Continuously developing since the early 2000s, social media has rapidly expanded into the society-defining behemoth it is today. During this time, two distinct generations of platforms have emerged, the early exploration generation and the current attention-based generation that has reigned since the early 2010's.
Sonata presents the concept of a new "third-generation" social media platform, which prioritises and fully commits to embracing responsibility.
1 Prioritise users’ free expression and access to information, particularly for journalism; discussion of human rights; educational materials; and political, social, cultural, religious, and artistic expression.
2 Prioritise and set as a continuing goal the improvement of society through top-level decisions, policies, actions and innovation.
3 In guidelines and terms of service, clearly and thoroughly explain what speech is not permissible, what aims restrictions serve, and how content is assessed for violations. Ensure that terms of service and mechanisms for reporting harmful content and appealing content decisions are translated into all languages where the company’s products are used.
4 Reject the attention economy and personalised algorithmic content recommendations.
5 Address the United Nations Bill of human rights and detail how the service will conform with it in all relevant articles.
6 Address corporate social responsibility and maintain a policy in respect of this.
7 Require only the necessary licenses for user-supplied data and content which permit the service to perform its primary purpose of storing, aggregating and distributing such material in a way that remains justifiably financially viable. Any further licences must be optional, opt-in and separately agreed upon.
8 Provide complete transparency regarding published policies. Detail all policy items in extreme detail. Reasoning should include the source of all rules, definitions of all potentially ambiguous terms and case studies or examples for any complex policy item.
9 Publish detailed transparency reports on content removals, both for those initiated by governments and for those undertaken on behalf of individuals or companies. If applicable, transparency reports should also address how machine learning is used to train automated systems that classify, recommend, and prioritize content for human review.
10
Publish regularly updated statistics on data relating to users of the platform by month, in total and total by country:
- Active users
- Total users
- Post creation
- Total views and engagements
- Number of people whose private data is being held
11 Publicise as much data as possible without breaching privacy or security responsibilities and justify each restriction.
12
Permit a yearly independent investigation and subsequent report into the company’s internal affairs, including:
- Legal matters (active case information may be redacted if advised)
- Finances
- Data protection & privacy
- Platform security
13
Open source all code which has involvement in the following:
- Content aggregation (including feeds)
- Content and user search
- Content and user discovery features (such as trending lists)
- Processing of data relating to users
- Content recommendation
14 Establish and publicise a full disclosure on platform ownership, data storage and data processing locations and responsibility for all platform aspects.
15
Openly make available to any self-identifying person over the age of 18 who makes a formal request:
- All publicly available content posted by any person.
- All publicly available data relating to any person.
16
Ensure an exceptionally high standard of data protection and privacy by:
- Developing security measures to prevent data leaks.
- Resisting requests for private information to be revealed by courts, state agencies and governments unless the user's actions on the platform constitute a major crime in the platform's jurisdiction.
- Removing or anonymising personal data is that no longer required.
- Requiring strong security measures for any administrator able to access personal data.
- Appropriately managing administrator access rights to limit exposure of data if a leak occurs.
17 Allow for and support fully anonymous users.
18 Actively and proactively prevent the exploitation of the platform to manipulate political elections.
19 Review and verify the truthfulness of all political advertisements.
20 Actively provide preventive guidance for those searching for topics dangerous to mental health or likely to cause direct physical harm to the user searching.
21 Actively provide functionality to aid those with suspected social media related or induced issues, such as social media addiction or social media induced depression.
22 Reject the use and support of tools that facilitate and aid the propagation of unrealistic personal appearances.
23 Provide optional functionality to reduce or restrict access to types of content as the user requests.
24 Provide an efficient and timely avenue of appeal for users who believe their rights were unduly restricted, including through censorship, banning, assignment of labels, or demonetization of posts.
25 Facilitate and permit the existence of third-party clients in a financially viable way. If charging a fee, break down the costs in detail and justify.
26 Promote healthy competition by interworking services with competitors.
27 Ensure continued value of investment in the event of platform decline.
28 Expand the capacity, geographic, and linguistic diversity of content moderation teams, and ensure they are sensitive to nuances in a language that is spoken across multiple countries or regions. Conduct human rights due diligence assessments to ensure that implementation of moderation does not lead to unintended consequences, such as disproportionately affecting marginalized communities.
29 Ensure advertisements cannot be tailored specifically to single users, especially through the use of AI.
30 Make freely available to anyone self identifying as a researcher any content which was removed as a result of moderation action, along with the reasons for removal, provided you are legally able to do so. If possible, redact non legal text based content sufficiently that it can be made available.
31 Provide a warning of no less than 7 days to relevant parties about any new factual labels which are to be applied publicly to an account representing an institution or business. Facilitate a system of appeal, allowing the entity to provide evidence that the label is inaccurate and should be amended, or withdrawn. When publishing a label, publicise relevant data concerning the decision including data from any appeals, if the entity grants permission. Ensure that all labels applied are accurate to the best of the platform's knowledge. Permit appeals from the entity at any later time for further amendment.
32 Allow any user to close their account and delete all personal data, without any consequence to any unrelated services.
33 Resist governmental pressure from any country on any issues that degrade these policies.
