Inside Noah's Ark: What the 4,000 Knowledge Packages Contain and Why They Were Chosen

Imagine you had to pack a library for a world that has lost most of its libraries. You could not take everything. You would have to choose what goes on board, and the choosing would say a lot about what you think a society needs to recover, and to stay worth living in.

That is the thought experiment behind "Noah's Ark," a pair of inventories in the NothingUnseen series. Set A (NU-SERIES-005) and Set B (NU-SERIES-006) each list 2,000 proposed knowledge packages, 4,000 in all, spread across 40 categories of 100 packages each. Set A starts with everyday foundations such as water, food, shelter, tools and energy, and sets them beside science, public institutions and cultural memory. Set B, in its own words, "adds 2,000 distinct complementary proposals, extending attention to care, communication, ecological observation, design, play and stewardship."

This article goes through what is actually in the two lists. It explains what a "knowledge package" is and what each field in an entry means. It walks through all 40 categories using real entries quoted by their identifiers. It shows how the packages depend on one another, sets out what the inventories leave out on purpose and how honest they are about their limits, and places the project next to real efforts to preserve seeds, languages, encyclopedias and machines for the long term. An FAQ closes the piece.

One point matters more than any other, so it comes first. These are proposals. The lists say plainly that no package has been acquired, licensed, tested or preserved. They are a starting inventory that can be checked and challenged. They are not an archive, and they are not a survival manual.

Why call it Noah's Ark?

The name comes from the flood story in the Book of Genesis, chapters 6 to 9. Noah is told to build a vessel and take his family and a pair of every kind of land animal on board, so that life can begin again after the waters go down. Scholars have long linked the story to older Mesopotamian flood narratives: the Sumerian Ziusudra, the Babylonian Atrahasis, and Utnapishtim in the Epic of Gilgamesh, which is the version closest to the biblical one.

The metaphor fits for three reasons, and it also has limits.

First, an ark is a selection under constraint. It is not a copy of the whole world. It is a deliberate, limited sample meant to be enough to start again. The Noah's Ark inventories accept that kind of limit openly. Each category gets exactly 100 packages. Set A says this quota "expresses a design decision, not a discovery that every subject deserves precisely the same space."

Second, an ark is about regeneration, not display. Animals were taken on board so they could breed and fill the land again, not to be looked at. In the same way, each package is meant to be usable: diagrams, worked examples and ways to notice failure, not just a title on a shelf.

Third, and this is where the metaphor bends, the inventories refuse to suggest that whoever loads the ark owns what is on it. A large share of the cultural entries say a community may decline to share some knowledge, and that a responsible archive may record only that the knowledge exists and how to ask permission. An ark that takes "two of everything" without asking would be the wrong model for culture. The lists say so repeatedly.

What is a knowledge package?

The introduction to Set A gives the most useful definition: "A package is larger than a title and smaller than an entire library." It is a bounded capability or cultural subject, with enough proposed contents that a future editor could commission or assemble something concrete: "diagrams with labels, worked examples, a glossary, a procedure's safe limits, a way to recognize failure, a reading selection or an account of competing interpretations."

Every one of the 4,000 entries has the same structure:

Field	What it records	What it does not mean
Identifier and title	A stable ID such as `A-W1-001` and a short name	Not a ranking; order within a category is not priority
Why	The editorial reason for including it	Not proof that it outranks every omitted subject
Proposed contents	What the assembled package would hold	Not a claim that the material already exists in one place
Dependencies	What a reader, workshop or institution needs first	Not certification that those prerequisites are met
Safety and rights	Review requirements, hazards, licensing and community permissions	Not clearance; it lists conditions still to be satisfied
Acquisition lead, not acquired package	A collection where an editor could start looking	Not permission to redistribute, and not evidence of acquisition

The first entry in Set A shows how this works. A-W1-001 — Household water balance is included because it "makes hidden daily demand visible before selecting equipment." Its proposed contents are an "end-use worksheet, seasonal demand scenarios, losses ledger and reserve calculation." Its dependencies are "Arithmetic; household census." Its safety field calls for "Public-health review" and says to "verify rights for each acquired work." Its acquisition lead points to the World Health Organization's water, sanitation and health program.

That is the format repeated 4,000 times. It is plain on purpose. It does not describe a water treatment recipe. It describes a way to understand a household's water needs before anyone makes decisions with consequences.

How the identifiers work

Every identifier starts with the set letter (A or B), then a block code (W1 to W4), then a three-digit number. In both sets each block covers five categories, or 500 packages. In Set A, for example, A-W1-001 to A-W1-500 run from water supply through energy, A-W2 covers the five science categories, A-W3 covers health, agriculture, administration, economics and preparedness, and A-W4 covers languages, literature, music, visual arts and social history. Set B follows the same pattern. All 4,000 identifiers are unique. The introductions describe them as "stable identifiers," so later corrections or crosswalks can refer to a specific package without confusion.

The forty categories at a glance

Set A (NU-SERIES-005)	Set B (NU-SERIES-006)
Water supply and sanitation	Textiles clothing and personal equipment
Food preparation and preservation	Transport navigation and field logistics
Shelter and building maintenance	Materials processing and repair
Hand tools and workshop practice	Measurement calibration and metrology
Energy access and household systems	Communications and information access
Mathematics and quantitative reasoning	Astronomy and space observation
Physics and physical models	Ecology conservation and field naturalism
Chemistry and matter literacy	Computation and software foundations
Biology and living systems	Engineering design and systems reasoning
Earth systems and local environments	Experimentation scientific reasoning and research methods
Public health and prevention literacy	Care accessibility and human development
Agriculture and managed landscapes	Comparative governance and conflict resolution
Administration records and civic institutions	Supply networks and operational coordination
Economics exchange and resource allocation	Education teaching and skill transmission
Safety risk and community preparedness	Ethics rights and collective decision making
Languages writing and translation	Humor play games and recreation
Literature storytelling and reading traditions	Oral traditions community memory and folklore
Music sound and performance	Architecture design and inhabited space
Visual arts craft and image literacy	Philosophies worldviews and intellectual traditions
Social history and historical reasoning	Archives preservation and cultural stewardship

Read across the rows and the two sets mirror each other in structure. Each opens with five practical categories, moves to five scientific or technical ones, then five about institutions and coordination, and ends with five cultural ones. The subjects inside those blocks are different, though.

