Page MenuHomePhabricator

Top read repeats
Open, MediumPublic

Description

For a while the following are almost always in the top read listing in English Wikipedia, plus other languages, on the iOS/iPadOS native app. One can see how some of the items may indeed have reliable reader demand and may be true positives, but would it be possible to remove any sources of bot / CSRF / misconfigured requestors to ensure an accurate tally for most read reflected in the app? It seems likely data should be addressed upstream instead of by override (excepting perhaps "XXXX", which may be a natural traffic pattern change if its three character sibling is being pushed downward).

  • "Microsoft Office"
  • "F5 Networks"
  • "Google logo"
  • "XXXX"
  • "Cleopatra"
  • "Bible"

Given in particular the historical and present significance of the final two items in the list a careful inspection seems in order.

Event Timeline

@JAllemandou sorry I missed your comment last week! Is this something that still needs looking into? Tagging with Product-Analytics so we can discuss this in our next board triage.

T270784 suggests that Cleopatra "is a default voice search test in spanish for hello google." I'm not sure where that info comes from; I'll reach out to Partnerships.

@kzimmerman my partner set up her voice search for her new android phone -- I suspect it has all of the vowels that are common in spanish.

Thanks @Astinson !

@JAllemandou I raised the question of defaults (like "Cleopatra") with Partnerships and @Nicholas_Perry said he's going to see if he can get more info from our partners.

Moving this to the backlog, though @cchen will keep this issue in mind as she looks at our top-level metrics.

Thanks @kzimmerman for the heads up :)
On our side we don't forget the improvement of heuristics.