notes: research/papers/papers-many-cubed

This data as json

path title type status classification words body
research/papers/papers-many-cubed Many Cubed; Solving the Many-to-Many-to-Many Problem for Scalable Reliable Secure Distributed Internet Applications summary v0.1 — 2026-10-15: drafted by Claude from the PDF in the Papers repository; unread by Mark; marked partner on Mark's word 2026-10-21, paraphrase only partner 481 # Many Cubed; Solving the Many-to-Many-to-Many Problem for Scalable Reliable Secure Distributed Internet Applications **Abstract.** A 2017 white paper by Samuel M. Smith on the design of secure, scalable distributed internet applications, arguing that they must solve a "many-to-many-to-many" problem in how components share information and proposing routed data flows over the RAET protocol as a solution. It predates KERI and is of historical interest for the security principles it sets out. For architects of distributed systems. ## Summary The paper says scalable, reliable and secure distributed internet applications must solve the "many cubed" problem in how their components share information. The first "many-to-many" comes from security. It rejects perimeter (defense-in-depth) security on the grounds that "network locality is not security", and proposes a diffuse-trust, perimeterless model with five principles: the network is always hostile; every interaction is authenticated and authorized with best-practice cryptography; communication is end-to-end encrypted and data stored encrypted; policies change dynamically with behavior (reputation); and policies are governed by distributed consensus. Each of these requires a many-to-many communication topology. The second constraint is performance and reliability: load balancing across machines and data centers, redundancy, and clustered M-of-N reliability, again many-to-many. The third "many" is the diversity of components, which the paper calls flows: logically concurrent flows of execution or data, each uniquely addressable and with its own requirements for security, reliability, latency and throughput. The paper then addresses mapping flows onto hosts: scaling, complexity, coroutines for computation, User Datagram Protocol (UDP) rather than TCP for the network, and data-flow routing with a four-part address (datashare, Unix domain socket, UDP and channel names) and an example routed message. It reviews DDS/RTPS, ZeroMQ and RAET (Reliable Asynchronous Event Transport), and describes RAET's CurveCP-based handshake, which authenticates name, role and mode, along with its flexible header, large-message support, presence detection and interprocess protocol. It closes by proposing a RAET Router, modeled on Erlang's port mapper, to extend RAET to languages other than Python. ## Key terms many-to-many-to-many, diffuse trust, perimeterless security, distributed consensus, flow, data flow routing, coroutine, RAET, CurveCP, RAET Router ## Claims and their evidence - Perimeter security fails because hostile users exist inside as well as outside the network; network locality is not security. [informative] - Distributed consensus, granular authentication and dynamic authorization each require a many-to-many communication topology. [informative] - UDP-based transport is easier to load-balance and tune at scale than TCP for this use. [informative] - RAET's handshake authenticates name, role and mode, which bootstraps per-flow authorization. [informative] ## Related Open Reputation Framework (2015), Decentralized Self-certifying Self-Identifying Data (2017), Keri Request Authentication Mechanism (KRAM). A markdown version of the same paper is in the repository. ## Provenance Drafted by Claude on 2026-10-15 from the PDF `whitepapers/ManyCubed.pdf` (version 1.1, 2017/01/05), text extracted: the overview and problem sections in full, the solution sections by their headings and the RAET sections in full. Not yet read by Mark.

Links from other tables