Why Set B deliberately differs from Set A

Set B is not "Set A, continued." Its introduction is direct about this: "A second inventory can reveal what the first one treated as background. Foundations are indispensable, but a community also needs ways to teach, repair relationships, navigate, care for people, preserve plural traditions and enjoy being alive."

Comparing the blocks makes the design clear:

Practical block. Set A covers basic survival: water, food, shelter, tools, energy. Set B covers what makes those foundations usable over distance and time: clothing, transport, materials repair, measurement, and communications.
Science block. Set A covers the classic disciplines: mathematics, physics, chemistry, biology, earth science. Set B covers observation and method: astronomy, field ecology, computation, engineering design, and experimental reasoning.
Institutional block. Set A covers how a community keeps itself healthy, fed, recorded, trading and prepared. Set B covers how people care for each other, govern, coordinate supply, teach, and make ethical decisions together.
Cultural block. Set A covers languages, literature, music, visual art and history. Set B covers play and humor, oral tradition, architecture as lived space, philosophy, and the work of stewarding archives.

The introductions are also careful about how this came about. Both sets "were produced through one coordinated editorial process commissioned by Austin Little," developed "through coordinated AI-assisted editorial work with a common taxonomy." Set B "is not secretly the work of a different named model, a blind replication or a superior replacement." The experiment is "a comparison of editorial priorities, not a measured contest of artificial intelligences." That honesty matters. It would be easy, and false, to present two lists as independent minds converging on the truth.

Set B also warns that adding more slots does not fix blind spots: "More slots can still reproduce the same blind spots if the selection rules remain unexamined."

Category by category: what the packages contain

What follows is a tour of all 40 categories, with real entries quoted by their identifiers. Titles are reproduced exactly, and descriptions draw on each entry's own "Why," "Proposed contents," "Dependencies" and "Safety and rights" fields.

Set A, block one: everyday foundations

Water supply and sanitation

This category opens Set A, and its acquisition leads come almost entirely from the World Health Organization (80 entries) and the US Environmental Protection Agency (20). It focuses on observation and record-keeping rather than treatment chemistry.

A-W1-013 — Collection-point accessibility audit "includes users who cannot lift or reach standard taps," with reach diagrams, queue space, handle-force observations and user interviews. Its safety note is "User-led design."
A-W1-049 — Gravity-fed water distribution model shows "how elevation creates both opportunity and pressure limits," using elevation profiles, head-loss examples and pressure-zone diagrams, and requires "Engineer review before construction."
A-W1-097 — Water test result uncertainty notebook "preserves detection limits and sampling context alongside numbers," because a lab number with no context is easy to misread.
A-W1-025 — Taste and odor complaint log turns subjective complaints into traceable maintenance evidence. Its safety field is short and firm: "Do not taste suspect water."

The pattern is clear. The packages help people see, measure, record and ask the right professional. They do not hand out improvised purification methods.

Food preparation and preservation

Leads point to the US FoodSafety.gov portal and the University of Georgia's National Center for Home Food Preservation. Again, the entries draw boundaries.

A-W1-101 — Kitchen clean-to-dirty workflow "prevents preparation stages from contaminating ready-to-eat food," with counter diagrams, utensil movement and handwashing points. Its first dependency is "Safe water."
A-W1-137 — Water-bath canning suitability explains "that processing method depends on the food and recipe," and its safety field reads "No operational canning recipe." It depends on "Current tested recipes."
A-W1-173 — Recipe scaling by mass teaches baker-percent examples and yield corrections, with a sharp warning: "Preservation recipes not freely scalable."
A-W1-200 — Recipe version and correction ledger "prevents a copied error propagating through community cooking notes," and insists: "Never alter safety-critical instructions."

Shelter and building maintenance

Most leads go to the US National Park Service (78 entries), whose historic-preservation material is a natural source for maintenance and documentation, alongside ENERGY STAR and the Consumer Product Safety Commission.

A-W1-201 — Building condition baseline survey "distinguishes new deterioration from longstanding irregularities" through dated, room-by-room photographs.
A-W1-225 — Limewash heritage documentation preserves "compatible finish traditions without chemical mixing instructions," and depends on community consent and a conservator.
A-W1-273 — Retaining wall warning signs collects observation evidence "without recommending amateur stabilization." The safety note is two words: "Collapse hazard."
A-W1-300 — Building knowledge provenance register separates "measured facts from inherited assumptions," while protecting "security-sensitive plans."

Hand tools and workshop practice

This category leans on OSHA (95 entries), and its packages are about understanding tools and joints, not dangerous fabrication.

A-W1-313 — Chisel edge and bevel concepts connects edge geometry to cutting loads, with an "edge-damage atlas."
A-W1-349 — Assembly sequence planning "prevents a joint becoming inaccessible after another closes," using exploded views and a reversible dry-fit checklist.
A-W1-397 — Low-resource workshop capability map "prevents promising repairs that depend on missing processes," and its safety field rejects the fantasy of full independence: "No self-sufficiency claim."
A-W1-400 — Metric and imperial socket segregation is a small, practical package that stops near-fit sockets from rounding off fasteners.

Energy access and household systems

Leads go mainly to ENERGY STAR (65) and the Consumer Product Safety Commission (32). One entry here gets the most careful wording in Set A.

A-W1-401 — Household energy end-use ledger shows "which services consume scarce energy," with appliance inventories and an uncertainty worksheet. Its safety note is "No electrical work."
A-W1-449 — Battery aging recordbook links runtime changes to age and conditions and forbids "damaged-cell testing."
A-W1-497 — Controlled fire fundamentals and permitted use keeps "foundational heat literacy while making containment and restrictions inseparable." Its contents cover a combustion model, fuel moisture, ventilation and extinguishment diagrams, with "no accelerants or chemical recipes." It depends on "Current local fire authority guidance; trained adult supervision; lawful designated site." Its safety field calls it an "Educational proposal only, not validated instruction."
A-W1-500 — Energy equipment end-of-life routing keeps retired refrigerants and batteries out of ordinary waste.

