Page MenuHomePhabricator

[] Establish a set of criteria for blacklisting most-read articles likely representing bot traffic
Closed, DeclinedPublic

Description

It's surprisingly often the case that one or more relatively un-notable article will persistently appear on the "trending" (most-read) feed card. Currently, an article of modest interest, AMGTV[1], has been in the top 5 for some time for English Wikipedia.

Earlier, we blacklisted two other relatively un-noteworthy articles on TV stations as likely inflated by bot traffic: https://gerrit.wikimedia.org/r/302720

From the outset, we have also incorporated the original blacklist in use in the iOS app from its original feed prototype. (I'll update with a link.)

We shouldn't cross the line into censorship, of course, but the feed's success depends on its surfacing content that's actually interesting, which articles that just happen to get hit by a bot a lot of times typically aren't. To be more objective in these decisions, let's establish a set of criteria for blacklisting rather than doing so on an arbitrary, case-by-case basis.

[1] https://en.wikipedia.org/wiki/AMGTV

Event Timeline

I thought this was discarded due to a fat-fingered backspace press but it looks like it was actually created. Anyway, it's a duplicate now (see T143990: [Feed] Establish criteria for blacklisting likely bot-inflated most-read articles).