We reprint this op-ed by Beeblebrox as it appeared in our August 2011 issue to illustrate the eccentricities of our system for obtaining consensus, rather than pending changes itself. His experience motivated him to later pen his excellent essay "The perfect policy proposal" – anecdotal, but a perfect tutorial for anyone wanting to start a Request for Comment (RfC) and not wishing to be 'the guy trying to be the ringmaster of an out-of-control circus'.

They say the road to hell is paved with good intentions. I had good intentions, and they led me straight into Wikipedia Hell. As most of you know, pending changes (PC) was a modified version of the "flagged revision" system used on other Wikipedia projects. It was deployed here as a trial: the trial period expired and ... nothing happened. That's where I come in.

I had applied PC to a few dozen articles during and after the trial period. I got a message on my talk page from a user who noted that I was still using it even though the trial period was over. I didn't think I was doing anything drastic, but it was still bothering some folks because there was no clear mandate to continue using the tool. I'd participated in a number of policy discussions in the past, so I took it upon myself to seek an answer to the question of whether we wanted to retain this tool or not. Six months later, the question remains unanswered.

It started with a simple request for comment ([1]). I really didn't know what to expect. I knew the original discussions had been heated, and that many people had believed this tool would create more class divisions on Wikipedia; but after the trial began, the furor seemed to have died down. My goal was to come up with a yes or no answer as to whether we should use the tool, but in retrospect it was naive of me to think it would be that simple. I opened the discussion on February 16. By the 19th it had grown into a long, disjointed conversation on a myriad of topics. There were many misunderstandings, and a lot of confusion regarding who was supposed to do what when the trial ended. That appears to be where this whole thing went wrong. Everyone was angry that nothing was being done, but nobody seemed to know definitively who was supposed to be doing what in the first place.

Things were getting a bit out of control as discussions were duplicated and new participants added new sections without apparently having read previous posts. On March 8 the second phase of the RFC began. The rate of participation was high, and disruption and factionalism were low. However, a small (it seemed to me) but very vocal group of users felt that we shouldn't have a conversation about whether to keep it until it was turned off. Gradually, this became the primary topic of discussion on the talk page. Contributors began to split into two camps: editors who wanted the tool turned off and those of us who felt this was irrelevant. I was dismayed by what I saw as the emergence of an adversarial relationship. The waters were becoming muddied and an unpleasantly confrontational atmosphere was developing on the talk page: a storm was brewing.

To try once again to organize discussion into a format that would yield usable results, I proposed yet another phase. The idea would be a survey for editors to complete. I'd participated in the Wikipedia:RfA Review/Recommend survey and liked the format. I believed this issue was not as contentious as RfA, and that we could use the combined results of the three phases to determine what the community wanted and move forward. I still believe that.

I tried to roll out the third phase. I asked for feedback on it, but got very little. Eventually it was clear the increasingly vocal users who wanted to switch off PC didn't like the existence of the third phase. For my part, while I didn't "own" the RFC I did feel it should focus on the particular purpose for which I'd created it: to determine whether or not we should continue to use PC. How could we craft a policy on the use of a tool if we couldn't even decide if we would use it, and how could we expect the Foundation to expend its resources to develop it if we were unable to tell them if it would end up being used? I decided to push ahead as only a few users out of the 100+ who had participated in earlier phases had objected to the final phase. The breakdown of what happened in phase two suggested that we had some fairly usable results, and I didn't want to lose the momentum we had. I wanted to get this over the finish line and answer what I was now calling "The Big Question".

I turned on the questionnaire phase after ten days of discussion that had resulted in changes to both the wording and ordering of the questions. Nobody had proposed an alternative procedure other than reverting back to open discussion, which had already proved to be too messy to yield any usable results in my view. Was I being pushy? Maybe, but I felt it was important to resolve this issue, which had by then been discussed for more than a month.

Two questionnaires had been submitted when a user decided to revert phase three and place it on hold pending further discussion. For the first time, I was actually feeling stress and getting angry about something on Wikipedia. I am usually able to keep my cool fairly well, but accusations were being leveled at me and I felt that irrelevant objections were sidelining a major policy discussion that would have far-reaching consequences. I repeatedly stated that if turning PC off was what it would take to get the conversation back on track that we should just do it. That wasn't good enough for some users, and a new third phase was created whose sole purpose was to discuss the temporary use of the tool. I admit that I began to make some intemperate remarks and some foul language crept into my conversation. I was frustrated with Wikipedia for the first time in years. My third phase was put on hold while the other issue was being resolved. By now the RFC had been open for 45 days.

There's a lot more I could say about what happened next, but it is all there in the archives for those who want the details. Eventually I decided I'd had enough: too much time was being spent debating my alleged motivations as opposed to the actual issues, and I quit the process—a process I'd initiated with the simple intention of answering one question. I un-watchlisted the related pages and haven't looked at them again until now. A couple of users expressed concern that I might quit Wikipedia altogether, and I re-assured them that I was just sick of the tactics used in the debate, and didn't want to be part of it anymore. The RFC was finally closed on May 27, 101 days after I opened it. In the end, all that happened was that PC was "temporarily" taken out of use, the same way it was temporarily turned on. It's still there, we just aren't allowed to use it until we finally answer that "big question" I set out to answer back in February. Nothing more substantive than that was decided. There's still no policy on PC. For all that effort, we failed to achieve the primary goal of deciding whether or not to use the tool, although, after a poll, it was eventually removed from all pages on which it was still being used.

When I got the discussion about PC going, I saw it as my opus, my great contribution to Wikipedia's policy structure. Whether PC was kept or not, we would finally have a policy on it one way or the other after many years of debate. Although I admit I had a preferred outcome, what I wanted most, what Wikipedia needed most, was a yes or no answer. I dedicated many hours to organizing the debate and engaging in discussion. In the end it was a bitter disappointment that accomplished nothing. There seems inevitably to come a point in any such attempt where there are simply too many voices, too many nonsensical objections, too much petty bickering to get anything done. This is a growing, systemic problem at Wikipedia, and eventually we are going to have to deal with it.

When people talk to me about Wikipedia I always tell them that the best thing about it and the worst thing about are the same thing. The consensus-based decision-making model works in a lot of cases, but sometimes it fails us because there are no controls. Nobody was able to keep this process moving in a forward direction once those who wanted to discuss a different issue had derailed it. Perhaps, when the tool has been switched off for long enough, we can look at this again and try to answer that one question without the psychological barrier of its simultaneous non-consensual operation. When that day comes, I'll be happy to be on the other side of the fence as a participant, not the guy trying to be the ringmaster of an out-of-control circus.