Fire is perhaps the oldest technology in any "rebuild civilization" list, and this entry shows how the inventory treats risky foundational knowledge: as understanding wrapped in conditions, not as a how-to.

Set A, block two: scientific understanding

The five science categories share one safety statement: "Check item-specific rights and attribution before acquisition. Educational models only; qualified review required before physical, chemical, biological, environmental or engineering application. No hazardous operational protocols." Their leads go to open textbooks and courses: OpenStax, MIT OpenCourseWare, Chemistry LibreTexts, the US Geological Survey and NOAA.

Mathematics and quantitative reasoning

A-W2-001 — Place value with regrouping blocks "makes arithmetic reconstructible without a calculator," using base-ten diagrams and error cards. Its only dependency is "Counting."
A-W2-013 — Significant digits versus exact counts "avoids assigning false precision to measurements."
A-W2-049 — Matrices as linear transformations connects arrays of numbers to geometric actions through grid deformations.
A-W2-100 — Counterexamples to plausible mathematical shortcuts "trains verification rather than pattern trust," with division-by-zero traps and hidden-domain cases.

The category runs from counting to probability models such as A-W2-085 — Poisson counts and exposure units, and keeps coming back to verification.

Physics and physical models

A-W2-101 — Position-time graphs and motion stories starts at the start: separating location from speed.
A-W2-161 — Kirchhoff laws from conservation rebuilds circuit rules from charge and energy, using "low-voltage paper circuits."
A-W2-173 — Lens equations with real and virtual images reconstructs optical imaging "instead of memorising magnification."
A-W2-200 — Physical impossibility checks using conservation "screens attractive claims before expensive experimentation" with energy ledgers and momentum audits. This is a practical defense against perpetual-motion schemes in any era.

Chemistry and matter literacy

A-W2-201 — Elements, compounds and mixtures sorting atlas "prevents everyday appearance determining chemical identity."
A-W2-213 — Dilution as conserved-solute bookkeeping teaches concentration change "without laboratory recipes," using paper volume balances.
A-W2-285 — Detection limits versus quantified amounts separates evidence that something is present from a reliable measurement of how much. It links straight back to the water-testing notebook in block one.
A-W2-300 — Chemical claim verification dossier keeps a repeatable way to assess advertised material properties.

Biology and living systems

All 100 leads go to OpenStax.

A-W2-301 — Cell theory and the boundaries of living units explains organisms "through evidence rather than appearance," including acellular edge cases.
A-W2-337 — Hardy-Weinberg proportions as a baseline makes departures interpretable "without assuming real equilibrium."
A-W2-373 — Viruses and dependence on host cells explains acellular replication "conceptually without protocols," using generic lifecycle cartoons.
A-W2-400 — Comparative biology uncertainty notebook ties generalizations to observed species and conditions.

Earth systems and local environments

A-W2-425 — Earthquake waves and arrival-time evidence infers source information from P-S delays on seismograms.
A-W2-461 — Humidity, saturation and dew point separates relative percentages from actual water content.
A-W2-485 — Geomagnetic field and secular variation "distinguishes magnetic north from geographic north." Navigation in Set B depends on exactly this.
A-W2-500 — Natural-hazard maps and decision boundaries "keeps hazard probability distinct from individual engineering approval."

Set A, block three: health, land, institutions and preparedness

These categories share a safety formula ending "verify individual source rights before reuse," along with category-specific limits.

Public health and prevention literacy

Leads go to the WHO, the US National Library of Medicine and the CDC. The word "literacy" matters: none of these packages is medical advice.

A-W3-001 — Absolute and relative health risk "prevents impressive percentages from hiding small baseline effects." Safety: "Clinical review; no treatment advice."
A-W3-013 — Hand hygiene decision literacy preserves "when hygiene matters rather than ritualized overcleaning," and depends on "Safe water access; public-health guidance."
A-W3-037 — Mental-health help-seeking literacy makes "routes to support understandable before a crisis." Safety: "Not diagnosis or therapy."
A-W3-100 — Historical public-health lessons "prevents old interventions being copied without context," because historical material is "not current guidance."

Agriculture and managed landscapes

Leads go to the USDA National Agricultural Library and the FAO.

A-W3-101 — Farm observation notebooks retain "seasonal evidence that memory loses."
A-W3-113 — Manure nutrient planning links nutrient value with runoff and pathogen concerns, with "No application rates or untreated-food contact."
A-W3-149 — Farm succession knowledge transfer "keeps tacit site knowledge from disappearing" through field-history interviews.
A-W3-185 — Drought contingency for farms links early decisions to water, labor and animal needs.

Administration records and civic institutions

Leads point to the US National Archives, the General Services Administration and International IDEA.

A-W3-201 — Administrative record lifecycle keeps "routine records usable from creation through disposal."
A-W3-225 — Address systems and service access shows "how ambiguous locations exclude residents," with "no surveillance mapping."
A-W3-285 — Anti-corruption evidence literacy separates suspicious patterns from substantiated misconduct. Safety: "No unsupported accusations."
A-W3-300 — Civic institution reconstruction checklist retains essential functions "without assuming one national model," and depends on a "Community mandate; local legal expertise."

Economics exchange and resource allocation

Most leads go to CORE Econ, an open economics curriculum.

A-W3-301 — Opportunity cost in household decisions makes forgone alternatives visible "without reducing values to money."
A-W3-337 — Club goods and exclusion uses tool-library congestion and fee scenarios to explain membership-based shared services.
A-W3-349 — Barter and double coincidence shows transaction limits through multi-person exchange exercises.
A-W3-397 — Policy counterfactuals separates observed change from policy-caused change.

Safety risk and community preparedness

The American Red Cross is the main lead (73 entries), with the International Labour Organization next.

