I generally agree with your analysis Marc, notwithstanding that there is blame to share on all sides - not just users who point to broken edge cases. The (quite predictable) behaviour you mention is why I was quite fond of the way the "usability initiative" from several years ago (the team that built the Vector skin) operated.
They made it very easy to opt-in AND out of the beta version of their work. They also made it very clear that when the proportion of people who stayed "in" reached a certain level (If I recall correctly it was 80% and certainly not 95%+), that would signal that the software had reached a certain level of acceptance, and that a sufficient number of edge-cases had been addressed. Until that point they would focus on working with the people who has tried the beta but had turned it OFF, to identify their personal edge cases - in the hope this would fix other people's problems. Much like you described.
The key here in my opinion is: - clear communication about what state constitutes "success" (e.g. "When 80% of people who have opted in have STAYED opted-in") - clear communication about the progress towards that state (e.g. Showing the "success" factor in the little statistics on the "beta features" tab, and showing how close we are to that.) - only moving to the next stage when that state has been reached, (not a fixed schedule but "it is ready when it is ready, not before") - making it easy to try, and withdraw from, new things, always starting with opt-in before making it opt-out.
Since the "usability initiative" there has now been introduced the "beta" preferences function for editors to test new software and provide feedback. I would like to see this used more consistently and with more clarity about what stage of "beta" the individual elements within that system are at. Currently all I know is that there is a list of items and that there is an absolute number of users for each item (e.g. "1,234 people are using this.") Would it be difficult to make it show what proportion of people who have tried any given beta feature are STILL using it? That would give an indicator of popularity/acceptance at least.
-Liam / Wittylama.
On Tuesday, 2 September 2014, Marc A. Pelletier marc@uberbox.org wrote:
On 09/02/2014 02:52 AM, Yann Forget wrote:
OK, I could buy that [fixing image pages]. But then why not fixing that *first*, so that any MV implementation coming afterwards would be smooth?
In the best of worlds, that would have been ideal.
Now, no doubt I'm going to be branded a cynic for this, but have you ever /tried/ to standardize something on a project? Obviously, my frame of reference is the English Wikipedia and not Commons; but in a world where there exists at least six distinct templates whose primary function is to transclude a single "<references/>" onto a page and where any attempt to standardize to one of them unfailingly results in edit wars, that doesn't seem like a plausible scenario.
Perhaps the problem is more fundamental than this, and we're only seeing symptoms. I don't know.
But I /do/ know that waiting until every edge case is handled before deploying (attempted) improvements to the site is doomed to failure. If only because most of the edge case won't even be /findable/ until the software is in place so that it can't work even in principle.
IMO, in practice, "get it working for the general case and most of the obvious edge cases" is a reasonable standard; and I'm pretty sure that MV qualified under that metric (and VE didn't).
I suppose much of my frustration over the MV keruffle is borne out of a reaction I see much too often for my taste: editors yelling "OMG, look, image X isn't properly attributed/licensed/etc in MV! Burn it with fire!!!" rather than figuring out why X's image page isn't properly parsed and /fixing/ it (and possibly an underlying template that could fix dozen/hundred others in one fell swoop).I'm pretty sure that if half as much effort had been spent fixing issues as was attempting to kill MV, its fail rate would already be at "statistical anomaly" levels.
.. but my inner cynic is also pretty sure that many of the loudest voices wanting to get rid of MV ostensibly because of its failings don't actually /want/ those failings to be fixed because being able to say "It's broken" rather than "I don't like it" sounds much more rational.
-- Marc
Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org javascript:; ?subject=unsubscribe>