THRIVING after the FALL
Book Three GOVERNING after the FALL

GOVERNING · Part I · The Levelling Floor

The numbers and sources

Chapter 2: The Scouts' Dance

The chapter tells its story with a handful of figures. Here is everything behind it: each figure with its study, its research document and its row in the log, the capture test and the scale ledger in full, the case for and the case against, where the sources disagree, and what is still open.

  • 63 open
  • 23 to check first

Read about the chapter · Its 63 rows in the log

Numbers to carry

The figures the chapter is built on, each with its research document and its row in the log.

  1. About 15 scouts present at one candidate site at the same time make a quorum, and the swarm commits without waiting for consensus (Seeley, in NA-01, range 10 to 20). Row C02-17.
  2. In 4 of 5 trials on Appledore Island, Maine, swarms offered five nest boxes chose the single good one (Seeley and Buhrman 2001, Behavioral Ecology and Sociobiology, P = 0.0064, NA-01). Row C02-07.
  3. A red deer herd moves when a mean of 62 per cent of the adults are standing, standard deviation 8 per cent (via Conradt and Roper 2003, Nature, NA-02). Row C02-27.
  4. Worker bees destroy about 99 per cent of the eggs laid by other workers (NA-01, from a single university page, behind Ratnieks 1988, The American Naturalist). Row C02-23.
  5. The mean of 787 guesses at the weight of an ox at Plymouth in 1906 came within a pound of the true dressed weight of 1,197 lb (Galton 1907, Nature, audited by Wallis 2014, PS-03). Row C02-53.

Every figure

One row for each figure or finding the chapter rests on. Status is how the finding stands in the research: robust, contested, open or fallen. It is not yet the log's verdict.

FigureWhat it measuresSourceDocumentRowKindStatus
Appledore Island, Maine, treeless, 39 hectaressetting of the opening scene and of the experimentsSeeley and Buhrman 2001, Behavioral Ecology and SociobiologyNA-01C02-01recordrobust
About 10,000 bees and one queensize of a swarm that leaves the hiveSeeley 2010, Honeybee DemocracyNA-01C02-02recordrobust
Several hundred scoutsshare of the swarm that searchesSeeley 2010 (NA-01 gives an area of about 70 square km, not used)NA-01C02-05recordopen
Five boxes, four mediocre (about 15 litres), one good (40 litres, 15 square cm entrance near the floor, facing south)the choice offered to each swarmSeeley and Buhrman 2001NA-01C02-04recordrobust
4 of 5 trials chose the good box, P = 0.0064accuracy of the swarm's choiceSeeley and Buhrman 2001NA-01C02-07recordcontested (five trials, called extraordinary accuracy in NA-01)
1949 and 1951, Munichwhen and where Lindauer saw scouts dancing on a swarmLindauer 1955, Schwarmbienen auf WohnungssucheNA-01C02-09recordrobust
1955, plebisciteLindauer's paper and his word for the choiceLindauer 1955NA-01C02-12saidrobust
Inspection of about 40 minutes (prose says only a long time)how long a scout spends inside a cavityblog source cited in NA-01NA-01C02-15recordopen
About 100 circuits for an outstanding site, about a dozen for a mediocre one (prose says only long and brief)dance strength against site qualitySeeley, via NA-01NA-01C02-16recordrobust
About 15 scouts, range 10 to 20, at one site at oncethe quorum thresholdSeeley, via NA-01NA-01C02-17recordrobust
2012, stop signal, headbutt of 0.21 plus or minus 0.10 seconds (prose says brief)cross inhibition between rival dancersSeeley, Visscher, Schlegel, Hogan, Franks and Marshall 2012, ScienceNA-01C02-18recordrobust
2010, five lessons (prose says lessons)Seeley's advice to human groupsSeeley 2010, Honeybee DemocracyNA-01C02-21saidrobust
Relatedness 0.5 to a son, 0.25 to a brother (prose gives no numbers)the worker's reason to cheatRatnieks 1988, The American NaturalistNA-01C02-22inferredrobust
99 per cent of worker-laid eggs destroyedstrength of worker policingUniversity of Sussex page cited in NA-01NA-01C02-23recordopen
1988, policingwhen the idea was formalisedRatnieks 1988NA-01C02-24inferredrobust
2003, democratic decisions cheaper in most conditionscost of despotic against shared decisionsConradt and Roper 2003, Nature 421NA-02C02-25inferredrobust
62 per cent (SD 8) of adult red deer standing, groups of about 10 to 50movement threshold in red deervia Conradt and Roper 2003NA-02C02-27recordrobust
Olive baboons, Kenya, collars logging every second, 25 animals, 123,773 pulls and anchors (prose gives no numbers)shared decisions in a ranked troopStrandburg-Peshkin et al. 2015, Science 348NA-02C02-28recordrobust
Compromise below 90 degrees, majority above (prose says not far apart and far apart)the troop's rule for disagreementStrandburg-Peshkin et al. 2015NA-02C02-29recordrobust
Golden shiners, groups of 150 and 300 (prose gives no numbers), 2011uninformed members restore the majority's choiceCouzin et al. 2011, Science 334NA-02C02-31recordrobust, model values blank in the PDF
1947, captive unrelated wolvesorigin of the alpha ideaSchenkel 1947NA-04C02-34recordrobust
1970, 1986 to 1995, 1999, Ellesmere Islandthe alpha idea popularised, the wild pack watched, the idea withdrawnMech 1970, The Wolf, and Mech 1999, Canadian Journal of Zoology 77NA-04C02-35recordrobust, place garbled in NA-04
Cheats within weeks (10 to 15 days of daily transfer)how fast lasR mutants appearPNAS 2015 study, attributed in NA-05 to Diggle 2007NA-05C02-40recordopen
Slime mould, Tokyo map, length 1.75 plus or minus 0.30 of the minimum tree against about 1.8 for the rail lines, disconnection risk 14 per cent against 4 per cent (prose gives no numbers)efficiency and fragility of a web grown without a planTero et al. 2010, Science 327NA-05C02-44recordrobust
Hamlet of 30 to 50, village of 150 to 500the rungs In practice speaks tothe book's ladderbrief, Concept NoteC02-52designrobust
Plymouth, 1906, mean within a pound (787 cards, mean 1,196.7 lb against 1,197 lb, median 1,208 lb)accuracy of independent estimatesGalton 1907, Nature, audited by Wallis 2014PS-03C02-53recordrobust
An evening a monthtime a village quorum meeting needsNA-01 village implementability (a monthly meeting)NA-01C02-62designopen