A-W3-401 — Household emergency contact plans keep "communication possible when ordinary channels fail," with out-of-area contacts.
A-W3-413 — Weather warning terminology distinguishes watch, warning and local action language, noting that "Definitions vary."
A-W3-473 — Fraud awareness after disasters helps people check unsolicited offers. Safety: "No scam scripts."
A-W3-500 — Spontaneous volunteer reception "prevents unassigned arrivals from obstructing established response."

Set A, block four: culture and memory

The cultural categories use a wider range of leads and the most detailed rights language. Common safety formulas include "Check every edition, image, recording and translation license before acquisition or reuse; no clearance claimed" (used 194 times across both sets) and "Community-led scope and permissions required; respect restricted, sacred and collective knowledge, attribution and refusal" (86 times).

Languages writing and translation

Leads include the International Phonetic Association, the Unicode Consortium, the Max Planck Institute for Evolutionary Anthropology, the Perseus Digital Library, Sefaria and Project Gutenberg.

A-W4-001 — Sound before spelling separates "audible contrasts from inherited letter habits," with mouth-position sketches and minimal-pair listening prompts.
A-W4-025 — Tibetan syllable stacks prevents stacked letters being mistaken for independent rows, and requires a Tibetan-literate reviewer.
A-W4-037 — Dialect boundaries and identity "keeps classification from overruling speakers' self-description."
A-W4-100 — Reader-tested translation colophon records "who checked a translation and what remains unresolved."

Literature storytelling and reading traditions

Project Gutenberg supplies most leads (64).

A-W4-101 — The Odyssey as return narrative preserves "a complex model of homecoming, recognition and unreliable telling," with translation comparison prompts.
A-W4-113 — Don Quixote reading and delusion examines "what stories do to expectations without reducing comedy to mockery."
A-W4-149 — Popol Vuh transmission dossier handles the major K'iche' Maya narrative through "attention to language, mediation and community authority," and depends on "Maya cultural and linguistic advisers."
A-W4-197 — Prison writing contextual reader "protects dignity when reading works produced under confinement."

Music sound and performance

Smithsonian Folkways Recordings is the main lead (73).

A-W4-201 — Pulse without a metronome gives an ensemble a shared reference "using bodies and attention."
A-W4-249 — Hindustani raga listening pathway "avoids reducing raga to a scale chart," using performer-approved examples.
A-W4-273 — Jazz collective conversation shows ensemble response "without equating improvisation with randomness."
A-W4-297 — Performer attribution ledger keeps "composers, arrangers, communities and performers from disappearing in reuse."

Visual arts craft and image literacy

The Metropolitan Museum of Art supplies 75 leads.

A-W4-301 — Looking before naming trains "sustained description before interpretation hardens into certainty."
A-W4-313 — One-point perspective as convention introduces a spatial system "without equating it with correct vision."
A-W4-349 — Arabic calligraphic composition avoids "reducing meaningful text to anonymous decorative shapes," with a practitioner's review.
A-W4-373 — Portrait photograph power relations makes "sitter agency, staging and circulation visible."

Social history and historical reasoning

Leads are spread widely: Europeana, the US National Archives, the South Asian American Digital Archive, the South African History Archive, the Qatar Digital Library and others.

A-W4-401 — Historical claim evidence ladder separates "what a record states from what a historian infers."
A-W4-413 — Historical maps and claims reads maps "as situated arguments rather than transparent territory."
A-W4-449 — Gulf pearling communities in records studies livelihoods "without romanticizing dangerous labor."
A-W4-500 — Calendars meeting in historical documents prevents date conversion "from erasing different ways communities organized time."

Set B, block one: making foundations portable and durable

Textiles clothing and personal equipment

Leads go mostly to Utah State University Extension (79), plus the Canadian Conservation Institute and iFixit.

B-W1-001 — Fabric fiber identification without burning supports care decisions "without destructive flame tests," using label glossaries and magnified examples.
B-W1-025 — Drawstring safety redesign keeps adjustability "while recognizing entanglement hazards," following current product-safety guidance.
B-W1-037 — Trouser knee patch design addresses "bending and abrasion rather than cosmetic coverage alone." Safety: "No PPE restoration claim."
B-W1-097 — Textile repair cost and time ledger makes repair feasibility visible "without equating low wages with value."

Transport navigation and field logistics

The leads mix the National Park Service, Park Tool, the FAA, NOAA's Office of Coast Survey and the Coast Guard's boating safety program.

B-W1-101 — Map scale and ground distance prevents "a copied map from retaining an invalid assumed scale," which is a real risk whenever maps are photocopied or reprinted at a different size.
B-W1-113 — Route card preparation preserves "decision points without requiring constant connectivity."
B-W1-149 — Bicycle chain wear measurement connects drivetrain longevity to measurable wear.
B-W1-137 — Trailhead transport coordination "prevents a successful walk ending without return transport."

Materials processing and repair

Leads go to the Canadian Conservation Institute (68) and iFixit (26).

B-W1-201 — Unknown material intake record prevents "repair methods being chosen from appearance alone."
B-W1-213 — Metal corrosion form atlas distinguishes uniform, pitting and crevice corrosion, with "No chemical removal instructions."
B-W1-225 — O-ring dimension records keeps cross-section and diameter distinct during replacement.
B-W1-300 — Object return-to-service evidence makes "cosmetic completion distinct from approved functionality," and forbids "self-certification of critical equipment."

Measurement calibration and metrology

Leads go to the UK National Physical Laboratory (74) and the US National Institute of Standards and Technology (23).

B-W1-301 — Measurand definition workshop prevents "measuring a convenient proxy instead of the intended quantity."
B-W1-313 — Ruler end wear and zero stops a damaged edge from shifting every measurement.
B-W1-337 — Measurement uncertainty budget makes "significant error sources visible rather than hidden."
B-W1-349 — Meniscus and parallax reading keeps viewing geometry in mind for ordinary volume measurements.

Metrology is one of the clearest places where Set B adds what Set A took for granted. Set A asks people to measure water, energy and food. Set B asks how anyone knows the measuring tool is right.

Communications and information access

Leads split between the American Radio Relay League (41) and the World Wide Web Consortium (41).

