Reworked diagram

I've reworked the diagram to point out the one way inputs as follows:

My version is ugly but at least I can find the stray ends.

There are parts of this scheme that make no sense to me, however. I don't understand how quality is supposed to deter "participation by a diverse community of contributors looking to fix things". I feel as if someone is skirting a derogatory word here, but if we're to collaborate on improving Wikipedia then honesty is crucial.

Likewise it would be nice to know what "gardening and community tools" are. I assume that this refers to deletion and/or admins?

Then there are the chapters. Do they really belong in this high-level diagram? I mean, I can't ever recall having seen any action on Wikipedia that was done by a chapter, though I assume they are invaluable with fundraising off-Wikipedia. I don't see them as a force measurably improving quality or reach, unless they are doing some things that would probably be called "meatpuppetry" by the various arcane regulations about canvassing.

Love your version. Would be happy if you replaced the graphic on the page with your version. We can do a few iterations, then do a rendered version when we've got good consensus.

Not guilty of skirting, but almost certainly guilty of trying to pack too many ideas into the nodes. What we're trying to say is, if you're looking to fix things, and there's nothing to fix, you probably won't stick around. It's possibly too granular and perhaps not representative of how things actually are working right now.

Gardening and community tools include Talk pages, meta pages, and various admin tools. But, they could also potentially include things like wizards, social features, etc.

Several chapters have made significant contributions by partnering with museums, libraries, etc. Additionally, many chapters run Wikipedia Academies to teach people how to use Wikipedia. And then, as you mentioned, there's the fundraising. The chapters contributed $1.8 million to last year's fundraiser (over 10% of the total), and that figure is likely to go up in the coming years.

I feel abashed to substitute my version, because as I say it's ugly, but if someone does I won't revert it. ;)

The idea that there will be nothing left to fix seems hopelessly optimistic to me. We see from large software manufacturers that no matter how big and sophisticated a program gets, they never run out of "critical updates", let alone minor fixes. Yes, where editing is concerned, some Wikipedia articles do get so good that it's hard to find something to do... but for every article that reaches that point there are always ten more in desperate need of proofreading. And that will remain true if there are three million articles or three billion.

Okay, this is a fair point, and we were probably guilty of overcomplicating the systems diagram. Do you think something needs to be said about how an increase in participation might have a negative effect on quality? Or do you think it's safe to just combine the two Participation nodes into one, and give it a positive feedback loop with Quality?

Honestly... I'd worry that if the diagram is labile enough that my one opinion is enough to change a negative feedback to a positive and combine two categories, based on phrases that I barely understand, then the whole "theory" should probably be dropped entirely as unproven and perhaps unprovable speculation. Sorry... but I'd been thinking there must be some sort of survey or brainstorming session behind this and that someone had a better idea of what was intended.

Even so, I do think that participation has a positive feedback on quality - with Citizendium being a classic demonstration that without participation, it doesn't matter how carefully the wiki is designed to promote quality. Every small wiki has short, vague articles - and even the worst designed wiki analogs, like Hudong and Baike Baidu, which don't generally mark up inline references and editing history as they should, still manage to include highly informative documents.

Trust me, your opinion didn't cause a total flip-flop. The problem with these system diagrams is that the arcs are not weighted. (You can actually specify weight, but I don't think we understand this well enough to do it accurately.) How we chose to represent participation has been troubling me for some time, so getting some additional feedback helped.

There has been a ton of research behind all of this work. But to suggest that anyone totally understands how this all works would be flat out wrong. At the end of the day, a theory is just a theory.

This is where our priority around innovation comes in. One reason for being innovative is simply to increase our understanding. Let's be thoughtful about what we do, but let's not get into the trap of analysis paralysis either. If Ward Cunningham had done a detailed analysis of whether wikis could work before he created them, he probably never would have bothered.

Your point about Citizendium and Hudong and Baidu are well taken. There all these content farms cropping up on the Internet these days, and they're getting great penetration, but the quality is horrific. We obviously don't want to be that.

I places this redrawn version of your diagram in the page: File:Virtuous cycle.png