Things came to a head last year when a prominent editor was harassing users for years and the English wikipedia failed to address it. It came to the attention of Wikimedia foundation who took action against the editor, and that caused a storm.
Personally I think they need to do a lot more than a CoC, especially if it's going to be applied by Wikipedia.
I don't get why projects think a top-down straighten-up-and-fly-right edict is how you retain driveby contributors (I get why they do it, I don't get why they make that the primary/only change). That's a highly visible but overall small part of the barrier to entry. Online volunteer collaboration projects often assume anyone motivated to participate will self-train. Making the surface level language nicer only keeps people reading long enough to find out that they've walked into a busy office full of people working with tasks, process and goals they don't understand. Lots of the day-to-day in an office puts efficiency above approachability, and I don't think that's always wrong.
Surely WMF would have better results if they actually worked on the barrier to entry by explicitly adding to staffing(/recruiting volunteers) to onboard new editors. e.g. shepherd their work through onboarding-focused helpers until the newbie is ready to drop into the office proper. Use the failures of conversion to identify where there was a lack of handholding (and build a system for that), vs a lack of interest.