The capture test

The captor: the most patient person in the room, who knows the threshold and knows most of the room is undecided. The first move: arrive early with allies so that a quorum appears before anyone has looked (a false quorum, NA-02), or frighten the undecided so that the neutral middle takes a strong preference and stops diluting the faction (NA-02, the golden shiner model run backwards). What stops it in animals: a steep quorum response, so that a few initiators are followed reluctantly and a crowd suddenly (Ward and colleagues 2012 on sticklebacks, values blank in the PDF), and a quorum made of individuals who each inspected the site (the swarm). In the hive the cheat is a worker laying her own eggs, and she is caught by policing keyed to the queen's scent (Ratnieks 1988). In bacteria the cheat is a lasR mutant, and cyanide made by the cooperators kills it. In wolves the would-be usurper appears only in captivity, where young animals cannot leave (NA-04). The chapter carries the human version as one turn in "Who moves the herd" and the defences in In practice: wait for the room to fill, put a night between a frightening question and the count.

The scale ledger

ArrangementHouseholdHamletVillageDistrictWatershedRegionNationPlanetDocument
Honeybee swarm, scouts and quorum (by bee headcount)worksworksworksworksworks, upper limit of one colonynever tried, bees split firstnever triednever triedNA-01
Herd quorum by visible posture or call (deer, sticklebacks)worksworksworks with a clearing where all can seefails, the vote cannot be seennever triednever triednever triednever triedNA-02
Uninformed buffer (golden shiners)fails, every member neededworksworks bestvulnerable to an inflamed middlevariablevariablevariablevariableNA-02
Vector compromise (olive baboons)worksworksworks with a formal rulefails, angles cannot be seennever triednever triednever triednever triedNA-02
Human face to face town meeting (Bryan, New England)worksworksworks best, under about 500 peopledegraded, vocal minorities dominatefailed, attendance below 2 per centnever triednever triednever triedPS-03

The chapter carries this in two sentences: a swarm reaches the size of a town by headcount but splits rather than federating, and nothing in the animal record runs a quorum above the size at which everyone can see the vote.

The case for and the case against

The case the chapter makes

Groups of animals and even cells reach good decisions without a ruler by a quorum, a threshold of independent judgements that the rest follow. It is faster than unanimity and needs no counter, unlike a majority. Shared decisions cost a group less than despotic ones in most conditions (Conradt and Roper). Uninformed members dilute an extreme minority (Couzin). Every public good breeds cheats and needs policing (worker bees, Pseudomonas). Leadership in wild groups tends to rest on knowledge, which is hard to seize (the wolf retraction, the matriarchs). The tool handed to a village is a threshold of independent judgements set in advance.

