All vendors

Vendor lineage

RAND Corporation - where packet switching was imagined

Paul Baran's 1964 'On Distributed Communications' argued a network with no center could survive anything - AT&T declined to build it; the internet did.

A Santa Monica think tank, not a vendor - included on merit no vendor matches. RAND built the postwar decision sciences (game theory's workshop, linear programming, the Delphi method), ran early AI on its own JOHNNIAC, and employed the engineer whose eleven 1964 reports specified distributed, message-block, store-and-forward networking: the conceptual root of every router on these pages.

The profile covers Project RAND's 1946 origins, the mathematical toolkit years, Baran's survivability argument and its parallel invention by Donald Davies, and the flow of the idea into the ARPANET.

Founding stories

1948

RAND Corporation

Santa Monica, California · Founders: Spun from Project RAND (Douglas Aircraft / US Air Force, 1946)

Project RAND began in 1946 inside Douglas Aircraft to keep wartime operations research alive - its very first report studied the feasibility of an artificial satellite, eleven years before Sputnik. Incorporated as an independent nonprofit in May 1948, RAND became the archetype of the think tank: mathematicians, economists, and engineers paid to think decades ahead. One of them, an engineer named Paul Baran, spent the early 1960s asking how a communications network could survive losing most of itself - and answered with the idea the internet is made of.

The timeline

  1. Game theory's workshop

    Flood and Dresher formalize the exercise Tucker soon names the prisoner's dilemma; Nash consults over the summers; Dantzig's linear programming matures here - RAND builds the mathematical toolkit postwar decision-making runs on.

  2. JOHNNIAC

    RAND builds its own von Neumann-architecture computer and names it for him - one of the machines on which systems analysis, early AI experiments (Newell, Shaw, and Simon's work), and the JOSS time-sharing system are hammered out.

  3. On Distributed Communications

    August 1964: Baran's eleven-volume series lays it out - break messages into blocks, route each independently through a redundant mesh, let nodes forward 'hot potato' style, and the network survives damage that kills any centralized design. AT&T reviewed the idea and declined to build it.

    RAND memoranda RM-3420-PR and companions, publicly available from RAND.

  4. Two inventors, one idea

    Donald Davies at the UK's National Physical Laboratory reaches the same design independently and gives it the name that sticks: packet switching. The ARPANET's planners draw on both lineages - the rare invention with two honest parents.

  5. Into the ARPANET

    The first ARPANET nodes light up carrying the design philosophy Baran's reports argued for: no center to kill, intelligence at the edges, packets finding their own way - the argument this site's routing tools still assume.

  6. Still thinking ahead

    RAND continues as a policy research institution - defense, health, education - its networking chapter closed but canonical: the conceptual root of everything these pages measure.

Flagship products and solutions

  • On Distributed Communications (1964)The report series that specified survivable, message-block, store-and-forward networking - packet switching's American birth certificate.
  • JOHNNIAC and JOSSRAND's own computer and its pioneering time-sharing system - the workbench of early systems analysis and AI.
  • The Delphi method and systems analysisDecision-making machinery exported to governments and industries worldwide - RAND's other lasting products are methods.

Key innovations

  • Packet switching, argued from survivabilityBaran did not optimize for efficiency; he optimized for surviving attack - and the resulting architecture (distributed, redundant, edge-intelligent) proved superior for everything else too.
  • The think tank itselfRAND invented the institutional form: interdisciplinary, long-horizon, publicly published - the reason an aerospace-funded nonprofit could seed the internet's core idea.

Main markets

RAND never sold networking products - its market is ideas, and this one shipped everywhere: every router, every datagram, every mesh on these pages implements a 1964 memorandum.

Analyst standing

  • Included among the pioneers on merit no vendor matches: the deepest ancestor. When this site's tools reason about paths, routes, and resilience, they are speaking Baran's language.