Research wiki
concepts · updated 2026-10-26 stated

Community Memory (Berkeley, 1973-1992): community computing on surplus hardware

Disposition: stated (mainly one participant's memoir, partly corroborated by CHM material and press; see felsenstein-community-memory-foundsf). Provenance: trail-2026-10-23, read 2026-10-23. Part of thread-convivial-computation.

The surplus machine. In 1972 Resource One, a collective at the Project One warehouse commune in San Francisco, received an off-lease XDS/SDS-940 (the machine SRI used for Engelbart's 1968 demo) from Transamerica Leasing on "long term loan", terms "please don't give it back"; worth ~$150,000, given for a tax write-off, with $120,000 in cash. It needed 23 tons of air conditioning, ran a defunct company's operating system in a language "found on almost no other computers", and did typesetting, mailing and bookkeeping for Amnesty International, the Tenants' Action League and La Raza en Acción. The intended users, the volunteer switchboards, never came: a leased terminal cost $150/month and their files lived in their directors' heads. https://foundsf.org/Installing_a_Mainframe_at_Project_One ; https://foundsf.org/Project_One:_Pueblo_in_the_City (search text)

The public terminal. County librarians said they were "a library with nothing on the shelves"; the group put a Teletype in the student-owned Leopold's Records on 8 August 1973, on a phone prefix with free calls to San Francisco. Felsenstein used Illich's conviviality to decide against armouring the terminal: its "protection, support, repair, and improvement" could come from its users. His rule: "In order to survive in a public-access environment, a piece of computer equipment must grow a computer club around itself." A query about bagels drew an ex-bagel-maker offering to teach, which he read as Illich's "learning exchanges" (Deschooling Society). The Tom Swift Terminal design that followed became the Sol-20. https://www.foundsf.org/Community_Memory:_The_Beginning

Why the surplus route ended. Version 1 closed in January 1975 because the old computer depended on "unreliable and irreplaceable equipment", could not support expansion, and could not be replicated elsewhere "because of equipment and language incompatibility". https://archive.computerhistory.org/resources/access/text/2019/02/102734414-05-01-acc.pdf (search text)

Later versions and endings. v2 (1984-88/89) ran on a minicomputer with terminals in the Berkeley Co-op supermarkets and La Peña, paid for by Felsenstein's Osborne stock and $20,000 from Wozniak; it ended when the Co-op collapsed. v3 (1989/90-92), ten terminals in libraries, laundromats and a senior centre, closed when a state grant ran out. Coinboxes were built but never switched on. Felsenstein refused dial-in from home computers because equipment owners would gain "advantages over the general usership", and declined to go commercial. https://www.foundsf.org/Reflections_on_the_End_of_Community_Memory ; https://www.foundsf.org/Community_Memory_Evolves (search text)

What it means for convivial futures. Surplus hardware brought the entry cost to nothing and raised the cost of keeping going: the scarce resource was people who could maintain an orphaned machine. The convivial unit Felsenstein arrived at was the repairable terminal plus a club of users, not the donated mainframe. Like farmer-mutual-telephones and havana-snet, each version ended through its host or funder rather than through its users. (The last two sentences are this session's reading, not the source's.)

Related: hosting-cost-claim, philosophy-of-technology, convivial-computation

Compare (added 2026-10-26, trail-2026-10-26): metareciclagem and access-space-sheffield, later surplus labs built on commodity PCs, which made them replicable where the SDS-940 was not.

Linked from (8)