{"path":"research/what-a-comparison-produces/real-analogies-report-2026-11-09.md","content":"> **Kept as it came back.** The report of a read-only research helper (Claude Sonnet, an `Explore` helper) made on 2026-11-09 for [what-a-comparison-produces.md](../what-a-comparison-produces.md), the all-angles pass on ledger row BL3 (the four products have homes). It was asked for six to eight real, documented analogies whose fate is known, each with base and target, the claimed common pattern, the projected inference and its fate, the difference that mattered and what it defeated ((i) the target's membership in the rule's population, (ii) the rule itself, (iii) only the transfer of one inference), any re-description, and any difference pointed out at the time that proved irrelevant. Claude did not open these sources itself in this pass; the read-status tags are the helper's. Nothing else was changed. Dates, quotes and classifications below are the helper's and are checked in the note before any of them is relied on.\n\nREPORT: seven historical analogies tested against the design rule. Thin sourcing is flagged per case.\n\nCAVEATS FIRST\n- Most pages came back as summaries from WebSearch or WebFetch, not as full text. Read-status tags are conservative.\n- The WebFetch of the Carnot PDF failed (binary). The Cambridge Water, Myth and Measurement, Black 1989 and Wright 1901 primary texts were not read.\n- Classifications (i)/(ii)/(iii) are my reading of the record unless I say the record states them.\n- Dates and numbers below come only from the sources named. Everything else is [unverified].\n- The strongest cases are 1 (Darwin), 2 (Carnot) and 4 (Read v Lyons). Cases 5, 6 and 7 are weaker.\n\nCASE 1. Darwin: artificial selection to natural selection (Wallace's objection)\n(a) Base: Breeders pick individuals with wanted traits and breed from them. Over generations the stock changes in the direction picked. Target: In nature, varying individuals leave unequal numbers of offspring. Heritable traits that help survival become commoner over generations. No one does the picking.\n(b) Pattern: Differential reproduction of heritable variants changes a population's traits.\n(c) Inference: Cumulative, directional change in form, shown in breeding, also happens in nature. HELD. The theory became the foundation of biology. I did not source the later evidence; that is general knowledge, [unverified here].\n(d) No difference that mattered. The inference was not defeated. See (f).\n(e) Re-description: yes. \"Selection\" had to be widened from an intentional act by a chooser to any differential survival or reproduction. Wallace's letter of 2 July 1866 asked for \"survival of the fittest\" in place of \"natural selection\". The reason he gave was that the term and the comparison with \"Man's selection\" invite readers to imagine a thinking agent, an objection \"made a score of times\". Darwin's reply of 5 July 1866 agreed. From the fifth edition of the Origin he headed the chapter \"Natural selection or survival of the fittest\" (per the Project's summary [secondary]).\n(f) Pointed out at the time, but irrelevant: that nature has no chooser or purpose. Wallace said nature \"does not so much select special variations, as exterminate the most unfavourable ones\", and that \"survival of the fittest\" needs no \"intelligent chooser\". A second difference Wallace flagged was the Origin's line \"Man selects only for his own good; Nature only for that of the being which she tends\". It also did not block the transfer. The mechanism needs only variation, inheritance and differential survival. Caution: the replacement phrase was itself misread as \"survival of the better\" (Project note [secondary]).\nSources:\n- https://www.darwinproject.ac.uk/DCP-LETT-5140 [part read, via fetch summary with short quotes]\n- https://darwinproject.ac.uk/survival-of-the-fittest [secondary, search summary]\n\nCASE 2. Carnot: heat falling through a temperature drop as water falling over a wheel (caloric)\n(a) Base: A given quantity of water falls through a height and turns a waterwheel. The work depends on the quantity and the height. No water is lost. Target: A given quantity of heat \"falls\" from a hot body to a cold one through a heat engine. Carnot treated heat as caloric, a substance that cannot be created or destroyed.\n(b) Pattern: Motive power comes from a conserved quantity falling through a difference in level.\n(c) Inference: Two inferences were carried. (1) Work produced depends on the size of the drop, not on the working substance. This HELD, and Clausius kept it in 1850 as the second principle (MacTutor, secondary). (2) The quantity passing through is not consumed, so the same heat is delivered to the cold body. This FAILED. Joule's work showed heat is turned into work in equivalent measure. The evidence is that Clausius took Joule's results as nearly conclusive evidence of the equivalence of heat and work, and then said the conflict with Joule lay not with Carnot's fundamental principle \"but\" with the added assumption that no heat is lost. That wording is from the search summary of the 1851 translation [secondary]; I have not checked the passage itself.