We are excited to introduce the Plural Technology Collaboratory (PTC), a cross-company and multi-partner collaboration with a home base within Microsoft Research (MSR) Special Projects.

Collectively, we are exploring how to responsibly create a technological foundation for Plurality (opens in new tab), a future where technology enables, recognizes and empowers trust and collaboration across social diversity, respecting intersectional personal identity and context in communication. The conceptual framework of Plurality was prominently articulated by digital democracy activist and Digital Minister in Taipei, Audrey Tang, and is being expressed in greater detail in a joint book (opens in new tab) with one of us and an online open source community. PTC brings together dozens of MSR researchers across disciplines from the social sciences and economics to mathematics and cryptography with a range of external collaborators: the Harvard Edmond and Lily Safra Center for Ethics (opens in new tab) (which is today launching an allied GETTING-Plurality (opens in new tab) initiative), the University of California Berkeley Center for Responsible Decentralized Intelligence (opens in new tab), Collective Intelligence Project (opens in new tab), Plurality Institute (opens in new tab), EY (opens in new tab), Haun Ventures (opens in new tab), and Protocol Labs (opens in new tab).  Together, we hope to build the largest, strongest, socially-focused, and industry-connected plural technology research group in the world to tackle the unique opportunities and challenges for Plurality posed by the unprecedented technological changes we are living through.

Plural Technology and Generative Foundation Models

The PTC relaunch is motivated by the potential of Generative Foundation Models (opens in new tab) (GFMs), such as GPT-4, to become general-purpose technology, comparable to electricity or computation, but emerging into broad use far faster than either of those examples. GFMs have the capacity to reshape and disrupt much of our society, including the economy, politics, governance, social structures, and the organization of production. Not all of these changes will be beneficial: as Brad Smith highlighted in his recent blog post (opens in new tab), AI-driven innovations can be used to powerfully target individuals with false information, compromise privacy, undermine democracy, and erode trust in society and institutions. As machines come closer to passing all Turing Tests (opens in new tab) (being indistinguishable from human beings), we must remember the capacity for deception this entails and thus we should expect the cost of deception to fall and its quality to rise, dramatically.    

Realizing the full potential of GFMs and reaping their widespread social benefits, while avoiding these risks, will thus require social and organizational adaptations, supported by technology, that keeps pace with this GFM revolution. As such, the PTC will research how Plural Technologies (PTs) can facilitate these sociotechnical transformations to ensure they protect and even strengthen social diversity and cooperation.

MSR is called to this work by the sense of responsibility we feel as developers of many foundational PTs, such as Proof of Work (developed (opens in new tab) by Cynthia Dwork), Data Dignity (developed (opens in new tab) by Jaron Lanier), and Quadratic Voting and Funding (developed (opens in new tab) by one of the authors), as pioneers of social science (such as contextual privacy (opens in new tab) pioneered by danah boyd, (opens in new tab) respect for digital labor (opens in new tab) emphasized by Mary L. Gray (opens in new tab) and Siddarth Suri (opens in new tab) and online community (opens in new tab) studied by Nancy Baym (opens in new tab)) that underlies the aspirations of Plurality. Our experience with such technologies has taught us that we can only maximize their potential if we address their real risks through broad interdisciplinary and cross-sector partnerships, which motivates the approach we are launching today.

Phase one collaborations

Microsoft researchers will be working with collaborators globally to study, design and deploy PTs, focused on the following pillars:

1. Verifiability and Provenance: As GFMs make persuasive and misleading content increasingly easy to create, verifiable content and identity will be increasingly essential across social, business, and democratic systems. PTC is researching cryptographic techniques to address these challenges, fostering secure, private, and trustworthy online experiences, grounded in meaningful and authentic community connections.
2. Economy: To foster a sustainable creator economy, we will research strategies like Data Dignity (opens in new tab) and non-speculative token designs (opens in new tab) (both concepts pioneered by PTC researchers) to protect creators’ data and ensure value returns.
3. Society: While GFMs face criticism for potentially reinforcing social hierarchies, they hold the promise of fostering diverse experiences when applied with contextual consideration and equitable distribution. GETTING-Plurality leader Danielle Allen developed a consortium around “next-generation badging” (opens in new tab) system to make educational evaluation more flexible harnessing PTs and GFMs.
4. Politics and Governance: GFMs can both centralize power and enable large-scale democratic participation, challenging traditional democratic structures. PTC plans to research technologies to foster deliberative discussions, implement innovative mechanisms.  A critical exploration is Gov4Git, (opens in new tab) a protocol designed by our partner Petar Maymounkov at Protocol Labs that harnesses the git version control protocol as a foundation for blockchain-like, but low cost, governance on which the Plurality book will be developed harnessing governance systems developed by our partners at the Collective Intelligence Project and Plurality Institute.
5. Organization of Production: GFMs reliance on centralized computing and public data restricts scalability, but decentralized protocols like Interplanetary File System (opens in new tab) and FileCoin (opens in new tab) designed by our partners at Protocol Labs can harness underutilized edge devices for distributed computation. PTC partners are developing privacy-aware and security-centric training approaches to access valuable private data, unlocking the full potential of GFMs.