B-W1-401 — Radio link component map makes communication "depend on an entire chain rather than one handset."
B-W1-425 — Simplex versus repeater operation explains infrastructure dependence "in apparently direct communication."
B-W1-461 — Color-independent status communication keeps warnings readable "without color perception."
B-W1-500 — Information access continuity plan preserves "multiple ways to reach the same essential material" through a print, audio and digital matrix, with "No access or preservation guarantee."

Set B, block two: observation, computation and method

These five categories share the same "educational models only" safety statement as Set A's science block.

Astronomy and space observation

All 100 leads point to NASA Science.

B-W2-001 — Celestial sphere and observer horizons makes sky positions interpretable from different locations.
B-W2-013 — Stellar classification from absorption spectra uses "line patterns rather than visual colour alone."
B-W2-037 — Galactic rotation curves and missing-mass inference separates "observational discrepancy from a unique explanation."
B-W2-100 — Astronomical catalogues, epochs and provenance preserves coordinates and classifications across revisions.

Ecology conservation and field naturalism

Leads include OpenStax, the National Park Service and iNaturalist.

B-W2-101 — Habitat mosaics and ecological boundaries explains "why map categories hide transitional communities."
B-W2-161 — Temporary ponds and specialised lifecycles shows that "drying can be part of habitat function."
B-W2-173 — Minimum viable population misconceptions "rejects a single safe number for all species."
B-W2-185 — Plant identification from multiple characters avoids relying on one leaf or color feature, and includes "no-consumption warnings."

Computation and software foundations

Leads include MIT OpenCourseWare, the Python Software Foundation, the SQLite project and the RFC Editor.

B-W2-201 — Binary representation and positional encodings explains digital quantities "without equating bits with meaning."
B-W2-249 — Checksums and accidental corruption detection "separates integrity checks from authentication." That distinction matters to any digital archive.
B-W2-261 — Indexes and query-plan interpretation shows why faster reads can cost storage and writes.
B-W2-300 — Computational reproducibility audit kit checks "whether results survive a fresh environment," with environment manifests and output hashes.

Engineering design and systems reasoning

Leads go to NASA and MIT OpenCourseWare.

B-W2-301 — Needs, objectives and verifiable requirements keeps a desired outcome from being confused with a design choice.
B-W2-373 — Maintainability as access and diagnosis connects service time "with design choices, not worker skill alone."
B-W2-385 — Alarm floods and actionable information separates many warnings from useful diagnosis.
B-W2-397 — Root-cause analysis without single-cause shortcuts connects proximate failures with the system conditions that enabled them.

Experimentation scientific reasoning and research methods

Leads include NIST/SEMATECH's statistics handbook, the US Office of Research Integrity and Cochrane.

B-W2-401 — Research questions with observable consequences turns "broad curiosity into answerable comparisons."
B-W2-413 — Factorial experiments and interaction detection shows that "combined effects need not equal separate effects."
B-W2-461 — Causal diagrams and adjustment choices makes assumptions about pathways explicit using directed acyclic graphs.
B-W2-500 — Research estimands and the exact quantity sought prevents analysis from answering a different question from the stated objective.

Set B, block three: care, governance, coordination, teaching and ethics

Care accessibility and human development

Leads go to the WHO, UNICEF, the US Access Board and UNESCO.

B-W3-001 — Person-led support profiles "keeps preferences ahead of institutional assumptions."
B-W3-013 — Braille service navigation explains when qualified transcription and proofing are needed, and depends on Braille readers.
B-W3-085 — Participation without speech preserves contribution "through multiple modes."
B-W3-100 — Life-story work in care settings preserves "identity beyond service needs," with "no forced disclosure."

This category answers a gap the Set B introduction names directly: a community needs ways to "care for people," not only to feed and house them.

Comparative governance and conflict resolution

Nearly all leads (97) go to International IDEA.

B-W3-101 — Written and unwritten constitutional orders prevents constitutions "being equated only with one document."
B-W3-113 — Opposition rights preserves "accountability when electoral majorities govern."
B-W3-137 — Consensus decision rules explains agreement "without pretending conflict disappears."
B-W3-149 — Active listening in disputes preserves understanding "without implying agreement." Safety: "No manipulative persuasion."

Supply networks and operational coordination

Ninety-nine leads go to the Logistics Cluster, the humanitarian logistics coordination body.

B-W3-201 — Supply network mapping reveals "dependencies beyond the immediate supplier."
B-W3-225 — Reorder point reasoning connects demand and lead time to replenishment, with "No optimal-stock guarantee."
B-W3-237 — Cold-chain excursion governance separates a recorded deviation from product release decisions.
B-W3-300 — Supply system exit planning "prevents temporary programmes leaving unmanaged stocks and obligations."

Education teaching and skill transmission

UNESCO supplies 90 leads.

B-W3-301 — Learning objective alignment keeps "lessons, practice and assessment aimed at the same capability."
B-W3-313 — Transfer of learning distinguishes "task familiarity from application elsewhere."
B-W3-373 — Open educational resource adaptation makes "permissions and changes traceable." This is the same rights discipline the inventories apply to themselves.
B-W3-397 — Teaching evidence appraisal separates promising studies from universal classroom rules, with "No neuroscience myths."

Education is arguably the category that makes all the others usable. A package nobody can teach is a package nobody inherits.

Ethics rights and collective decision making

UNESCO supplies 97 leads.

B-W3-401 — Facts and values in public choices prevents "empirical uncertainty from concealing moral disagreement."
B-W3-425 — Withdrawal and reversibility makes exit options part of ethical design.
B-W3-437 — Ethics of uncertainty communication makes "omissions and confidence levels morally consequential." Safety: "No false reassurance."
B-W3-500 — Ethical omissions audit asks "whose interests disappear from a decision frame."

Set B, block four: play, memory, place, thought and stewardship

Humor play games and recreation

The Strong National Museum of Play supplies 92 leads. This is the category most likely to surprise readers of survival lists, and that is intended. The Set B introduction lists "enjoy being alive" among a community's needs.