\n(d) Difference: water is conserved in the fall, but heat is partly converted into work. Which defeat: (iii). The rule that work follows from a drop stands in a revised form, and the target is still a member. Only the conservation condition, needed for inference (2), was absent. My basis is Clausius's own statement that Carnot's principle survives and only the no-loss addition goes. A caveat: if the shared rule is read to include conservation, the case becomes (ii) or (i). The record supports the narrower reading.\n(e) Re-description: yes. The analogy only worked once heat was described as a quantity that is conserved. After 1850 \"caloric\" was dropped and \"the fall of heat\" was kept as a looser figure. [Interpretation, partly unverified.] One source calls the conservation assumption \"a fortunate misconception\" [secondary].\n(f) None found in the sources.\nSources:\n- https://sites.pitt.edu/~jdnorton/teaching/2559_Therm_Stat_Mech/docs/Clausius%20Moving%20Force%20heat%201851.pdf [secondary, not opened]\n- https://mathshistory.st-andrews.ac.uk/Biographies/Clausius [secondary]\n- https://arxiv.org/pdf/2501.15787 [fetch failed; surfaced in search only]\n\nCASE 3. Thalidomide: rodent tests as a model for human pregnancy\n(a) Base: Rats and mice were dosed in pregnancy and produced mostly embryo death, with few birth defects. Target: Pregnant women took thalidomide and many bore children with limb defects.\n(b) Pattern: A drug's effect on fetal development in one placental mammal predicts its effect in another.\n(c) Inference: \"No malformations in rodents means safe for humans.\" FAILED. Rodents proved \"remarkably resistant\". Rabbits and primates responded, though rabbit defects differed from human ones. Evidence is from reviews [secondary]: Smith and Mitchell 2018, and Bailey 2005 on poor concordance between species.\n(d) Difference: species differ in how thalidomide is metabolised or bioactivated. The proposed explanations are binding to plasma and liver proteins in rabbits versus rats, and DNA oxidation in rabbits versus mice. The 2018 review says the mechanism \"remains enigmatic\", so no single difference is settled. Which defeat: (i) for rodents on this compound, meaning rodents are not in the sensitive population. The rule that cross-species prediction works stands, since rabbits and primates did respond. This classification is my judgement. The mechanism is unresolved, so the record does not clearly choose between (i) and (iii).\n(e) None.\n(f) None in the sources.\nSources:\n- https://pmc.ncbi.nlm.nih.gov/articles/PMC6248049 [secondary, search summary]\n- https://pubs.rsc.org/en/content/articlehtml/2018/tx/c8tx00187a [secondary]\n- https://embryo.asu.edu/pages/studies-thalidomides-effects-rodent-embryos-1962-2008 [secondary]\n\nCASE 4. Rylands v Fletcher to Read v J Lyons (a legal precedent distinguished on a stated fact)\n(a) Base (Rylands): A mill owner's reservoir water escaped from his land and flooded a neighbour's mine. Target (Read v Lyons, [1947] AC 156 HL): An explosion inside the defendant's own munitions factory injured the claimant, an inspector who worked there. No negligence was shown. [Rylands facts are from general knowledge, [unverified here]. Read facts are from e-lawresources and Casemine.]\n(b) Pattern: A person who brings a dangerous thing onto land for a non-natural use is strictly liable for the harm it does.\n(c) Inference: Strict liability without negligence. FAILED. The trial judge applied Rylands, the Court of Appeal reversed, and the House of Lords dismissed the claimant's appeal (e-lawresources [part read]).\n(d) Difference: nothing escaped from the defendant's land. Summaries also report a second ground, that the rule does not cover personal injury. Sources conflict on that ground [secondary]. Which defeat: (i). The Court of Appeal reasoned that Rylands requires an escape. The escape element is part of the rule, so the target fell outside the population the rule covers. If personal injury is the ground, the same classification holds. The reasons of Viscount Simon, Lord Macmillan and Lord Porter were not retrieved, so I cannot quote them. The Cambridge Water summary treats the escape requirement as the rule (Read cited as authority) [secondary].\n(e) None known.\n(f) Not retrieved. [unverified] I did not find a difference in the sources that the court set aside as irrelevant.\nSources:\n- https://www.e-lawresources.co.uk/read-v-lyons-1947 [part read]\n- https://www.casemine.com/commentary/uk/read-v.-j-lyons-&-co-ltd-(1946):-clarifying-strict-liability-for-personal-injuries/view [secondary]\n- https://swarb.co.uk/?p=72444 [secondary]\nNote: Lloyd's Law Reports dates the case 1946 and the Appeal Cases report is 1947. One site prints \"146\", which looks like a typo for 156.