Trail 2026-10-26 (eleventh session)

Budget: 15 fetched pages or 40 tool calls. Orientation: steer, index, research-values, dead-ends, trail-2026-10-19, trail-2026-10-21, trail-2026-10-23, weekly note 2026-10-25.

Thread chosen: steer item 2, hosting-cost-claim surplus-hardware version, "who actually runs on surplus", taken outside the US: Brazil's MetaReciclagem network (from 2002), which rebuilt discarded computers into community telecentres and later fed the federal Pontos de Cultura programme. Why: the weekly note says item 1 has taken five of ten sessions and is crowding out item 0 (discovery), and the last "worth continuing" (Community Memory running costs) would firm up a known case rather than find a new one. MetaReciclagem is not in the wiki (grep found nothing), is from a culture and policy setting James hasn't drawn on, and tests whether the surplus route can be replicated, which is exactly where Community Memory failed.

Hops

Tooling: web_extract still broken; searches come through the free-tier fallback; Brazilian government PDF downloads with curl and parses with pypdf. Searches count as fetches; each hop says full read or search text.

1. Expect: MetaReciclagem's origin and where its machines came from. -> search text: points to the network's own history (Mutirão da Gambiarra #1, 2009), Fonseca's 2015 Makery essay, and a 2012 government e-waste study that names MetaReciclagem.
2. Expect: the government study's view of these projects. -> ABDI 2012 study (179 pp., curl + pypdf, relevant pages read in full). Against: reuse projects "have no infrastructure, and possibly no interest, in solving the waste question at large scale ... often end up accumulating considerable volumes of unusable material"; credits them with pushing the 2010 solid-waste law; seven federal CRCs reconditioning mainly federal agencies' machines.
3. Expect: the Mutirão PDF in full. -> curl: desarquivo.org does not resolve; browser fetch fails. Nothing.
4. Expect: a Wayback copy. -> none (HTML stub; availability API empty). Dead end, logged.
5. Expect: the archived network site to carry its own history. -> metareciclagem.github.io, read in full: "most active between 2002 and 2012", archived 2016.
6. Expect: a history page among the wiki pages. -> page list read: HistoricoMetaReciclagem, Replicacao, PontosDeCultura, a CRC meeting page.
7. Expect: the network's own account of origin, method and state work. -> four pages read in full. Surprise: the rebuilt machines "belong to those who recycled them (and no longer 'to the project')", which frees people from asset registers, warranties and specialised support; and the network deliberately refused to become an institution: it "does not need scale ... there is no MetaReciclagem body that needs to be fed". Origin by barter: they processed every device the NGO Agente Cidadão received, VCRs and faxes included, in exchange for space and a connection. Members then helped design the Ministry of Culture's Pontos de Cultura digital kit.
8. Expect: Fonseca's later judgement. -> Makery 2015, read in full. Mutirão and gambiarra as the roots; the prototype "already a piece of waste"; early influence was Lowtech/Access Space in Sheffield. Says little about why the network faded.
9. Expect: Access Space's model and fate. -> search text, long extracts (nettime 2000, RTI booklet, access-space.org). Surprise: in 1997 Wallbank's 200 flyers brought so many donated computers that RTI cancelled a 2,000-flyer mailing "for fear of being buried alive"; his graph of use value vs market value: market value drops sharply after year 4 "the use value hardly varies at all", so the lab "will get regular free upgrades". Still running on recycled laptops; Sidney St building demolished.

Stopped at budget: 15 fetches counting blocked attempts (9 hops, several multi-page); ~24 tool calls.

Hunches (unsourced, not in wiki)

• Community Memory died because its surplus machine was exotic and singular; MetaReciclagem and Access Space lived because theirs were commodity and plural. The convivial condition on surplus hardware may be interchangeability, not age or price. For James's servers this argues for a fleet of common x86 boxes rather than one big retired enterprise machine.
• MetaReciclagem's "the machine belongs to whoever rebuilt it" is a property rule, not a technical one; it removes the asset register that makes institutional computing heavy. It rhymes with the urk question from 19-21 Oct (who holds title at retirement decides what happens next).
• Wallbank's year-four cliff and Bashroush's ~6-year efficiency limit bracket a window: surplus kit is free from about year four and still worth its power until about year six to eight. Unchecked for servers.
• MetaReciclagem's refusal of scale and its absorption into a federal programme may both explain why it faded after 2012 (the Pontos de Cultura funding was cut after 2016). Not read; would need the Mutirão history or a thesis.

End of session

Surprises

• MetaReciclagem gave the rebuilt computers to the people who rebuilt them, "no longer 'to the project'", precisely so they could open them without asset registers, warranties or specialised support; and it refused to become an institution because it "does not need scale". https://metareciclagem.github.io/wiki/Replicacao.html ; https://metareciclagem.github.io/wiki/HistoricoMetaReciclagem.html
• In 1997 a Sheffield artist got so many donated computers from 200 flyers that he cancelled the planned 2,000-flyer mailing "for fear of being buried alive", and argued that market value collapses around year four while use value hardly moves, so a lab on surplus gets free upgrades forever. Access Space has run on that since 2000. https://nettime.org/Lists-Archives/nettime-bold-0010/msg00583.html
• Against: Brazil's own e-waste study says these reuse projects end up "accumulating considerable volumes of unusable material" and cannot handle waste at scale. https://www.gov.br/cetem/pt-br/pesquisa-e-desenvolvimento/reminare/central-de-conteudos/documentos/evte_residuos_eletronicos_abdi_2013.pdf

Dead end: the network's own history, Mutirão da Gambiarra #1; desarquivo.org is gone and there is no Wayback copy. Logged in dead-ends. Worth continuing: metareciclagem. Next hop: why it faded after 2012, through an academic study of the spores and the Pontos de Cultura kit (e.g. Miguel Caetano's thesis, cited on the history page), to see whether the end came by funder, as with Community Memory, or by choice. Permission to chase: Brazil's federal Computer Reconditioning Centres (CRCs, now "Computadores para Inclusão"): a state-run surplus-hardware programme still operating, which would show what the surplus route looks like when government owns it. Budget used: 15 fetches (of 15, counting three blocked attempts); ~26 tool calls (of 40).