On behalf of everyone who worked on Documentation, I want to thank all 15,451 users who contributed. We particularly want to acknowledge the 294 people who tested the private beta and the 2,361 who pounded on the public beta in its first month. Your efforts taught us so much about how collaborative editing works and how it fails. We also appreciate the dozens of people we interviewed over the past year. These were eye-opening conversations and immeasurably useful for improving the product. (I also enjoyed being able to get to know some of you over a hangout.) Our top concern at this point is to honor the trust you put in us. We hope you will be willing to participate in future betas, but even if not, we remain in your debt for your efforts with this one.
While it might not seem obvious, we are deeply grateful for both the positive and negative feedback we’ve received on meta. Many people told us this project was too ambitious and that we hadn’t considered some of the deep challenges inherent in what we were attempting. You were, of course, correct. Most likely I wouldn’t be making this announcement if we’d aimed our sights a bit lower. On the other hand, we wouldn’t have learned nearly as much.
We were really excited about this project and I'm disappointed that this is the choice we had to make, but it's the right decision and something that we always knew was a possible outcome. Last December I wrote a series of contingency plans in case we needed to shut Documentation down. Some of the options I considered made it into the plan described after the linebreak. But the entirety of one plan was “Just succeed instead.” That is no longer a possibility.
We still think Stack Overflow Documentation is a good idea. Kevin Montrose’s initial research has mostly been proven correct. Not only did our own survey show that developers rely on official documentation to learn, GitHub's Open Source Survey showed "incomplete or confusing documentation" to be the top pain point. Unfortunately, we can't afford to work on the problem at the moment. While we have an exceptional team of engineers, there just aren't enough of them to support all the projects Stack Overflow is working on.
In order to hire more people, we need to make more money. That might mean helping more developers find a great job or selling more ads or signing up more businesses to use Enterprise. In the future, it might mean selling Channels to new teams. The business pitch for Documentation was that it'd bring in new users who might be in the market for a job. If the feature were particularly successful, it would create new opportunities to sell advertisements. At the end of 2016, we established a metric to aim for: substantially increase the number of Documentation users.
By May, it was clear we weren't on the right path. New users weren't coming to Documentation. So we went back to the drawing board and started another round of user interviews focused on Transact SQL. We brought on a user experience researcher to help us interview technical writers. The results were encouraging in the sense that we know a lot more about what makes for great documentation and how we might support that effort. But it was also clear fixing Documentation would require a significantly larger team.
In addition, it’ll be a very long time before that work will pay off in terms of bringing new users to Stack Overflow. Our interviews showed even very experienced users of T-SQL felt inadequate to contribute documentation. Users with less Stack Overflow experience tended to be intimidated by the prospect of making even trivial edits. So the programmers most likely to become Documentation contributors were already heavily engaged in using Stack Overflow.
Finally, our research showed that while a lot of developers were dissatisfied, the current state of programming documentation is not universally broken the way Q&A was when Stack Overflow started. In particular, we heard over and over that Stack Overflow has become de facto documentation for many technologies. As many of you pointed out**, Stack Overflow is already good enough at providing documentation of obscure features**. Even when considering just the company's mission of helping programmers “learn, share their knowledge and build their careers”, Documentation isn’t the most efficient use of resources.
Even though this outcome has been a disappointment, most of us agree it was the right decision. If so, I hope we'll be able to revisit documentation in the coming years. However, as the ancient Hebrew proverb says, “Hope deferred makes the heart sick; but desire fulfilled is a tree of life.” This was a mature and responsible decision, but that doesn’t cure the heartsickness.
Yes! As Shog pointed out, we've already learned quite a bit from doing things we couldn’t do otherwise. It's too soon to know exactly what we'll be able to port over to Q&A, but I'm excited about the possibilities of CommonMark, technology versions support, shared drafts, better collaborative editing and dated links to previous revisions. Before we leave Documentation, the team will compile what we learned so that we can draw from the knowledge in the future. And, as I will describe below, we’ll publish all of the content generated in Documentation.
More importantly, we've changed our approach to product development across the company. When we started Documentation, our discovery phase was largely done without interviewing typical users. Up to that point, most new features on Stack Exchange were developed with input from Meta or entirely internally. If a feature got used, we'd refine it or just declare victory. If a feature wasn't used, we typically removed it only if was actively causing harm. However, after our experience with the Documentation Beta, we've learned to focus more of our efforts on pre-development research.
For instance, the Documentation team already decided the outline of the feature before coming to meta with the initial announcement. By contrast, our newly formed Developer Affinity & Growth team asked for help setting priorities based on themes and user stories. Documentation’s private beta fleshed out most of the functionality. By contrast, the new mentoring proposal is a minimal viable product that is arguably too minimal. We’re certain to have failed projects in the future, but we’re working to fail a lot earlier in the process.
After August 8, 2017 at 1700 UTC, we’ll disable proposed changes and wait for pending changes to be reviewed. Once the review queue is cleared, it'll be removed from the review menu and page. We’re still working out the other details, which I will try to explain below. For some of this, we need your feedback in order to do right by the community. This probably won’t be the last time we discontinue a beta feature and now is a good time to set a healthy precedent.
As with all content contributed by the community, you own it and we merely have a license. Since we want to make it as easy as possible for you to recover your work, we'll be listening for suggestions in the answers below. At a minimum, we'll provide a JSON archive similar to what we already provide for live content. Let us know what you'd prefer.