So here's the basic deal. It's easy to forget that at one point wikipedia was as much a software innovation as it was an encyclopedia. Back when you couldn't physically build an app like google docs (let alone expect it not to run like a dog on most browsers), creating a site where people could edit content, have it versioned permanently, build their own presentation and not have the whole thing fall down when more than 20 people looked at it was a big software advance. The fact that we're having discussions about flow, VE, and so forth in 2014 is a sign not just that the old system basically works but that the problems it was trying to solve were hard. Now we're at the point where the above requirements can be met (basically) without the artifice of the older system, so we can look at what restrictions mediawiki placed on editors and try to fix some of them. Among those are the frankly arcane markup system we use for content and the fact that we use the same system (highly modified) for discussion (VE and Flow, respectively). We're also no longer living in an era where wikipedia is close to the easiest way for someone to contribute to knowledge generally and it isn't unreasonable to expect the devs to try to claw back some of that difference. It's not just chasing google docs for the sake of it either. I work with a lot of new editors (either through the education program or editathons) and the biggest stumbling block (aside from the difficulties anyone has writing what is essentially a literature review) is the pretty forbidding syntax in the editor. One you see less often at editathons (just because there's less use of the talk page) but all the time working on AfDs and what-not is the literally arbitrary structure of discussion. To make it appear that I'm "replying" to you I've prefaced this paragraph with a colon. I could've used an asterisk--though as you'll note from that AN/I thread I followed your two asterisks with 3 but got caught up in an edit conflict (more on that in a sec) and ended up making my own indentation rather than following yours. Most errors are a bit more prosaic (if that's even possible). People start a newline to reply to comments because there's no indication anywhere (save studying the structure of extant replies, which can mix in a lot of different syntax), don't sign their comments (3 tildes to sign undated, 4 to sign with a date, 5 to date without signing!) or generally mistake rendered text (e.g. @example vs {{ping|example}}) for markup necessary to render it. Those are understandable errors but they make people look like idiots when really they just haven't learned a very specific set of skills which serve them nowhere else. And they aren't necessary.
We can lull ourselves into thinking this is the way it should always be because it's been that way since we've all been here. The edit window and the markup system are (for insiders) synonymous with wikipedia. But they're not. They are solutions for a set of specific technical problems which no longer exist. We couldn't edit and have rendered HTML in the same space because browsers in 2001 didn't support it (minimal support didn't come along until 2004). The system we're all using now is a software product built around the constraints of the time. If Wikipedia were created a few years later we'd be using some flavor of markdown (BTW this is top of my list when I get a time machine). A few years earlier and we would've had to make a hard transition to using the browser as an input device after the groundwork was already laid down.
All of that is nice information, I guess, but it doesn't speak to the heart of the problem you're worried about. For that we need to jump to 2007. Some of this history is (thankfully) reported in VisualEditor (home to one of the more hilarious "expert needed" tags), but the broad idea is that after nearly a decade of year over year growth, Wikipedia's editor growth and retention started to flatten out. We can look at it now and suggest that it was a combination of a few things: wikipedia is basically "done" (that is to say, the low hanging fruit for anglophone articles has been picked already), wikipedia is on the right side of the adoption curve, wikipedia now has to compete with hundreds of different means to contribute things to the world ephemerally (twitter, facebook, easy blogging platforms, etc.). But I think if you worked at the foundation at this time you were scared shitless. We were all (foundation and community alike) congenitally disinclined to believe the first cause: it is still an article of faith that wikipedia isn't close to finished. I don't think it is as clear cut but it's also not a stance that someone responsible for the long term health of the site is likely to adopt. The second cause looks tempting now, but 2007 was an inflection point. It's always very hard to judge what is going to happen to growth of any kind when facing an inflection point. Maybe growth and churn is idiosyncratically important to a project like wikipedia and once that falters all kinds of other mechanisms (like, I dunno, RfA) fall apart. Or maybe everything works out fine and wikipedia just grows slower over time. The last cause is also one that the community and the foundation were less likely to take seriously, though if they did the prognosis is not good. We talk a lot about people being here to write the encyclopedia without interrogating what that actually means or why people do it. To take an analogy, a company like Blockbuster once thought they were in the video rental business. And in that business they had a pretty strong grasp on the market. Most of the competition was local and lacked the market power to keep prices high enough to combat diseconomies of scale. But it turns out they were in the entertainment business. People didn't want to rent videos, they wanted to watch shit. When someone came along and let them watch shit without the video rental artifice, Blockbuster discovered their hold over the video rental part of it wasn't worth much. See also BlackBerry. As far as competing for time, we're in a broader business than writing an encyclopedia.
