometimes legitimate words may be disallowed unfairly. For example, if someone writes the word like 'rape', it will be disallowed, even it is being used for legitimate reasons.

When citing the filter name, we propose adding the word "Filter: " for clarity (replacing the text string "Feedback: " if it exists at the start of the title, to keep it short), if that's easy to do.

We currently disallow several hundred posts per day with AFT5 at 10% -- so this message would appear a thousand times per day once we deploy to 100% on the English Wikipedia. This represents about 5-10% of total feedback posted per day.

If you would like to see what kinds of words are being disallowed, we are tracking the 7 abuse filters that are now disabled on this Google spreadsheet:

https://docs.google.com/a/wikimedia.org/spreadsheet/ccc?key=0AiGAdIp7VYlbdDdKUm9naXhxOXVweWZ5YkU3Wk5lSlE#gid=0

This has been highest priority for four months now without getting fixed or seeing any progress, so I have doubts that this is an urgent item.
Please reconsider and reduce the number of open "highest priority" tickets for AFTv5. See http://www.mediawiki.org/wiki/Bugzilla/Fields#Priority

Sorry - should've closed this one already.
Code is at https://gerrit.wikimedia.org/r/#/c/32208/ and has already been deployed.