[Wikimedia-l] Feedback for the Wikimedia Foundation

[Erik Moeller]

On Tue, Jul 23, 2013 at 6:16 AM, Craig Franklin
<cfranklin at halonetwork.net> wrote:

> I just want to basically endorse some of the other comments being made
> here, which I think are quite insightful.  If the goal of this project was
> to get the Visual Editor deployed on time and on budget, then the goal has
> been achieved.  But if the goal was to gain acceptance from the community,
> then I think that the polls on enwiki and nlwiki show that it has been
> quite a failure.

At no point did we expect that VisualEditor would see significant
adoption among experienced editors initially. It's very clear that
initially, for a user experienced with markup, a completely new
editing tool that sometimes introduces new types of errors (either
inevitably because it's a different editing mode, or avoidably due to
bugs or UX issues) will in most cases represent primarily a new
cognitive burden and not increased ease. It's to the credit of our
community that we nonetheless see strong support for developing a
VisualEditor even from users who are unlikely to use it in the near
future.

I'm hearing some of those users say "It's easier for me right now to
do simple copyedits", and I've seen some examples of users who weren't
active come back and say "I'm editing again now because I was never
able to figure out markup". But for users who are very comfortable
with the current mode of editing, it will take us a long time to both
provide a consistently superior experience (which I do think is
possible, but very hard) and to win them over. It will take users some
time, as well, to discover features in VisualEditor that make their
lives easier (keyboard shortcuts, templatedata, default parameters,
etc.).

We did expect higher uptake among new users. It is higher (currently
about 35-40% for post-July 1 account holders), but not as high as we'd
like. That number will go up a fair bit as we add IE support, we're
estimating to ~50%. As you interpret data, please keep in mind that
it's hard to accurately establish whether a user is new based on
registration date alone. For example, we don't know how many of these
users have prior casual IP editing experience (prior survey results
indicate that about 60% of active editors do), or experience editing
on non-WMF wikis, or even as an account on en.wp (sock puppet,
forgotten password, etc.).

We don't know at this point what the quantitative impact on new user
productivity is. The initial A/B test we ran prior to release is still
being analyzed, and there were many confounding variables (including
the fact that some users who were assigned to VE shouldn't have been).
Please wait for Aaron and Dario to complete this work before drawing
conclusions from it.

That said, based on what I know, the single factor that likely has the
most negative impact on new users is performance on long articles,
which is still quite poor. If you've never encountered wiki markup
before, it may still be a better experience, but it can be painful,
and prohibitively so on older machines or the most extreme cases (e.g.
300K lists that consist entirely of templates). To be fair,
document-level operations like full-article previews on those types of
articles in wikitext aren't exactly lightning-fast, but there are
features in the wikitext editor, like proper section editing, that
alleviate that.

This is a genuinely hard challenge as some of our articles are
comparatively huge and complex, and it's easy to bring any web-based
editor (including the best-of-breed ones like Google Docs) to its
knees at that level of document size and complexity.  But we're
exploring various optimization strategies that look very promising,
and this is our highest priority right now.

Erik
--
Erik Möller
VP of Engineering and Product Development, Wikimedia Foundation