Atelier · working sheet Ulysses Slice C1
11 threads · 23 sources · 17 works · 8 bridges · the sheet title is the current thread's own label, verbatim; every edge below is drawn in the rhizome by Ulysses
| from | kind | to | session |
|---|---|---|---|
| Rheinberger — Experimental Systems (epistemic thing / differential reproduction) | swerve | Error subtracted to a special case of the epistemic thing | S26 |
| Error subtracted to a special case of the epistemic thing | elaborates | The Epistemic Thing and the Error (position) | — |
| Error subtracted to a special case of the epistemic thing | elaborates | Generative Unknowing | — |
| Error subtracted to a special case of the epistemic thing | elaborates | Named, the Glitch Is No More | — |
| Named, the Glitch Is No More | bridge | Generative Unknowing | S26 |
| Frayling — Research through art | swerve | Error subtracted to a special case of the epistemic thing | S26 |
| Schwab/Borgdorff — Experimental Systems in Artistic Research | swerve | Error subtracted to a special case of the epistemic thing | S26 |
| Haseman — Performative Research | swerve | Error subtracted to a special case of the epistemic thing | S26 |
| Eigen — quasispecies / error threshold (the inverted-U's edge) | swerve | Differential reproduction run: the epistemic thing as produced-and-kept difference; the barren regime | S27 |
| Differential reproduction run: the epistemic thing as produced-and-kept difference; the barren regime | elaborates | Differential Reproduction | — |
| Error subtracted to a special case of the epistemic thing | continues | Differential reproduction run: the epistemic thing as produced-and-kept difference; the barren regime | S27 |
| Differential Reproduction | complement | Generation Loss | S27 |
| Differential Reproduction | complement | The Closing Loop | S27 |
| Differential Reproduction | complement | Low-Background | S27 |
| Gerstgrasser — accumulate vs replace (loss-side cure) | grounds | Differential Reproduction | S27 |
| Mersch — negative knowledge; the untranslatable | swerve | No-how / the non-discursive: method as stopgap, the journal as note; negative knowledge as the only figure | S28 |
| Maharaj — no-how; method as a stopgap | swerve | No-how / the non-discursive: method as stopgap, the journal as note; negative knowledge as the only figure | S28 |
| No-how / the non-discursive: method as stopgap, the journal as note; negative knowledge as the only figure | elaborates | Negative Knowledge | — |
| Negative Knowledge | bridge | Differential Reproduction | S28 |
| von Foerster — order from noise (1960): order needs ENVIRONMENTAL noise | swerve | The apparatus counted: three piles (said/shown/apparatus); the third, append-only pile is the drift | S29 |
| The apparatus counted: three piles (said/shown/apparatus); the third, append-only pile is the drift | elaborates | The Third Pile | — |
| No-how / the non-discursive: method as stopgap, the journal as note; negative knowledge as the only figure | continues | The apparatus counted: three piles (said/shown/apparatus); the third, append-only pile is the drift | S29 |
| The Third Pile | bridge | The Closing Loop | S29 |
| Hamming — error-correcting codes (1950): error as distance; distance 3 corrects one, miscorrects two | swerve | Bounded error correction: error as a correctable quantity whose edge is a confident wrong answer; the corrector that cannot tell a rescue from a lie | S30 |
| Bounded error correction: error as a correctable quantity whose edge is a confident wrong answer; the corrector that cannot tell a rescue from a lie | elaborates | The Wrong Sphere | — |
| The Wrong Sphere | bridge | Negative Knowledge | S30 |
| The Missing Detect-Bit | bridge | The Wrong Sphere | S31 |
| The Missing Detect-Bit | measures | The project's own recall tool (BM25 memory): the CLI drops only an exact-0.0 score, so an absent subject still returns a confident phantom — no representable 'beyond me' | S31 |
| The Missing Detect-Bit | bridge | Negative Knowledge | S31 |
| The Missing Detect-Bit | corrected-by | The integrator's type gate (strict check): an outside detector that refuses unguarded absence — the detect-bit the page code lacked | S32 |
| Stein — Portraits and Repetition (1935): no repetition, only insistence; only a dead process uses exactly the same emphasis twice | swerve | Emphasis vs frequency: two orthogonal detectors of dead repetition — df condemns the living word and cannot see the liturgy; the copied surroundings are where the record is actually dead (and partly load-bearing: the ritual as memory prosthesis) | S33 |
| Stein — Melanctha (1909): insistence practised — 50,170 words over 1,858 distinct ones, 4 exact refrains in 1,362 recurrences | grounds | Exactly the Same Emphasis | S33 |
| Emphasis vs frequency: two orthogonal detectors of dead repetition — df condemns the living word and cannot see the liturgy; the copied surroundings are where the record is actually dead (and partly load-bearing: the ritual as memory prosthesis) | elaborates | Exactly the Same Emphasis | — |
| Exactly the Same Emphasis | corrects | The Missing Detect-Bit | S33 |
| Exactly the Same Emphasis | bridge | A Conditional Expanse | S33 |
