Negotiations between strike representatives and Stack Exchange, Inc. have come to a close, with a mutually acceptable agreement being reached and announced. With that agreement being reached, we'd like to clarify how this agreement meets our conditions for ending the strike and what's happening moving forward.
Concluding the strike
Several polls were held since the agreement was reached in negotiations, including a poll for diamond moderators, a poll for the community at large, and polls by the various groups and organizations who went on strike. The results of these polls clearly indicated that the results reached are sufficient to end the call to strike at this time:
- The moderator poll resulted in 50 votes to end the strike, 0 votes to continue (on Discord).
- The community poll resulted in 130 votes to end the strike, 7 votes to continue (on Discord).
- The SOBotics poll resulted in 15 votes to end the strike, 1 vote to continue (in Stack Overflow chat).
- The SOCVR poll resulted in 50 votes to end the strike, 3 votes to continue (in Stack Overflow chat).
- The internal Charcoal (SmokeDetector) discussion resulted in the decision to resume operations.
As of Monday, August 7th, 2023, the call for a strike has ended, and involved groups and individuals are resuming operations. However, each individual will have to decide for themselves if the result is satisfactory enough to stop striking on an individual level; the coordinated call to strike has ended, but nobody involved has the authority to bind any individual into striking or not.
In Moderation Strike: Stack Overflow, Inc. cannot consistently ignore, mistreat, and malign its volunteers, we laid out several conditions for ending the strike. This section will reiterate these and clarify how the agreement reached meets our expectations.
The AI policy change retracted and subsequently changed to a degree that addresses the expressed concerns and empowers moderators to enforce the established policy of forbidding generated content on the platform.
The updated mod-agreement-policy about handling AI-generated content has been released to Meta.SE. The policy outlines that moderators are empowered to take action in cases where suspected AI-generated posts match either one strong heuristic or several weak heuristics. If the company receives an appeal about an action taken on suspected AI-generated content, the moderator who took action is expected to be able to justify the action taken according to the documented heuristics.
This also plays into the agreement reached in negotiations where staff will understand that moderators are willing to discuss actions they've taken and be challenged when an action was taken mistakenly or in error.
The heuristics are still under development and will be for the foreseeable future as the technology in this space changes and evolves. The current set of heuristics is the most restrictive, as only the initial list approved in negotiations are in force right now, but as this process matures, the heuristics defined will grow and evolve to allow for consistent, justified, and effective moderation of AI-generated content.
Reveal to the community the internal AI policy given directly to moderators.
This has been done and is publicly viewable at (Historical) Policy on the use of GPT detectors. The policy as outlined there is no longer in practice and is provided for inclusion in the historical record. Philippe has also, of his own initiative, issued an apology for how some of the language in that policy was interpreted, available in the updated mod-agreement-policy post.
Clear and open communication from Stack Exchange, Inc. regarding establishing and changing policies or major components of the platform with extensive and meaningful public discussion beforehand.
A few bare minimum policies for establishing policy or software changes have been agreed to, with a commitment to go above and beyond that minimum whenever possible.
The bare minimum for binding policy changes is a review period where moderators can provide feedback on the policy for seven business days (as defined by standard Stack Exchange, Inc. corporate business days) before any such policy goes into effect. This will be enshrined in the Mod Agreement.
The bare minimum for significant software changes is to gather community feedback before finalizing any decisions on making those changes, and before any such change is released for beta testing. It's also been established that almost no such change will ever be irreversible if it is detrimental to the platform.
Honest and clear communication from Stack Exchange, Inc. about the way forward.
While this is less of a concrete action to take and more of a change in the culture of communication between the company and the community, Stack Exchange, Inc. has started to improve their communication processes. The company will increase their communication and collaboration with the community, particularly moderators, recognizing that the conflicts here could have been avoided with enough two-sided communication.
Collaborate with the community, instead of fighting it.
Related to the above, the agreement reached in negotiations includes several commitments to increasing communication and collaboration in many areas. This is hard to quantify with individual actions, and will have to be demonstrated on an ongoing basis.
Stop being dishonest about the company’s relationship with the community.
The negotiation agreement has been featured network-wide, demonstrating to everyone on the platform that this conflict happened and that a resolution was reached. The VP of Community has posted an official answer acquiescing to the agreement reached in negotiations (aside from being the primary negotiator in private, this publicly establishes the company's agreement). On August 17th, moderators will have a virtual meeting with the Community Management Team, along with several senior members of the company: Philippe (Vice President of Community); a senior director of Community Products; and Prashanth (the CEO). The moderators will have the opportunity to frankly discuss these issues with the people involved. This is a long-term project, and we'll have to keep tabs on how it develops, but the early signals are pointing in the right direction.
With this progress being made, and as a result of the various polls held, we consider the goals of the strike to have been achieved and the call for a strike ended.
The original version of the above text, posted here as the first revision, was written by Mithical in the above linked GitHub Gist and released under both a CC BY-SA 4.0 and CC0 license. Mithical asked that it be posted by someone else.