B-W4-001 — Play agreements before competition "makes opting out and changing difficulty normal parts of recreation."
B-W4-013 — Mancala family comparison avoids "treating many distinct sowing games as one uniform tradition."
B-W4-025 — Senet interpretation limits preserves fascination "while admitting that ancient rules are incompletely known," with an "explicitly hypothetical play model."
B-W4-100 — Game ending and repair ritual helps a gathering "recover from frustration without diagnosing participants."

Oral traditions community memory and folklore

Leads go to the Oral History Association, Smithsonian Folkways and the South Asian American Digital Archive.

B-W4-101 — The right not to tell "makes refusal part of a healthy storytelling gathering."
B-W4-149 — Language-loss stories without shame keeps grief and adaptation "from becoming accusations against individuals."
B-W4-161 — European fairy-tale collector mediation makes "editors and collectors visible between tellers and printed books."
B-W4-197 — An interview that stays private preserves "the value of listening even when nothing enters a public collection."

Architecture design and inhabited space

Archnet supplies 64 leads.

B-W4-201 — Reading a building as inhabited space starts architectural understanding "with use rather than stylistic labels alone."
B-W4-225 — Madrasa spatial learning histories reads architectural arrangements "without claiming they determine teaching outcomes."
B-W4-249 — Tenement space and social judgment avoids "romanticizing deprivation or repeating stigmatizing housing descriptions."
B-W4-300 — Places intentionally left undescribed "makes non-disclosure a legitimate outcome of cultural interpretation."

Set A's shelter category asks whether a building is sound. This category asks how people live in it.

Philosophies worldviews and intellectual traditions

Leads include Project Gutenberg, the Göttingen Register of Electronic Texts in Indian Languages, Sefaria and the Perseus Digital Library.

B-W4-301 — Plato's cave as contested image keeps a famous allegory connected to its dialogue "rather than motivational slogans."
B-W4-325 — Pirkei Avot commentary plurality makes "layered interpretation visible around short inherited sayings."
B-W4-349 — Mencius and cultivated dispositions preserves argument and story without pretending a historical theory settles psychology.
B-W4-397 — Comparative philosophy with asymmetry notes prevents "familiar terms from deciding in advance what another tradition must mean."

Archives preservation and cultural stewardship

The Northeast Document Conservation Center supplies 41 leads. This final category turns the inventory back on itself.

B-W4-401 — Cultural collection purpose charter defines why materials belong together "without claiming ownership of a culture."
B-W4-449 — Glass photographic plates caution protects fragile images through supported professional assessment, with a "no-cleaning boundary."
B-W4-473 — Orphan recordings uncertain authorship prevents "missing attribution being mistaken for free reuse."
B-W4-500 — Preservation limits openly declared is the last of the 4,000 entries, and it "ends with an honest boundary between selected knowledge and an actually secured cultural archive."

The whole project closes on a package about admitting what has not been done.

How the packages depend on each other

Set B's introduction has a section titled "Distinct does not mean disconnected." It explains: "Scientific observation depends on measurement; teaching depends on language; a repair may require materials knowledge. Deduplication should remove redundant selections, not pretend that knowledge falls into isolated boxes."

The dependency fields make those connections traceable. A few chains stand out.

Water, then health, then food

The clearest chain starts with water. Set A's water category explains how to understand supply, contamination and storage. The public health category's A-W3-013 — Hand hygiene decision literacy lists "Safe water access" as its first dependency. In food, A-W1-101 — Kitchen clean-to-dirty workflow depends on "Safe water; kitchen plan," and A-W1-113 — Cereal cooking water ratios needs a "Pot; heat source; safe water." Nine entries across Set A name "safe water" in their dependencies, including A-W1-054 — Handwashing station usability, A-W1-117 — Leafy-green handling chain, A-W1-122 — Steamed staple preparation and A-W1-154 — Rehydrating dried staples.

The order matters. A cooking package is not much use if the water is unsafe. A hygiene package means little without reliable water to wash with.

Counting, then arithmetic, then everything quantitative

A-W2-001 — Place value with regrouping blocks depends only on "Counting." Arithmetic then appears as a dependency in 41 entries across both sets, from A-W1-001 — Household water balance to A-W1-173 — Recipe scaling by mass and A-W1-401 — Household energy end-use ledger. Statistics or statistical reasoning appears in 76 dependency fields. A-W1-097 — Water test result uncertainty notebook, A-W2-285 — Detection limits versus quantified amounts and A-W2-300 — Chemical claim verification dossier all rely on it. Set B's metrology and research-methods categories rest on the same base.

Magnetism, then maps, then routes

A-W2-485 — Geomagnetic field and secular variation explains magnetic versus geographic north. B-W1-101 — Map scale and ground distance depends on a "Ruler; verified map." B-W1-113 — Route card preparation depends on "Current route information." Navigation in Set B uses the earth science in Set A.

People before records

The most common kinds of dependency are not tools at all. "Consent" appears in 182 dependency fields, "community" in 129, and "qualified" (as in a qualified reviewer, adviser or professional) in 231. A-W1-225 — Limewash heritage documentation needs "Community consent; conservator." A-W4-149 — Popol Vuh transmission dossier needs "Maya cultural and linguistic advisers." B-W3-001 — Person-led support profiles needs "Person consent; trusted supporter."

This may be the inventories' most important structural claim. Much of the knowledge a society needs cannot be pulled out of people. It can only be preserved with them.

A simple dependency ladder

Layer	Example packages	Depends on
Base skills	A-W2-001 Place value with regrouping blocks	Counting
Household systems	A-W1-001 Household water balance; A-W1-401 Household energy end-use ledger	Arithmetic; records
Health and food	A-W3-013 Hand hygiene decision literacy; A-W1-101 Kitchen clean-to-dirty workflow	Safe water
Measurement and method	B-W1-337 Measurement uncertainty budget; B-W2-401 Research questions with observable consequences	Statistics; scientific literacy
Coordination	B-W3-225 Reorder point reasoning; A-W3-401 Household emergency contact plans	Demand history; verified contacts
Transmission	B-W3-301 Learning objective alignment; A-W4-100 Reader-tested translation colophon	Subject expertise; translator agreement
Stewardship	B-W4-401 Cultural collection purpose charter; B-W4-500 Preservation limits openly declared	Community consultation; team review

What is deliberately left out

The inventories are as notable for what they refuse to include as for what they include.

