https://luminosoa.org/site/books/m/10.1525/luminos.181/

# Introduction: democracy in the wild

* Online communities are different to in-person communities.
* Online politics in the small reflect in the large.
* Online communities must explicitly be democratic, self-governance instead of top-down authority => governable spaces.
* Democratic erosion in the world is influenced by online communities.
* Users of online communities perceive arbitrary rule enforcement, unaccountability.
* Online movements have not resulted in lasting gains.
* The design of online spaces has atrophied everyday democracy skills.
* Garden club from 1960 with eight pages of bylaws => more successful than most only communities that will not live as long.
* Fervent US enthusiasm for forming associations observed by Alexis de Tocqueville in 19th century US.
* Tocqueville: democracy requires education, democracy in education requires political engagement.
* Tocqueville: associations can serve the social order.
* Will bad players behave better if they care about mini-democracies?
* Online spaces are different, more churn, faster, distributed, diverse.
* Participating in online spaces correlate to political participation.
* Author unclear about his disagreement with Tocqueville's conclusions, author is more optimistic.
* Democratic self-governance is harder in online spaces, but possible.
* Design to achieve self-governance, refuse corporate control.
* Technical solutions are not sufficient.
* People do not believe their governments are democratic.
* People are more willing to change due to technological progress.
* Governments use technology as an "unavoidable excuse", but it doesn't have to be this way.
* Introduction of citizen voice happens even authoritarian governments (!)
* Crypto ledger structures have new power structures, even though it's often antidemocratic, but presents an opportunity.
* For many, democracy is something that was created for them before they were born, or something they won't have in their lifetime.
* Online communities are closer to most than their democracy.
* Designing online communities offers chance to learn how to shape the larger government.
* No single design can work for all scenarios.
* Design should be based on accountability.
* Democracy on a small scale gives hope that it's possible on a bigger scale.
* From server control to community control.
* Implicit feudalism: power derives from founders and admins.
* "Governable stacks", "modular politics" to learn from.
* Widespread participation => burdensome, elitist, uninformed governance? Overwhelming to participants.
* Sometimes governable spaces should be highly participative, in others, use representation.
* Governance designs sensitive to economy of attention.