The case against

Philip Ball, in Critical Mass and How Life Works (NA-05): people have foresight, language and malice, and statistical laws of crowds assume identical parts. The naturalistic fallacy, voiced by the deep-time critics in SY-04 (Ellingson on the noble savage, SY-a principle 10): disease, infanticide and senilicide are as ancient as the swarm, and deep time is an inventory of survival adaptations, not an archive of political virtues. The swarm's unity rests on kinship and lethal policing (NA-01). Moving together may be shared preference, not following (Philosophical Transactions B critics, 2018, NA-02). Melanie Mitchell, Complexity 2009: nature's decentralised methods work through the death of countless failures (NA-05). Mansbridge: bees have unitary interests, villages adversary ones (NA-01). Michels and Freeman on hidden elites in structureless groups (NA-01, NA-05) were not used in the prose.

Where the sources disagree

The Appledore result is five trials. NA-01 calls it extraordinary accuracy. The chapter says a good record.

NA-01 says the swarm "unanimously" agrees and, a page later, that it "does not seek consensus". The quorum account is the one the chapter follows.

The 99 per cent figure for policed eggs rests on a single university page.

NA-05 attributes the lasR cheats appearing within 10 to 15 days to Diggle 2007. The figure comes from a PNAS 2015 study that NA-05 also cites for the cyanide. NA-05's claim that a 1 per cent share of cheats collapses the population looks garbled and was not used.

NA-05 praises the slime mould's resilience, but its own numbers give the mould a 14 per cent disconnection risk against 4 per cent for the rail lines. The chapter follows the numbers.

NA-02 calls the gorillas mountain gorillas and gives the Latin name of the western species. Not used.

NA-02's model values and equations for Couzin 2011 and Ward 2012 are blank in the PDF, and the authors of Ward 2012 differ between text and list.

NA-04 places Ellesmere Island in the Northwest Territories. It is in Nunavut, which is general knowledge, and the chapter says only the Canadian Arctic.

NA-02's claim that Venice went 529 years without internal capture rests on one paper and belongs to chapter 7.

PS-02 gives an interval for Dunbar's number from 3.8 to 520 (Lindenfors). The chapter does not use the number.

Groupthink as a whole is weakly evidenced (Aldag and Fuller 1993). Only directive leadership is a robust predictor (PS-03). The chapter says so.

Open questions

How long does a swarm take to choose? No report gives a figure.

Does the deer's 62 per cent describe a vote or a shared state of fatigue? The 2018 critics say the second is possible.

Does an uninformed buffer survive in a village where everyone hears the same news? NA-02 says the buffer is vulnerable at district scale and variable above. Nobody has tested it in people.

Do the sand-glass, the pebble ballot after a night's delay and the sunset rule work in a village? They are proposals in PS-03, not practices on record.

What is the quorum in the human sense, and is consent its better form? Chapter 12.

The documents

Primary: NA-01 (honeybee democracy, Lindauer, Seeley, worker policing), NA-02 (animal decisions, Conradt and Roper, Couzin, baboons), NA-04 (elders and the alpha wolf), NA-05 (control without a controller, Pseudomonas, harvester ants, slime mould, Ball), PS-03 (wisdom and madness of crowds, Galton, cascades, hidden profiles, groupthink). Consulted: SY-04 (the naturalistic fallacy), ME-03 (consent against consensus, for the turn to chapter 12), the PS documents (Bryan's town meetings, the warning on Dunbar's number).

Key works named in the reports: Lindauer 1955, Schwarmbienen auf Wohnungssuche. Seeley and Buhrman 2001, Behavioral Ecology and Sociobiology. Seeley 2010, Honeybee Democracy. Seeley, Visscher, Schlegel, Hogan, Franks and Marshall 2012, Science. Ratnieks 1988, The American Naturalist. Conradt and Roper 2003, Nature 421. Couzin et al. 2011, Science 334. Strandburg-Peshkin et al. 2015, Science 348. Ward et al. 2012. Schenkel 1947. Mech 1970, The Wolf, and 1999, Canadian Journal of Zoology 77. Diggle et al. 2007, Nature 450. Tero et al. 2010, Science 327. Gordon 2007, Behavioral Ecology 18. Ball 2004, Critical Mass, and 2023, How Life Works. Mitchell 2009, Complexity. Galton 1907, Nature, and Wallis 2014. Bikhchandani, Hirshleifer and Welch 1992. Stasser and Titus 1985. Janis, and Aldag and Fuller 1993. Mansbridge, on unitary against adversary interests.

Its documents in the Library

287 288 290 291 309 322 323 353