No hazardous operational recipes. Set A says: "The entries deliberately do not provide hazardous operational recipes or claim successful physical rebuilding tests." The safety fields repeat this in many forms: "No operational canning recipe," "No pesticide instructions," "No electrical work," "No improvised dosing," "No chemical mixing," "No pathogen work or disease concealment." The science categories all carry "No hazardous operational protocols." Even the virus entry, A-W2-373 — Viruses and dependence on host cells, explains replication "conceptually without protocols."

No medical, legal or financial prescriptions. Health packages are "literacy," not treatment. A-W3-049 — Cardiovascular prevention discussions helps people "prepare for risk conversations rather than prescribe themselves." Economics packages carry notes such as "Not personal financial advice" and "Not contract drafting advice."

No self-certification. Entries such as B-W1-300 — Object return-to-service evidence forbid "self-certification of critical equipment," and A-W1-249 — Storm shutter compatibility assessment offers "No wind-resistance guarantee."

No taking sacred or restricted culture. Eighty-six entries carry the formula "Community-led scope and permissions required; respect restricted, sacred and collective knowledge, attribution and refusal." Some packages, such as B-W4-300 — Places intentionally left undescribed and B-W4-197 — An interview that stays private, exist specifically to preserve the right not to put something in the archive.

No claim of completeness. B-W4-400 — Worldview plurality shelf statement makes "selection limits explicit without claiming exhaustive representation of human thought," and includes a "missing-tradition register." B-W3-500 — Ethical omissions audit asks whose interests are missing.

No reconciliation with the earlier list. Set A explains that "The historical 200-item Vesper Quill candidate list could not be recovered through the accessible routes used for this task." The new inventories are "explicitly unreconciled with that earlier list." If it turns up, a later crosswalk should "preserve it, map exact and substantive overlaps, and record additions and exclusions rather than silently overwriting the older work."

The honest limits

The inventories are unusually frank about what they are not. Here are the main limits in their own terms.

Limit	What the inventories say
Proposed, not acquired	"None of those fields means the package has been acquired, licensed, tested or preserved."
Leads are not packages	A linked collection "is not a promise that one ready-made document there contains the exact custom package described here."
Access is not a license	"Public access alone is not a redistribution license." A lead "is neither permission to redistribute an entire collection nor evidence that an acquisition has already happened."
Selection is not readiness	A package about water, health, electricity or workshop practice "is not itself a safe instruction manual."
Culture needs permission	"A cultural package is not permission to detach a story, image, ritual or recording from its community."
Quotas are choices	100 per category "expresses a design decision."
Order is not priority	Titles' "order is not a universal priority ranking."
Local needs vary	"A society's immediate needs depend on its environment, available tools, languages, institutions and people."
Not an AI contest	The sets are "a comparison of editorial priorities, not a measured contest of artificial intelligences."

There is also a structural limit the lists acknowledge only indirectly. Their acquisition leads lean heavily on US federal agencies and English-language institutions, especially in the practical and institutional categories: the EPA, FoodSafety.gov, the National Park Service, OSHA, ENERGY STAR, the CDC and the US National Archives. The cultural categories are noticeably more international, with Sefaria, the Qatar Digital Library, the National Diet Library of Japan, Te Papa Tongarewa and the South African History Archive. Even so, a reader in another country would need to substitute local standards and authorities. The introduction's warning that "a fixed inventory cannot represent every community adequately" applies here.

What should happen next

Both introductions describe the same next step. Set A: "select a small pilot, acquire materials lawfully, document dependencies, test whether another reader can use them, and revise the package after failures." Set B adds tests: "Could they locate the prerequisites, understand the diagrams, recognize the limits and identify a missing step? Could they explain whose perspective the package represents and whose it leaves out?"

That standard is stricter than "does the archive exist." It asks whether a stranger can actually use what was saved.

How Noah's Ark compares with real preservation projects

The inventories are a paper exercise, but they belong to a real tradition of trying to carry knowledge and living material across time and disaster. Each precedent below shows a different side of the problem.

The Svalbard Global Seed Vault: backup, not replacement

The Svalbard Global Seed Vault opened on 26 February 2008 in Norway's Arctic archipelago, with its storage area more than 100 meters inside a mountain. The rooms are kept at minus 18 °C, and the vault can hold 4.5 million seed samples. The Norwegian government owns it, NordGen runs it, and the Crop Trust provides most of the annual operating funding. Its own website reports 1,417,799 seed samples from 135 depositors, covering 6,587 species.

Two features matter for Noah's Ark. First, the vault stores duplicates. It is a backup for genebanks around the world, not a replacement for them. Second, depositors keep ownership of their seeds, and the vault "functions like a safe deposit box in a bank." The model was tested in 2015, when the research center ICARDA withdrew backup samples to rebuild collections after the Syrian war disrupted its genebank at Tel Hadya.

The Noah's Ark inventories reason the same way. They point to existing institutions as sources and insist on keeping provenance and permissions. They do not claim to own what they would save.

The Long Now Foundation's Rosetta Project: languages in nickel

The Rosetta Project, run by the Long Now Foundation, describes itself as "a global collaboration of language specialists and native speakers" building a public library of human languages. Its Rosetta Disk is a three-inch nickel disc holding more than 13,000 pages of documentation in over 1,500 languages, etched so small it must be read under a microscope at 650x. A prototype with 6,500 pages flew aboard the Rosetta comet mission, launched on 2 March 2004, and ended with the spacecraft's hard landing on comet 67P/Churyumov–Gerasimenko in September 2016.

Set A's language category is close in spirit. Packages such as A-W4-001 — Sound before spelling and A-W4-037 — Dialect boundaries and identity take seriously the idea that languages are carried by speakers and can be lost. Several of the language leads point to the Max Planck Institute for Evolutionary Anthropology, a major center for linguistic documentation.

