The Mobile Web team has decided to hide the uploads features (upload & add
to article + upload from left nav) on the mobile site until we have the
time/resources to rebuild them into a more productive contribution stream.
Before wikitext or VE editing, the mobile web team built an
upload-to-Commons pipeline as the first proof-of-concept of mobile
contribution. When we first launched the two upload features (upload and
add to article + upload to Commons via the left nav) ~2 years ago, we saw a
high ratio of these images being deleted (because they were copyright
violations, test images, selfies, etc.). Since then, we've continued to
make incremental improvements to address these issues: we added interactive
tutorials and instruction screens, and also gradually increased the
permission levels required to see these features (from everyone –>
logged-in-only –> autoconfirmed only –> 10+ edits).
However, despite all these changes, the ratio of kept to deleted uploads
has not changed significantly; though the absolute number of uploads per
month has gone down, 70-80% of these files still get deleted.
This is both a crappy experience for the end-user and a major headache for
the team – in addition to the pure engineering effort of continuing to
adjust the parameters of the feature, every incremental change to the
workflow requires a browser test rewrite, analysis time to figure out if
the improvements have actually made a difference, and lengthy
back-and-forth communication with a very unhappy set of Commons admins on
Bugzilla. And none of the changes to date seemed to have made much of a
In trying to address these issues, we've shifted from focusing on the new
user persona to the power user... but we're not explicitly revisiting the
interactions, messaging, or feature set, because we don't have the
bandwidth to make larger changes. I.e., despite the fact that the feature
is now not even being shown to brand-new users, we still show a tutorial
targeted at people who've never contributed before. I think we've reached
diminishing marginal returns on incremental improvements at this point. If
we want uploads to succeed, we need to start from scratch: decide who the
persona we want to target is and then build the set of features that this
user is going to need.
But rethinking how to instruct newbies (since tutorials don't seem to work)
or coming up with a whole new workflow aimed at experienced users isn't
something the team can take on at this point. It requires dedicated product
and design attention and quite a bit of engineering work, none of which we
have the resources for.
Since our focus for the year is new active editors and uploads are not part
of our annual targets, I recommended to the team that we hide the mobile
web uploading features for now and revisit them either later this year or
next fiscal year. The team agreed to this at today's planning meeting, and
we'll be making this change in the coming days.
I know it's not a great feeling when the software we create isn't a rousing
success, but I think it's really important to be upfront (with ourselves as
much as with the community) when we see that a feature just isn't doing
what we want it to do. Lila has been talking a lot lately about how it
seems like we've been trying to do! all! the things! in WMF engineering –
which comes at the cost of fragmenting our attention and making it hard to
really excel at any one of those things. I think she's totally right, and
I'd like to see our team lead by example and strive for more focus and
rigor in terms of what we work on and how :)
As always, if you have questions/concerns, feel free to voice them here.
I'll probably communicate this more broadly sometime early next week.
Product Manager, Wikimedia Foundation