About the BlueSky Meta-ProposalBlueSky Meta-ProposalBSMP_e145e2d4-2225-11ec-8c54-24131b83ea00Berkman Klein CenterFrom members of the Berkman Klein Center communityBoaz SenderCrystal LeeEthan ZuckermanJad EsberNana NwachukwuOumou LyPeter SuberPrimavera De FilippiSahar MassachiSamuel KleinTom ZickUsers are empowered to take control of their experiences_e145e3ce-2225-11ec-8c54-24131b83ea00To support market- and community-based curation of content_e145e482-2225-11ec-8c54-24131b83ea00CommunitiesMarketsContentCurationDiscoveryNormsModerationDiscovery & CurationSeparate the infrastructure from the content elements of curation_e145e522-2225-11ec-8c54-24131b83ea001Content curation ~ At the outset, proposals should separate between the infrastructure & content elements of curation. There is a difference between content that is violent and infrastructure that is violent - and infrastructure can become violent when we assume that content is neutral...Key questions to address:● How do we design community-based curation systems with different subjective curation schema?● How do we let people make recommendations in a community, collect system-wide signals and protect privacy?● How can we enable these things to be community-driven without introducingopportunities for excessive manipulation?User Sovereignty & ControlSupport a network for community-based curation_e145e5b8-2225-11ec-8c54-24131b83ea001.1A key theme that came out of discussions was user sovereignty and control. Realizing the difficulty of having individual personalization without a “centralized operator”, the proposals should allow for an interface that supports a network model for community-based curation, where there is a true marketplace of curation options.FiltersEmpower individuals and organizations to create their own topic filters and choose among those created by others_e145e64e-2225-11ec-8c54-24131b83ea001.1.1MIT’s Gobo ProjectHowever, projects such as MIT’s Gobo project, which gave people the opportunity to “filter in” and “filter out” content to curate their feed, found that, despite the high conceptual interest, there was low practical interest. This is an interface-level design challenge that requires attention.Instead of having one filter for any given topic, the proposals should focus on letting individuals and organizations create their own and users should be able to choose among those already created.VanityEnable users to apply vanity metrics_e145e6ee-2225-11ec-8c54-24131b83ea001.1.2A key factor in this ‘marketplace’ is the presence of vanity metrics. Metrics around popularity and user engagement should not be center screen, however users should have the option to tune this.Echo Chambers & ClustersMake “echo chambers” and clusters visible_e145e78e-2225-11ec-8c54-24131b83ea001.1.3On the high risk of homophily, it is important to be able to see the formation of “echo chambers” or clusters. We have seen that the homophily effect is stronger when you’re opting into conversations within your network (e.g. Facebook) vs. when you’re following people you don’t know (e.g. YouTube).Privacy & SignalingEnable users to relinquish degrees of privacy to gain better system-level signaling_e145e82e-2225-11ec-8c54-24131b83ea001.1.4Hence, system-level signals can be important. It is challenging to get a system-level signal without giving up privacy. Filters that depend on utilizing user data must give users the option to consent to relinquishing that degree of privacy for the benefit of better system-level signal.ModerationRestrict harassment and harm_e145e8ce-2225-11ec-8c54-24131b83ea002The role of moderation isn’t just restricting bad words or racist content. In designing the protocol and reviewing proposals, the conversation around moderation should center around restricting harassment & harm...When considering moderation filters, it’s important to consider some of the macro-risks and address the following questions:● Will certain bodies with money and influence be able to build “filters” that areperceived to be higher integrity?● How will institutions with financial means fair alongside individuals when it comes to these moderation filters?● Who decides if a “filter” should be listed -- will Bluesky have to manage the “app store” for moderation filters? Will it list all user-created filters, or exclude (for example) a white-supremacist filter? ^Perhaps there is an equivalent to ‘real world’ civil rights committees that must step in here to audit and ‘moderate’ the moderation filters. We recommend an entire set of proposals on this question.NormsFrame conversations under norms_e145e978-2225-11ec-8c54-24131b83ea002.1In considering the topic, the conversation should be framed under macro norms which are universal to the protocol; meso norms that are shared across certain clients of the protocol; and micro norms that are specific to a specific client.Virality & AmplificationAvoid virality and amplification_e145ea18-2225-11ec-8c54-24131b83ea002.2It is well documented that our current systems that rely on the virality of user-generated content end up amplifying harmful content - and there is only so much that moderation efforts we tack on can do to mitigate this. In reviewing BlueSky proposals, we must engage with the question of virality and amplification and whether the protocol design avoids this.FrictionDesign in friction on the protocol-level _e145eacc-2225-11ec-8c54-24131b83ea002.2.1Among the beauties and challenges of free flowing online space is the lack of physical boundaries. Traversing “geographies” by jumping from one conversation to another presents no restrictions. However, from a bad actor perspective, this presents an opportunity to scale harassment efforts and disrupt many events at once. Bluesky is an opportunity to “bring in more physics”, designing in friction on the protocol-level as a proactive way to avoid downstream moderation issues.New AccountsIntroducing a monetary cost to start a new account_e145eb76-2225-11ec-8c54-24131b83ea002.2.1.1Without getting into the complex issue of identity, increasing the cost of creating a new account, including introducing a monetary cost to start a new account, might be effective.NeighborhoodsEnable users to see each other's “neighborhood”_e145ec2a-2225-11ec-8c54-24131b83ea002.2.1.2Enabling users to see which “neighborhood” other users are coming from could help users identify a provocateur and take action themselves.User Tagging & InterestsEnable visible ‘tagging’ of users while highlighting their shared interests_e145ed38-2225-11ec-8c54-24131b83ea002.2.1.3In addition to helping avoid brigading, ways of visibly ‘tagging’ users could help identify “sock-puppet accounts'' and make bots easily identifiable. However, visibly tagging users could present the risk of short-circuiting judgments, and so the system should also present opportunities to identify any cross-cutting cleavages - for example by highlighting shared interests between users.DecentralizationEmpower users to take control of their own experiences_e145ee5a-2225-11ec-8c54-24131b83ea002.2.1.4Modular moderation filters In general, moving power down to the ends of the network so that users have the ability to decide between competing moderation filters in an open marketplace empowers users to take control of their experience. In their design, we highlight the importance of allowing users to freely “plug in” a moderation filter so that users can see what the resulting content looks like before committing to it.Business Model(s)Structure BlueSky as a non-profit entity and maintain a commitment to open-source_e145ef9a-2225-11ec-8c54-24131b83ea003WikimediaWikimedia’s structure is worth drawing inspiration from.In terms of legal business structure, BlueSky should be structured as a non-profit entity and should maintain its open-source commitment. At the outset, we do not think BlueSky, as a base protocol, needs a revenue model. To fund its operations, BlueSky should rely on the various avenues non-profits pursue to fundraise.Open StandardsServe as a bridge for interoperability among open standards_e145f0ee-2225-11ec-8c54-24131b83ea003.1BlueSky is well-positioned to bridge between the many open standards (which are currently incompatible) and evolve with them. Ambitiously, BlueSky should be interoperable with as many major social platforms’ standards.MonetizationSupport a decentralized monetary scheme for the collection and distribution of funds_e145f1ca-2225-11ec-8c54-24131b83ea003.2In considering the question of tokenization, attempts thus far demonstrate that social tokens have deflationary behaviour that makes them poor for engagement. Instead, BlueSky should have a decentralized monetary scheme that can support the collection and distribution of funds.This would enable clients that build on BlueSky to decide on whether to monetize - especially considering that these actors would be providing important value in building on the protocol.Additionally, this would provide incentives to the community to contribute to maintaining BlueSky. Importantly, BlueSky should not extract rent here.Monetization on the client-level can take many forms and ‘templates’ can be provided. These can span the range of proven approaches from subscription (pay to access the system) or membership (everyone has access to the system, but some people *can* pay - people can step up and pay more than their fair share) to more transactional models like tipping contributors or purchasing products or services on the platform.Marginalized CommunitiesAvoid harm to marginalized communities_e145f292-2225-11ec-8c54-24131b83ea004Marginalized CommunitiesFinal Principle ~ In wrapping up this meta-proposal, we want to underscore how important it is to be cognizant of the potential for a new protocol to harm marginalized communities.FocusCenter the people who are most hurt/marginalized by these technologies_e145f36e-2225-11ec-8c54-24131b83ea004.1We urge all parties to consider the need for the protocol to be designed and implemented in a way that centers the people who are most hurt/marginalized by these technologies.This should include asking ourselves:● Is the team designing the protocol and assessing proposals representative of the people who are most hurt/marginalized by these technologies?● Are the initial clients of the protocol diverse? This shouldn’t be another case ofUS-based initial clients and the rest of the world follows.2021-03-312021-09-30https://poseidon01.ssrn.com/delivery.php?ID=473081090103074074071006112011008091117005056063054005028094067064085105104113064122018063022011024019026116121030067078076122013040042001029100104107086096112091111070080003097017066097122092012012006095098006100091104080109083089067094025096086115103&EXT=pdf&INDEX=TRUEOwenAmburOwen.Ambur@verizon.net