On Sep 1, 2014 5:10 PM, "Marc A. Pelletier" marc@uberbox.org wrote:
Warning, tl;dr rant below in which live my personal opinion.
Thank you for that. A heartfelt rant feels a lot better than being told my "call is important to you."
(snip)
The fundamental issue is that the WMF is attempting to provide some direction, and the communities do not want any (for various and divergent reasons).
I don't really have all that much of a problem with direction. I have a problem though with strong arming change, which is snap happened here.
(snip)
The process *does* need community engagement. That'd seriously increases the value of what (and how) the WMF does things, and likely reduce the number of bad ideas from the outset.
But the community engagement it needs is one that is done in good faith;
Yes, from both sides. The flow example cited in another email shows this well. There is a large contingent of "thank you for your concern. We won't do that because we believe x rather than y", effectively closing discussion. That's not a great atmosphere to share ideas in. Another frequent problem is saying in the early stages not to worry, it's only the early stages, and all that will be fixed, just come back in a few months, and you'll see how great it'll be, and when it isn't say that earlier feedback would have helped, but nobody showed interest in the early stages.
something which I have yet to see more than exceptions here and there. It also needs fewer reactionnary hotheads. Editing sucks. Reading is lacking. Most of the tooling is crap. That X editors have gotten used to it and have implemented piles of workarounds doesn't justify keeping the old shit around.
MV is a perfect example. 99% of the problems it objectively has (we ignore here matters of taste) derive from the difficulty of parsing the multitude overcomplicated templates living on File: pages to work around the fact that a wikitext page is complete and utter crap at storing metadata. It's not an argument against MV, it's an argument for getting rid of the horrid way we handle File: pages with ad-hoc workarounds.
Yes! This must be said more often!
The *correct* solution is to fix the damn image pages, not to remove MV.
The *correct* solution is to make MV bail completely on pages it fails to parse, falling back to the known bad-but-sufficient behaviour, and maybe adding a hidden category unparsable by MV to the image, so that it can be addressed. If 10% of the effort spent on the long tail of template madness was spent on implementing "when in doubt, bail" much debate would have been unnecessary. Doing the right thing 90% of the time and nothing 10% of the time is preferable over doing the right thing 98 % of the time and the wrong thing 2%.
The same, by the way, goes for VE, which should have had "bail and give me what you have now as wikitext" from the onset, and Flow which needs a "bail and convert this thread to ye olde talkpage thread" (which I fear will be batted away as reactionary crank talk, and "by the time flow will be done unneeded anyway")
-- Martijn
How is it that the old saying goes? "'We've always done things this way' is the most dangerous statement in any language?"
-- Marc