[Wikimedia-l] Let's have the courage to sit down and talk about VisualEditor

[Erik Moeller]

Hey Tomasz,

this is a good way to start a new thread here, so let me respond.

We've done the following with regard to the VE beta so far:
- We've overall slowed down the beta rollout schedule;
- We've excluded nlwiki from the phase 2 beta rollout;
- We've switched dewiki back to opt-in;
- We've offered an easy way to hide VisualEditor (it was always
possible not to use it).

We've also pushed out lots of fixes & improvements, including a first
round of performance improvements, a better (still not awesome)
citations dialog, improvements to the template dialog, etc.

If we have to globally go back to opt-in for a while, or some
middle-ground (instant opt-out, or advertised beta) we'll do that too.
We'd prefer not to, because having developers quickly see the impact
of their changes at scale, positive and negative, is essential to
making the product better quickly. The steady stream of feedback has
been invaluable, and I think the changelog of the last few weeks
demonstrates that beyond all doubt.

We see very few roundtripping errors. We do see incidents of
problematic edits that are unique to a visual editing environment
(e.g. backspace-deleting an infobox vs. having to select & delete all
the markup representing it; reduced precision in applying formatting
to content due to mouse selection). Some of those we can help reduce,
some of them are inherent and we'll likely have to accept as a cost of
introducing a new editor. And we do see the much-discussed issues with
complex templates where the question isn't really who is "at fault"
but what the right long term solution is.

There are quite a few ugly bugs (it's a beta), but most of those
aren't that hard to squash. It'll take longer to make really big gains
in performance, though -- that's a genuinely hard problem.

We don't think going back into opt-in mode is in the interests of
advancing Wikimedia's mission -- maintaining VE as the "edit" button
while making it trivial to edit source forces us to really stay at a
high pace of continuous improvement that meets user needs. Nothing
constrains choice and imposes urgency like reality.

Our preference is therefore to work with the community to stay in
continuous improvement mode. If the overwhelming community sentiment
is that the cost of continuous improvement with a large scale user
base is larger than the benefit (as it was on dewiki), we'll switch
back (or to a compromise), and use a more rigid set of acceptance
criteria and a less rigid deadline for getting back into large scale
usage later in the year.

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