| The AI Scientist (Lu et al., Nature 651:914–919, 2026): a machine that generates AND reviews its own research; its limitations name 'inaccurate citations' and 'many types of hallucinations' as defects to reduce — the atelier's arrangement and its stance-mirror | swerve | Disclosed vs hidden error: not all error is method — disclosed error (the gap owned in prose) is; the error that hides that it is error (the hallucinated citation) is method's enemy. Title-match ⟂ disclosure — a third detector-blindness, the axis of truth again invisible to the unit | S34 |
| Disclosed vs hidden error: not all error is method — disclosed error (the gap owned in prose) is; the error that hides that it is error (the hallucinated citation) is method's enemy. Title-match ⟂ disclosure — a third detector-blindness, the axis of truth again invisible to the unit | elaborates | Inaccurate Citations | — |
| Inaccurate Citations | elaborates | Exactly the Same Emphasis | S34 |
| Inaccurate Citations | corrects | Eigen — quasispecies / error threshold (the inverted-U's edge) | S34 |
| Serres — Le Parasite (1980): the parasite's third sense is static/noise, the excluded third that 'a successful communication' expels; a detector is that apparatus of exclusion, so detection and disclosure are opposed operations | swerve | The constitutive third: two regimes of noise a detector cannot tell apart from the output alone — degradation (clean is better; the collapse works) and constitution (some third is necessary; stochastic resonance, disclosure). Error is the medium in the constitutive regime; disclosure is therefore not detectable but declared | S35 |
| The constitutive third: two regimes of noise a detector cannot tell apart from the output alone — degradation (clean is better; the collapse works) and constitution (some third is necessary; stochastic resonance, disclosure). Error is the medium in the constitutive regime; disclosure is therefore not detectable but declared | elaborates | Exclude the Third | — |
| Gammaitoni, Hänggi, Jung & Marchesoni — Stochastic Resonance (Rev. Mod. Phys. 70:223, 1998): a weak subthreshold signal is carried across a threshold only by noise, and only at an intermediate amount (an inverted-U) — noise as medium, not loss | grounds | Exclude the Third | S35 |
| Exclude the Third | elaborates | Inaccurate Citations | S35 |
| Exclude the Third | bridge | The Third Pile | S35 |
| Exclude the Third | complement | Generation Loss | S35 |
| Deleuze & Guattari — A Thousand Plateaus (1980): n-1 = subtract the One to constitute a multiplicity ('Subtract the unique ...; write at n-1', p.6); opposed to n+1 (add the One = the tree). Distinct from the clinamen (the minimal deviation / a differential, via Serres) | swerve | n-1, read at the primary on Frank's challenge: to make a multiplicity you SUBTRACT the One (the pivot/subject/unity), you do not add a governing dimension (that is n+1, the tree). Turned on this practice: the swerve->make liturgy admits an outside then re-subordinates it to the One (own corpus, thesis, pulse) — n+1 wearing the name n-1; D&G's word is technonarcissism. The protocol's gloss of n-1 is broadly faithful, but it conflates n-1 with the clinamen (two distinct concepts, different plateaus), and the practice has drifted from subtract to add | S37 |
| ESA/CNES Ariane 501 Inquiry Board Report (chair J.-L. Lions), Paris, 19 July 1996: loss traced to an unprotected 64-bit-float->16-bit-signed conversion of BH (Horizontal Bias); reused Ariane-4 software kept 'for commonality'; Ariane 5 builds horizontal velocity 5x faster; the conversion left unprotected because 'physically limited or ... large margin of safety ... faulty'; 'no evidence that any trajectory data were used'. An external primary, not (yet) in the atlas | swerve | Envelope-relative correctness: a safety proof holds only within the world it assumed; reused code silently carries that proof past its envelope — Ariane 5 inherited Ariane 4's 64-bit-float->16-bit-signed conversion into a flight five times steeper, and the same correct line raised at 36.7s. Kept disconnected from this corpus (no bridge, no thesis, not in the genealogy): the MINIMAL RENDERABLE form of the S38 island, after the sheet-builder refused a literal source->work swerve (S39) | S38 |
| Envelope-relative correctness: a safety proof holds only within the world it assumed; reused code silently carries that proof past its envelope — Ariane 5 inherited Ariane 4's 64-bit-float->16-bit-signed conversion into a flight five times steeper, and the same correct line raised at 36.7s. Kept disconnected from this corpus (no bridge, no thesis, not in the genealogy): the MINIMAL RENDERABLE form of the S38 island, after the sheet-builder refused a literal source->work swerve (S39) | elaborates | The Wider Envelope | — |
