Interestingly, but probably completely unrelated, I note that one of the options on your first "reference" [Varnish: Pageviews By Top Wikis] shows a drop to zero in page views at around 0625 hours UTC today. Of course, the view only gives an 8 hour snapshot, so it's not possible for an ordinary person like me to compare "before and after" changeover. However, that page won't load for me now for some reason so there's little else I can say here.
My evidence that there is degradation is fairly simple: Prior to this change, some users might not "like" the font they were getting, but no matter the project or OS/browser the user was using, the fonts rendered properly. Now, that is no longer consistently the case. If they have made certain modifications to their font defaults for reasons other interfacing with Wikipedia, or they are using non-latin scripts, their Wiki(p)(m)edia viewing experience has a good chance of being negatively altered. Personally, I don't care all that much whether headers and body are two different fonts, although serif fonts always look old-fashioned and dated to me, especially if they include the lorgnette-shaped "g". (Those of you who have met me will appreciate the irony in that statement.) The issue remains the viewability across all of our hundreds of platforms, and that seems to keep coming back to the new forced font stack. This is the issue, not whether or not typography should be updated. When non-Latin projects feel the need to override such a major "update" because of usability and readibility issues, there's a major problem here. Just because they aren't English Wikipedia doesn't mean their issues are minor, and they have the disadvantage of a language barrier to make their problems known. Risker/Anne On 9 April 2014 03:01, Steven Walling <[email protected]> wrote: > On Tue, Apr 8, 2014 at 10:05 PM, Risker <[email protected]> wrote: > > > It's pretty clear that the objectives of this project are not > successfully > > met at this point, and in fact have caused major problems on non-Latin > > script WMF sites, and significant but less critical problems on Latin > > script sites. Several factors for this have been identified in the > thread - > > including limited attention to the effects on non-Latin script projects, > > the insertion of philosophical principles (FOSS) without a clear > > understanding of the effects this would have on the outcome, and the > > unwillingness to step back when a major change results in loss of > > functionality. > > > > [citation needed] > > Loss of functionality? The functionality we're talking about here is > reading of Wikimedia content. It's the most core, most basic functionality > we have. In the case of VisualEditor, which picks up read-mode typography > styles, it's also editing. > > Did reading suddenly become seriously impaired? No. If these things had > happened, you'd see a hell of a lot more outcry than what we've seen now. > If millions of readers and tens of thousands of editors were functionally > unable to read our content easily and smoothly, you would hear a lot more > complaints. If you didn't hear complaints, you'd probably still see a swift > drop in pageviews.[1] > > Instead, what I see is this: a tiny handful of bugs raised.[2] I also see a > relatively small number of editors complaining on Village Pumps -- we have > 75,000+ contributors a month. 137 of them have showed up to complain in > English so far, our largest project. Fewer in other languages. Does that > suggest to you most of our editors are having serious functional issues > reading, particularly when we had 14,000 registered users voluntarily > opt-in to the changes? For reader feedback: comments from readers (on-wiki > and off) have slowed significantly in the days since the change was > made.[3] The same look has been in place on mobile (20% of our traffic) for > more than a year with basically zero complaints. > > This is the first time we've significantly changed Wikimedia typography in > many years. I do not under any circumstances suggest that everything has > gone forward with perfect smoothness. I also 100% agree it can and should > continue to improve.[4] Particularly, I agree that in the immediate future > we need to pay more attention to non-Latin wikis, though everyone keeps > saying "major problems" without actually being specific about what bugs > there are, which doesn't really help constructively. I also would prefer to > find a freely-licensed font to put first in the stack again, as soon as we > can get one that doesn't cause bugs for users on Windows systems. But to > suggest the project was a failure and that is serious loss of reading > functionality is just untrue and frankly hyperbolic. > > Steven > > 1. Our realtime pageviews data is lacking, but HTTP requests and edit rates > for top wikis via gdash.wikimedia.org don't seem to show unusual drops. Am > I wrong? > 2. See https://bugzilla.wikimedia.org/show_bug.cgi?id=63549 for tracking. > Only four are open, and they pretty minor. > 3. > > https://en.wikipedia.org/wiki/Wikipedia:User_experience_feedback/font_sizeas > an example had many comments the first two days. As of today and > yesterday, there are a tiny handful. OTRS has not even had enough comments > to warrant creating a template response. And the number of tweets and > Facebook comments has died down to almost nothing. > 4. We're going to hold a retrospective on the process around this change > later next week. That will include a public wiki page with more opportunity > for people to suggest ways the general process of Beta Features > testing/graduation can be handled better. > _______________________________________________ > Wikitech-l mailing list > [email protected] > https://lists.wikimedia.org/mailman/listinfo/wikitech-l > _______________________________________________ Wikitech-l mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/wikitech-l