Road Ahead

We are only at the beginning of the journey of developing plural technologies.  For example, today many people ask how to know if a piece of content was created by a GFM; tomorrow they are likely to ask how they can know if anything is created by a human; and not soon thereafter even these questions may seem quaint as the patterns of recombination of humans and machine may strain the way we think about agency.  Where privacy is today seen as a fundamental human right but mostly separable from other rights, soon preserving the integrity of communications and their context may be necessary for the protecting all property, identity and access control, as public information can increasingly be used to synthesize personalities, voices and more.

The complex sociotechnical issues surrounding privacy, authentication and provenance have historically taken decades to adjust.  The rapid ride of GFMs may force a dramatic acceleration of this timeline if we hope to preserve and enhance core principles of pluralism, sustainability, and accountability.  The PTC is embracing this generational challenge and welcomes every partner willing to join us in confronting it.

The post Introducing the Plural Technology Collaboratory appeared first on Microsoft Research.

Open-source software, as exemplified by the communities that have formed around the “git” protocol for collaborative software development and version management, is one of the great wonders of digital technology.  Communal open-source projects have found applications well beyond collaborating on code.

They have been used in various capacities to source and share information, such as tracking bugs, sharing scientific data, and writing books and papers. They are also used, in a somewhat ad-hoc manner, for various decentralized autonomous organizations (DAOs) tasks in the Web3 community, such as planning, road mapping, protocol design, policymaking, and others. Most existing applications either serve very small communities or otherwise resort to exogenous mechanisms for the management, security, and credibility of said communities.

We believe that the lack of an accessible, standardized, and secure governance solution for open-source projects is preventing many projects and communities from being formed in the first place, or from growing beyond the Dunbar size (opens in new tab) above which ad-hoc methods of organization are believed inadequate. Additionally, as Generative Foundation Models (GFMs) like GPT-4 become more popular, malicious actors who pose as collaborators will negatively impact community maintenance.

This observation compelled us to attempt a secure and cost-effective framework for community governance, which can be tailored to the specific needs of any one community and can be deployed by non-technical users anywhere where access to git is present. In approaching this goal, we have relied on the learnings of the Web3 communities and have adapted various Web3 ideas and primitives — such as trusted computation or non-interactive zero-knowledge proofs — in our implementation and future roadmap.

To this end, Microsoft Research, and Protocol Labs have been collaborating to develop a git-native protocol with many of the transparency, decentralization, and security benefits of blockchains but that harnesses the power of formal governance to avoid costly approaches to validation and dispute resolution like Proof of Work or to a lesser extent Proof of Stake. 

Gov4git is a decentralized protocol for governing open-source communities based on git. It was created to experiment with organizational and incentive mechanisms that stimulate happy, participatory, productive, and fair open-source communities.  A collaborative, open-source book project we describe below will pilot the protocol.  By designing governance software and protocols, we aim to help any git-based open-source project bootstrap and grow pluralistic, transparent, open-source communities with configurable governance rules. A primary focus of the initial proof of concept is using quadratic voting to prioritize pull requests (PRs) and supporting community-scoped currencies such as voting credits and soulbound tokens (SBTs).

The Gov4git governance framework is aimed at human-centric deliberative processes, is easily extensible and designed to be highly scalable and reliable. Additionally, Gov4git is highly secured where encryption and authentication is used to protect Git-related information.

For our pilot, Gov4git consists of an arbitrarily programmable command line interface and a user interface optimized for a scoped set of features:

• Polling, using configurable and programmable tallying strategies Quadratic Voting (opens in new tab), using persistent voting credits
• General-purpose community-scoped token accounts for members (used for voting credits, badges, and so on)
• Account services for members, including qualitative identity (sometimes called Soulbound) tokens
• Polls and other governance motions are targeted to specific subsets of members using logical member groups

Features like referenda for approval of changes to a source repository, changes to the governing logic itself, and experimental flavors of quadratic voting aimed at promoting diversity of opinions are planned for later releases.  A key application of these polls will be prioritization and eventual approval of suggested changes to the project, often called “pull requests” and of areas for further development, often called “issues.”

E. Glen Weyl, founder and research lead of the Microsoft Research Special Projects Plural Technology Collaboratory, (opens in new tab) and Audrey Tang, Taiwanese free software programmer and the inaugural Minister of Digital Affairs of the Republic of China (Taipei), will debut Gov4git as the platform for a book they are co-authoring Plurality: Technology for Collaborative Diversity and Democracy (opens in new tab).  The book will be built like an open-source software project, with the content being available on GitHub.  Translation to languages other than English will run in parallel on Gov4git as forks of the underlying English version. They will write the book in public and gradually transition management of the content to the community that helps create it.  The book will, among other things, describe the governance principles that are used to create the book itself.  Then, Tang and Weyl will solicit editorial and research assistance openly from the community and enlist those who contribute to help prioritize pull requests and eventually control the contents of the book. This application will both explain the ideas around the Gov4git platform and act as a perfect test of its capabilities, given both its match to the spirit of the project and the security challenges given existing geopolitical dynamics.