So we were facing this potentially permanent decline in editors and retention with no "good" explanation. What did the foundation do? They conducted user surveys and discovered that most people who hadn't already bought into the mediawiki model of writing a big, collaborative resource were flummoxed by the user experience. And make no mistake, it's fucking terrible. We (as experienced editors) don't notice it, but it's awful. The big pain points identified were those I mentioned in the first paragraph: poor/no indication of what content would look like while editing, no easy way to discover how to actually write something that doesn't look terrible and a confusing discussion system. Unlike the three causes I suggested above, those are solvable problems. They're not easy to solve, but they're solvable. VE attempts to solve some, Flow attempts to solve others. Because they're not easy, solutions like VE/Flow can cause regressions. Last year's introduction of VE was a huge mistake and was partially driven the notion that complaints were less about functionality and more about the locality of cheese. But that mistake doesn't imperil the whole idea of moving forward. I'd like to still be contributing to Wikipedia in 10 years time. At that time I'd like to not still be correcting indentation in talk page posts, fighting signbot when I reply, conflicting with editors on active pages or explaining to new users the distinction between {{this}} and [[that]] as though it were as fixed as the north star. I'd like to help people write articles by teaching them how to structure a neutral summary of a topic, not how to speak to a parser. VE, when it works, will let me do that.
I said above "Ok. Go get your pencil and paper 'cause I got some bad fuckign [sic] news for you" Believe it or not, I didn't mean that I was planning to spill 2000 words onto the page and you should get to transcribing. I meant that the project itself is fundamentally both. It's a software project which seeks to solve the problem of collaborative resource creation. Hell, even inside the constraints of the encyclopedia we've got templates, which are fundamentally software (in the worst possible way), designed to get around the same constraints we had in 2001. All of the metadata for a page was embedded in the page itself, infoboxes and references rendered alongside structured text and stored in the same input format; plumbing and porcelain mixed in a terrifying melange of curly braces and pipes. We're all writing software to speak to a parser when we should be writing an encyclopedia.
Some of this stuff isn't going to work perfectly until it all works. Until we can lift metadata and presentation out of content (basically the WikiData project) the visual editor will remain a vexed problem. And it feels reasonable to say "well, it works now so just test all of these systems until they work together better than the current system before you deploy it". There's a few problems with this view. First, the current system "just works" for the narrowest definition of "works" possible. It's difficult to learn (and unlearn, every time I go to a forum where markdown is the tool of choice I have to unlearn our patterns and relearn theirs), brittle and mixes content and logic in unpleasant ways. And the longer we wait to solve problems which are actually solvable the more risk we run of being consumed by problems which aren't. Every long term editor is here because they've A: mastered the parser and B: figured out how to contribute to a collaborative resource. To attract new editors we would really like to focus on B because A is an implementation detail. And if we believe the technology adoption curve story, we've already attracted those editors who take the time to master the parser--anyone who meets A and B is either already an editor or has left for other reasons. If we believe the idea that wikipedia is basically "done" cataloguing stuff of interest to anglophone editors it gets even worse. Now we want to attract people willing and able to write more than Elizabeth is the monarch of the commonwealth nations on subjects which have less attention, in a way that doesn't attract scrutiny from editors who no longer really believe that wikipedia is the wild west of internet resources. So they have to master sourcing, presentation and content all while translating that through a parser which exists because contentEditable wasn't deployed to browsers at the time. And if we believe the story that wikipedia is competing for scarce time and attention from other services which allow people to easily present their thoughts to the world--if we're in the business of being an internet vocation not building an encyclopedia--it gets even worse. Then the pool of potential editors shrinks not just as wikipedia covers more things or as more people adopt it but as the web gets better around us.
We can reject some of this stuff (and the community does, often and loudly) by saying "well, we're all here to build an encyclopedia, so the people who just want to update facebook aren't valuable to us", which I think misses the point. Plenty of people don't know they want to write an encyclopedia until they try it. Some people contribute meaningfully to wikipedia for years without "writing an encyclopedia article". Some people want to add sources to things or update current events (I nominate NOTNEWS as the silliest core content policy based on practice vs. proscription). Keeper76 (now basically retired) showed up to add some information to local events and ended up reverting vandalism and stuck around doing stuff that was interesting to them for 4 years. But even if we did want to restrict our outreach to "encyclopedia editors" there's no real reason why making it easier and more sensible to edit conflicts with that.
