Session Date/Time: 28 Jul 2022 20:00

dinrg

Summary

The dinrg (Decentralized Internet Infrastructure Research Group) session served as a joint meeting to discuss the ongoing centralization of the internet. The primary focus was a summary of a workshop held a year prior, which aimed to understand the root causes of centralization, particularly at the application layer. The workshop concluded that the main drivers are economy of scale and security threats, leading to significant concerns about control, influence, and resilience. The subsequent discussion highlighted diverse perspectives, including the "internet in the small" (IoT), the role of network effects, loss of privacy, and the potential need for revolutionary change over evolutionary solutions. A key proposal was made for dinrg to become a focal point for independent submissions and discussions on internet centralization.

Key Discussion Points

Workshop Recap (Lixia Zhang):
- A previous dinrg workshop (summary report in progress) focused on understanding root causes of internet centralization, specifically in application provider consolidation.
- Key Drivers Identified: Economy of scale and security threats are the top two factors.
- Historical Context (Jeff Houston): Centralization is a recurring pattern in industrial history, driven by profit maximization that often conflicts with user/society interests. The internet today exploits personal data for advertising.
- Application Layer Dominance (Christian Hudema): The internet's evolution has shifted dominance from network carriers to operating systems, and now to application services. Application services are easier to centralize due to lack of standardization requirements and faster deployment cycles for monopolies.
- Protocols vs. Deployment: While protocols facilitate data movement, it's application deployment choices (e.g., move from institution-run to revenue-generating services) that drive centralization.
- Security as an Accelerator: Centralized services can afford greater investment in security (e.g., spam filtering, DDoS mitigation), making it difficult for smaller, decentralized entities to compete and offering a compelling reason for users to adopt centralized solutions.
- Concerns: The real concern is the increasing control power and influence of centralized entities over users and society's security, exacerbated by limitations in current web security frameworks for direct user-to-user communication.
- Conclusion: Technology solutions alone are insufficient; effective regulations and legislation, guided by technical input, are crucial.
Open Discussion Insights:
- IoT/Edge Perspective (Jose): Centralization affects "internet in the small" (IoT). Vertically integrated systems force local devices to communicate via cloud, hindering efficient local data-driven applications.
- Local Communication (Dino): Users might value direct, secure, local communication (e.g., iPhone-to-iPhone via Bluetooth) if offered as a feature, challenging centralized messaging paradigms.
- Critique of "Economy of Scale" (Vittorio, Andrew Campbell): For "free" consumer-facing products, network effects, walled gardens, and self-preferencing might be more significant drivers than traditional economy of scale. Lixia suggested rethinking how services like "search" could be implemented in a decentralized way.
- Email Decentralization Challenges (Sachin): While technically feasible for individuals to run email servers, social/technical barriers like IP blacklisting and spam prevention effectiveness push users towards large centralized providers.
- Additional Missing Factors (Andrew Campbell): Loss of resilience in internet infrastructure and loss of user privacy (due to network effects and forced data sharing) are critical concerns not explicitly highlighted in the initial summary.
- Layer-Specific Analysis (Mallory Noodle): A comprehensive mapping of the benefits, detriments, and mitigations of centralization at different internet layers (from hardware to applications), from both provider and end-user perspectives, would be highly valuable.
- Pessimistic Outlook (Jeff Houston): Argued that markets, regulation, and users alone are unlikely to reverse centralization. Evolutionary change is insufficient; revolutionary change, breaking existing structures (like the AT&T breakup), may be the only path.
- Role of "Free" Services & Big Data (Christian Hudema): The accumulation of power stems from big data, which is largely driven by the advertisement-funded "free" internet model. Regulating big data and user data trafficking could be a crucial avenue, suggesting the internet might need to stop being "free."
- IETF's Role (Lenny): Engineers can focus on building solutions that lower the bar to entry and democratize contributions. Acknowledged that consumers choose value, even if it comes from giants. Regulation is viewed skeptically, potentially empowering giants further.

Decisions and Action Items

The report summarizing the previous workshop (with Jeff Houston and Christian Hudema) is in the process of being finalized by Lixia Zhang and Dirk Kutscher.
A proposal will be made on the dinrg mailing list by Mark Fedor and the co-chairs to position dinrg as a focal point/home for independent submissions and further discussions on various aspects of internet centralization and consolidation (economic, architectural, protocol, etc.).

Next Steps

Finalize Workshop Report: Complete the written report from the previous workshop.
Mailing List Engagement: The proposal to make dinrg a home for centralization discussions will be sent to the mailing list for feedback and support.
Identify Activity Topics: Structure the inputs received during the discussion to identify specific follow-up activities.
Interim Meetings: Co-chairs intend to organize interim meetings to continue the conversation and foster contributions.
Technical Solutions Focus: Explore avenues for technical work, such as minimizing data across protocols, developing open solutions for problems like distributed DoS, and enabling direct, secure user-to-user communication, potentially lowering the bar to entry for decentralized alternatives.
Cross-Layer Analysis: Consider initiating work to map out the effects of centralization at different internet layers, as suggested by Mallory Noodle.