Upcoming milestones

• Gov4Git open-source command line demonstration, releasing March 2023
• Plurality open-source book, Chapter 1+2, releasing March 2023
• Gov4Git with user interface implemented, releasing June 2023
• Gov4Git beta state, releasing July/August 2023

The purpose of Gov4git is to help open-source organizations manage their Git repositories more effectively, transparently, and securely, therefore empowering communities to make the self-management so often discussed in the open-source community reality. It allows any git-based open-source project to be governed with the transparency, credible neutrality, flexibility, and security that is prized in the Web3 ecosystem at a tiny fraction of the cost. We believe it is critical that collaborative open-source development remains a sovereign process, which is why our governance solution does not depend on any non-git technologies (such as blockchains). By using this decentralized voting mechanism system, communities can overcome many of the challenges open-source communities run into in being true to their ideals and managing themselves sustainably as their complexity grows beyond what an initial “benevolent dictator” can be reasonably expected to manage unilaterally.   Eventually, we hope this will be a powerful tool for all open-source communities to maintain their ethos in this new GFM enriched environment. To learn more please visit the Gov4git repository (opens in new tab) or follow us on Twitter (opens in new tab).

The post Gov4git: A Decentralized Platform for Community Governance appeared first on Microsoft Research.

Generative foundation models (opens in new tab) (GFMs) like GPT-4 will transform human productivity.  Yet the astonishing capabilities of these models will also strain many social institutions, including the ways we communicate and make sense of the world together.  One critical example is the way that human communication depends on social and community context. “Context” is used in a variety of ways, so we will aim to be precise about what we use it to mean: background data that is roughly “common knowledge” (known, known to be known, known to be known to be known, etc.).

As Nancy Baym (opens in new tab) and danah boyd (opens in new tab) of the Sociotechnical Systems Research Lab (opens in new tab) at Microsoft Research have studied, the applications built on top of the internet have already made it challenging to establish and preserve context for two reasons. First, as we talk to a wider range of people with less shared experience, it is increasingly hard to know what context we may assume in communicating. Secondly, the speed and ease with which information moves has made it hard to ensure that statements remain within the context in which they were made, often leading to “context collapse (opens in new tab),” causing different audiences to misunderstand or misjudge the information based on their disparate contexts.

GFMs risk dramatically exacerbating the challenges of constructing and maintaining context because they make the possibilities for informational spread so much richer.  Soon most of the information we consume may come from global foundation models (rather than from specific organizations with known audiences) and information we share can be carried out of context, but used to train models that recombine that information in ways we are only beginning to understand.  One particularly extreme example is highlighted by (opens in new tab) recent use of related models to synthesize a family member’s voice in a spear phishing/confidence attack, showing that taken to an extreme, decontextualization and recombination of snippets of content can threaten the very integrity of personhood. As GFMs and synthetic attacks like these become common, we expect the use of shared secrets to become increasingly important for authentication, making clear the critical importance of establishing and defending contexts.

Plural Publics

We recently published a paper titled ”Plural Publics (opens in new tab)” with our partners at the GETTING-Plurality (opens in new tab) project of the Harvard Edmond & Lily Safra Centre for Ethics. This paper argues for moving beyond the dichotomy between “public” and “private” information to seek technical designs that foster “plural publics”, the proliferation of firmly established and strongly protected contexts for communication.  We discuss some technologies, leveraging tools of both “publicity” and “privacy”, that may help achieve these goals, including group-chat messaging, deniable and disappearing messages, distributed ledgers, identity certificates, and end-to end encryption.  In the coming months, the PTC will work to build integrated suites of tools to help better achieve these goals, building off our pre-existing, open-source research on systems like U-Prove, which enables cryptographically verifiable attributes that can prove identity, including pseudonymous identities, while protecting privacy and preventing user tracking.

Yet these explorations are just the beginning of work needed in this area and open questions abound. What is the social interface between what a machine knows and what a human knows? To what extent and in what cases is deniability sufficient for preserving context?  When, and in what amount, is common knowledge necessary or desired as humans begin to communicate alongside artificial agents? Can we measure the necessary context needed to enter a public?  How do we facilitate communication and interoperation across many both nested and intersecting publics without undermining contextual integrity?  What is the most efficient way to scale digital certificates on the internet today? What identification methods do we need so that members of a public can communicate securely among each other? What lessons can be drawn from the digital rights management ecosystem to the development of publics today?

PTC understands the urgency of answering these questions as GFMs continue to improve and spread beyond boundaries where their use can be policed by responsible actors. The PTC will continue to embrace this generational challenge and welcome every partner willing to join us in confronting it.

The post Plural Publics appeared first on Microsoft Research.