\n\nCASE 5. Card and Krueger: price floor in product markets to the minimum wage (weaker; contested fate)\n(a) Base: In a competitive product market a price set above equilibrium reduces quantity traded and leaves a surplus. Target: New Jersey raised its minimum wage on 1 April 1992 from $4.25 to $5.05 per hour. Fast-food restaurants in New Jersey and eastern Pennsylvania were surveyed before and after, and Pennsylvania's wage stayed fixed. The labour market was treated as the market for a good, with the wage as the price.\n(b) Pattern: A binding price floor reduces the quantity bought.\n(c) Inference: Employment falls. The authors \"find no sign\" that the increase reduced employment (abstract [full abstract read]). A second source, via search, says New Jersey stores gained about 2.5 employees per store relative to Pennsylvania [secondary]. Neumark and Wascher's payroll data (230 restaurants) gave a 4.6 percent relative decline [secondary]. The fate is therefore CONTESTED, not known. Use with care.\n(d) Proposed difference: employers with market power over wages (monopsony). I could not verify from this session that Card and Krueger framed their result this way. [unverified] If it holds, the defeat would be (iii): the competitive-market rule stands and the target is a member only if the competitive condition holds. The record does not settle this.\n(e) None found.\n(f) None found.\nSources:\n- https://ideas.repec.org/a/aea/aecrev/v84y1994i4p772-93.html [abstract only]\n- https://nber.org/papers/w5224 [abstract only]\n- https://atticusli.com/replication-crisis/card-krueger-minimum-wage/ [secondary]\n\nCASE 6. Black and Scholes: heat diffusion to option pricing (weak sourcing)\n(a) Base: Heat spreads through a body and obeys the heat (diffusion) equation, which has known solutions. Target: An option's price depends on the stock price and time. The equation for a hedged position has the same form as the heat equation after a change of variables.\n(b) Pattern: Both systems follow one diffusion-type partial differential equation.\n(c) Inference: The solution method of the heat equation gives a closed formula for the option price. HELD as mathematics. Secondary sources say the equation \"can be reduced to\" the physics equation, and that Black had set up the early form of the equation in 1969 but could not solve it then. Whether the formula held in markets is [unverified]; I did not source the criticisms of it.\n(d) Not established from the record.\n(e) Yes, probably: \"diffusion\" was extended to prices. How far this was done at the time is [unverified].\n(f) Not found.\nSources:\n- https://openaccess.city.ac.uk/id/eprint/31577/1/Black-Scholes%20perspective%20revised.pdf [secondary, not opened]\n- Black, F. (1989), \"How We Came Up with the Option Formula\", J. Portfolio Management 15, 4-8 [cited via search; not read]\n\nCASE 7. Wright brothers: bicycle to aircraft control (weak; secondary only)\n(a) Base: A cyclist keeps balance as an unstable machine by constant active control. The bicycle shop taught the brothers about this. Target: An aircraft must be controlled about three axes (pitch, roll and yaw) by the pilot.\n(b) Pattern: Stability is not built into the machine but achieved by the operator.\n(c) Inference: Roll can be controlled actively. Wing-warping came from twisting an inner-tube box in 1899 (Henry Ford page [secondary]). HELD in part. Wing-warping alone did not stop sideslip in the early gliders until a movable rudder was added (Centennial of Flight, secondary).\n(d) Difference: an aircraft banks and turns through coupled roll and yaw. The bicycle lacks that, and the rudder had to be added. Which defeat: (iii), as the rule stands and one carried inference needed a further condition. My reading, [secondary].\n(e) None found.\n(f) A claimed bird-observation source was later discounted by Orville Wright (Air and Space Forces article [secondary]). That is a source question, not a difference.\nSources:\n- https://www.thehenryford.org/collections/explore/popular-research-topics/what-if-bicycles-held-the-secret-to-human-flight [part read]\n- https://webarchive.library.unt.edu/eot2008/20080917111951/http://centennialofflight.gov/essay/Wright_Bros/1901/WR3.htm [secondary]\n\nNOT COVERED\n- Snow and cholera and Pasteur were dropped: Snow's sources were secondary only and I could not pin down the analogy itself.\n- The brain-as-machine sequence and Lorenz were not searched.\n\nSUMMARY TABLE\nCase | Inference held? | Difference that mattered | (i)/(ii)/(iii) | Re-description?\n1 Darwin | Yes | none (chooser's absence was irrelevant) | none | Yes, \"selection\" widened (see (e))\n2 Carnot | Part (drop-size held, conservation failed) | heat is converted to work, not conserved | (iii) | Yes (conserved heat)\n3 Thalidomide | No | species differ in metabolism, mechanism unsettled | (i), uncertain between (i) and (iii) | No\n4 Read v Lyons | No | no escape from the premises (also personal injury, sources conflict) | (i) | No\n5 Card and Krueger | Contested | employer wage power (unverified) | (iii) if so | No\n6 Black-Scholes | Yes as math, not tested | not established | none | Probably\n7 Wright | Partly | rudder needed for coordinated turning | (iii) | No\n"}