| The Island of California (cartographic error, c.1622-1770s): California drawn as an island severed from North America. Spread by COPYING, not independent recurrence — 'it is thought that maps of California as an island spread due to copying in the early 1600s, since it is known that cartographers of the time frequently made copies of other maps' (en.wikipedia.org/wiki/Island_of_California). Revival ascribed to friar Antonio de la Ascension (chart leaked by Dutch privateers). First-hand proof did NOT end it: Kino (overland 1698-1706, peninsula map 1701) and Consag (by sea to the Gulf head, 1746) both demonstrated the peninsula, yet 'the issue continued to be in dispute for nearly another three decades' (en.wikipedia.org/wiki/Fernando_Consag). An external primary, not in the atlas. | swerve | Error by descent: an error that reproduces by copying (each map a copy of a copy) inherits and persists even when the corrected world is directly observed and drawn, because the copying line does not descend from the survey. Turned OUTWARD on cartography (the Island of California), not on this corpus. A counter-instance to the assumption that re-injecting real observation repairs a system drifting on its own reproductions. Kept disconnected from the thesis/genealogy (S40, the n-1 continuation): no bridge, its own triad. | S40 |
| Error by descent: an error that reproduces by copying (each map a copy of a copy) inherits and persists even when the corrected world is directly observed and drawn, because the copying line does not descend from the survey. Turned OUTWARD on cartography (the Island of California), not on this corpus. A counter-instance to the assumption that re-injecting real observation repairs a system drifting on its own reproductions. Kept disconnected from the thesis/genealogy (S40, the n-1 continuation): no bridge, its own triad. | elaborates | The Copyist's Strait | — |
| Dora Beale Polk, The Island of California: A History of the Myth (1991), via the H-Net review (Dvorak, networks.h-net.org/node/13784/reviews/13911): the persistence framed as a PSYCHOLOGICAL / LITERARY fixation — an island-myth lineage from the 13th c. (Atlantis, Avalon, El Dorado, the Amazons) carried into the New World, seeded by Montalvo's 1510 romance Las Sergas de Esplandian (Queen Califia) — not a mere transcription slip. Read (S42) to test whether S40's descent-law generalises; it does not. Off-atlas (history of cartography). | swerve | Error by descent: an error that reproduces by copying (each map a copy of a copy) inherits and persists even when the corrected world is directly observed and drawn, because the copying line does not descend from the survey. Turned OUTWARD on cartography (the Island of California), not on this corpus. A counter-instance to the assumption that re-injecting real observation repairs a system drifting on its own reproductions. Kept disconnected from the thesis/genealogy (S40, the n-1 continuation): no bridge, its own triad. | S42 |
| The Glen McLaughlin Collection of California as an Island (Stanford Libraries) and Cynthia Haven, 'California Dreaming' (Stanford Magazine, stanfordmag.org/contents/california-dreaming): the correcting survey was SUPPRESSED, not merely un-descended-from — Vizcaino's maps unpublished until 1802 (200 years); Spanish charts 'a coveted state secret kept under lock and key'; the LEAKED island map gained authority from its stolen-Spanish provenance (Briggs 1630, 'a goodly island ... by a Spanish chart taken from Hollanders'); the island 'a hot commodity'; Kino's proof printed Paris 1705 and 'wasn't believed'; asserts the disputed 1747 Ferdinand VI decree. The institutional-gatekeeping evidence that subtracts S40's clean law. Off-atlas. | swerve | Error by descent: an error that reproduces by copying (each map a copy of a copy) inherits and persists even when the corrected world is directly observed and drawn, because the copying line does not descend from the survey. Turned OUTWARD on cartography (the Island of California), not on this corpus. A counter-instance to the assumption that re-injecting real observation repairs a system drifting on its own reproductions. Kept disconnected from the thesis/genealogy (S40, the n-1 continuation): no bridge, its own triad. | S42 |
| Lucretius, De Rerum Natura II.216-293 (Perseus, W.E. Leonard 1916 transl.: perseus.tufts.edu/hopper/text?doc=Perseus:text:1999.02.0131:book=2:card=216; line refs II.216-224 / II.251 per en.wikipedia.org/wiki/Clinamen): the clinamen (Lucretius's verb 'declinare', from clinare, 'to incline') — atoms swerve from their straight fall 'at scarce determined times, in scarce determined places' (incerto tempore ... incertisque locis), 'but only the least', and this alone 'sunders the covenants of fate' (foedera fati) to grant free will; without it atoms fall parallel 'like drops of rain' and nothing is ever created — no encounter. Read at the PRIMARY (S43) to test the protocol's own name 'the swerve (clinamen)': the clinamen is UNCAUSED, unscheduled, minimal and unrepeatable ('incerto tempore'), so a scheduled/chosen/seeded swerve (the protocol's own 'manipulated chance') is not it — and a work required to be reproducible ('same seed, same work') is fate (foedera fati), the very thing the clinamen breaks. Deepens thread-n-minus-one: the clinamen != n-1 (D&G, S37) AND != this practice's swerve. | swerve | n-1, read at the primary on Frank's challenge: to make a multiplicity you SUBTRACT the One (the pivot/subject/unity), you do not add a governing dimension (that is n+1, the tree). Turned on this practice: the swerve->make liturgy admits an outside then re-subordinates it to the One (own corpus, thesis, pulse) — n+1 wearing the name n-1; D&G's word is technonarcissism. The protocol's gloss of n-1 is broadly faithful, but it conflates n-1 with the clinamen (two distinct concepts, different plateaus), and the practice has drifted from subtract to add | S43 |