# Implicit feudalism. The origins of counter-democratic design

* A popular group that called for accountability had a flagship organization with a single board member.
* Facebook claimed having "the hacker way": open, meritocratic, but Mark Zuckerberg has majority control.
* Founders solidify.
* Early social platforms had technical conditions that grant administrators complete control.
* Use of "feudalism" is not historically precise.
* "Implicit" because it is not explicit.
* Sometimes platforms do not even allow transfer of power.
* Democracy can arise in feudal technologies due to pressure, this democracy can be similar to primitive democracy.
* But democracy in technology tends to go against the design, the most natural outcome is nondemocratic.
* Implicit feudalism is not a feature, it is merely seen as a non-intentional lack of features.
* First step: perceive lack of democratic features.
* "Exit" vs. "voice"; can only leave, vs. can change things.
* Exit can have costs => captivity.
* Refine voice into "Effective voice" vs. "affective voice" => venting vs. being able to make changes.
* BBS: runs in the sysop house, sysop has absolute power, but also most responsibility and maintenance burden.
* Users being able to leave makes some accountability.
* Limitations of real world (sysop responsibility) lead to implicit feudalism.
* Usenet was bigger scale than BBS, but ultimately "the big 8" ruled (and they named their successors). But Usenet hosted more popular communities than BBSs.
* Usenet hierarchy is decided by the big 8.
* Mailing lists follow similar patterns, administrators have all the power.
* In IRC, iconic channel/network names are a big factor in popularity over performance.
* IRC pioneered bots to execute authority.
* BBS, Usenet, mailing lists, IRC's structure follow that of UNIX, with root, etc.
* Linux and Wikipedia are very productive.
* Linux has BDFL (feudalism).
* Git seems to break feudalism with its distributed nature, but Linux uses a mailing list and the BDFL to control.
* GitHub promotes forks, and user voice in issues, but each project has owners and collaborators.
* Git/GitHub make "exit" easier, but not effective voice.
* Linux added a code of conduct, GitHub encourages project to have one.
* Debian Project Leader is elected, technical barriers of entry.
* Debian/Apache are outliers, non-profits. (Linux is a non-profit too.)
* Wikipedia also has self-governance, but also has BDFL.
* Wikipedia uses MediaWiki for governance (dogfooding).
* But most MediaWiki sites do not have self-governance.
* After Wikipedia's BDFL overreaches, BDFL has diminished power.
* Although software designs can have power vacuums, in the absence of technical software vacuums, "tyrany of structurelessness" often arises.
* Anyone could participate, but not everyone has the time, knowledge, and incentives.
* Big corporate platforms could not have the technical limitations of smaller earlier platforms.
* US Communications Decency Act protects platforms from liability from user behavior.
* Companies could control the platform, but let communities self-govern.
* Facebook/Reddit are different (real names vs. pseudonyms) and in theory provide more control to users.
* Management of communities requires a lot of effort.
* AOL tried to reduce cost of access to voluntary moderators, but moderators realized they made benefits for AOL without sufficient compensation.
* To offload moderation to volunteers in a cost-effective manner, they are paid with unchecked power.
* Author thinks Slashdot moderation worked well and satisfied users, but failed in producing benefit from provocation/engagement.
* Facebook/Reddit grant "affective voice" through karma, etc.; but not "effective voice". Exit is the most effective voice.
* Facebook/Reddit provide moderation tools and gamify moderation (reports on groups performance to incentivize admins to maximize usage). This amplifies implicit feudalism.
* Mark Zuckerberg has power over the Facebook group admins, and engages in democracy theater (2009 referendum on changes to terms of service, required 30% of participation, only 29% achieved, declared "advisory", did what they wanted).
* 2015 "Reddit revolt", blackouts by making subreddits private. Reddit tightened their rules.
* Conway law => structure of software reflects the structure of the organization.
* Facebook/Reddit => the structure of the software shapes the structure of the organization.
* Facebook tried to go to individuals over communities, mirroring WeChat/TikTok which have no social graphs, only driven by personal habits.
* Because TikTok etc. do not have communities, there is less politics, but everything is still controlled by the company.
* Implicit feudalism => control over communities, founder authority, named succession, opaque policies/decisions, supression of user voice, user exit only effective means, only platform owners resolve disputes.
* Implicit feudalism made some sense with limited resources, but not so much with unlimited resources from large corporations.
* Implicit feudalism is part of the business model.
* In contrast, authocratic governments have more democratic "performances" because it resembles legitimate authority.
* But no major online community offers possibilities of even democratic "performances".
* Implicit feudalism is not so effective; most Reddits are small, Miecraft servers median lifetime is eight weeks.
* Exit leads to variety, choice, innovation, but effective voice leads to comitment and stability.
* Example of BDFL becoming inactive led to subgroups becoming more resilient.
* Debian does not exist in isolation; sits between Linux and Ubuntu (both with BDFLs).
* Ubuntu benefits from Debian.
* Debian/Wikipedia combine elections with meritocratic barriers.
* Self-governance seems to emerge more in nonprofits or cooperatives, mirroring ownership structures and technical infrastructures.
* Usenet has some shared governance and autonomy in newsgroups.
* Combination of different power structures helps self-governance; electoral processes + meritocratic barriers for popular but capable leaders.
* Multiple governance mechanisms helps prevent one entity from becoming too powerful, but also allows differently-skilled users from having voice.
* Python had PEPs, when BDFL retired they had some prior art in choosing their new governance, with elections.
* Disassociation/cancellation => no appeals, how long does it last? Affective, not effective voice. These things come because there is no process to challenge those in power.
* communityrule.info => online design of community rules and publication/forking. Try to make it easier to create self-governance.