The Arch Mission Foundation's Lunar Library: a backup off-world

The Arch Mission Foundation built the Lunar Library, a 30-million-page archive on 25 nickel discs, each 40 microns thick, made by NanoArchival. Four analog layers hold more than 60,000 page images readable with magnification. Twenty-one digital layers each hold a DVD master. The contents included English Wikipedia, tens of thousands of books as PDFs, the Long Now Foundation's PanLex datasets covering 5,000 languages, and the Wearable Rosetta disc. The library flew on SpaceIL's Beresheet lander, which crashed on the Moon in April 2019. The foundation says its scientific advisers believe the library survived the crash and is intact, based on NASA Lunar Reconnaissance Orbiter imagery.

The Lunar Library shows how far people will go to place a copy beyond Earth's risks. It also shows a limit that Noah's Ark takes seriously: a copy nobody can reach or decode helps nobody. That is why the Lunar Library included "technical specifications for decoding the digital layers," and why Set B includes packages such as B-W2-201 — Binary representation and positional encodings and B-W2-249 — Checksums and accidental corruption detection.

Kiwix: offline knowledge that people actually use

Kiwix was started in 2007 by Emmanuel Engelhart and Renaud Gaudin to make Wikipedia available offline. It compresses websites into indexed ZIM files that are much smaller but still searchable. A server version, Kiwix-serve, lets one computer share content over a local network, and Kiwix-hotspot turns small plug computers into plug-and-play Wi-Fi libraries. Kiwix has been used on Raspberry Pi installations in Tanzanian schools without reliable electricity, in European and US prison education programs, and aboard ships in Antarctic waters. The organization's own about page says it "brings the world's knowledge to places where the internet can't reach."

Kiwix is the closest real counterpart to Set B's communications category. B-W1-500 — Information access continuity plan asks for "multiple ways to reach the same essential material." Kiwix is one working answer: knowledge that does not depend on a live connection.

Lewis Dartnell's The Knowledge: the rebuild thought experiment

Astrobiologist Lewis Dartnell's The Knowledge: How to Rebuild Our World from Scratch was published by The Bodley Head on 3 April 2014, with a US edition from Penguin Press. It asks what knowledge survivors of a global catastrophe would need to rebuild. Dartnell describes a "grace period" in which survivors could "salvage food, materials and tools from the ruins of society" before having to produce their own food, tools and energy and maintain hygiene and health. The book covers agriculture, food and clothing, substances, medicine and transport, and argues that applying the scientific method to basic knowledge could let an advanced society "reappear within several generations."

Noah's Ark overlaps heavily with Dartnell's practical subjects: water, food, shelter, energy, textiles, materials, transport. It shares his emphasis on method, seen in Set B's whole category on experimentation and reasoning. It differs in two ways. It refuses to give procedures, where a book can explain processes in prose. And it gives play, oral tradition, philosophy, care and archival ethics as much space as technology.

Open Source Ecology's Global Village Construction Set: machines as a kit

Open Source Ecology was founded in 2003 by Marcin Jakubowski, with its prototyping base at Factor e Farm in Maysville, Missouri. Its Global Village Construction Set is an open-source platform for building the 50 industrial machines the group considers necessary for a small, self-sufficient community, "a lifesize, scalable, modular LEGO construction set." The first machine, a Compressed Earth Brick Press, was built in 2007. Jakubowski presented the project in a 2011 TED Talk. The project's own page said in 2018 that the machines were "about 1/3 done."

The GVCS is the strongest real example of a fixed list of capabilities meant to rebuild a working economy, so it is an obvious reference for Noah's Ark. The contrast is useful. Open Source Ecology builds and tests physical machines and measures success by "demonstrating a replicable, open source, social production model for a given machine." Noah's Ark is explicitly earlier in the process. Its entries "do not... claim successful physical rebuilding tests," and A-W1-397 — Low-resource workshop capability map carries the note "No self-sufficiency claim."

Memory of Mankind: ceramic tablets in a salt mine

Memory of Mankind was founded in 2012 by ceramicist Martin Kunze in the salt mine at Hallstatt, Austria, described as the oldest salt mine continuously worked by humans. It stores text and images on ceramic tablets 20 by 20 cm, each holding up to 5 million characters, which resist temperatures up to 1,200 °C as well as chemicals, water and radiation. Its aim is to "preserve knowledge about present human civilization from oblivion and collective amnesia." The project's site describes a plan for token holders to gather every 50 years, starting in 2070, to keep the memory of the archive alive.

Memory of Mankind is mainly a portrait of an era: personal stories, poems, newspaper articles. That links it to Set B's oral tradition and archive categories. Its plan to keep people remembering where the archive is matches a point Noah's Ark makes repeatedly: an archive is a social practice, not just a durable object. B-W4-500 — Preservation limits openly declared and A-W3-497 — Preparedness plan revision cycles, which "prevents a completed binder from becoming obsolete," both depend on people continuing to look after the work.

The Internet Archive: universal access as a goal

The Internet Archive was founded in May 1996 by Brewster Kahle, with Bruce Gilliat. Its stated mission is "universal access to all knowledge." The Wayback Machine opened archived web pages to the public in 2001, and in October 2025 the Archive announced it had archived one trillion web pages.

The Internet Archive is the opposite strategy to Noah's Ark: collect as much as possible and sort it later. Both approaches are needed. A huge archive preserves what nobody thought to select. A curated inventory makes sure the essentials are findable, explained and connected. The Noah's Ark leads point to many large open collections, including Project Gutenberg, Europeana, MIT OpenCourseWare and OpenStax, because a curated package has to be built from somewhere.

Precedents at a glance

Project	What it preserves	Medium or method	Main lesson for Noah's Ark
Svalbard Global Seed Vault	Seed duplicates from genebanks	Frozen storage inside a mountain	Back up, don't replace; depositors keep ownership
Rosetta Project (Long Now)	Documentation of 1,500+ languages	Micro-etched nickel disk	Language is foundational and fragile
Lunar Library (Arch Mission)	30 million pages incl. Wikipedia	Nickel discs on a lunar lander	Include decoding instructions
Kiwix	Wikipedia and other sites offline	Compressed ZIM files, local servers	Access without connectivity
The Knowledge (Dartnell)	Principles for rebuilding	A book-length thought experiment	Method matters as much as facts
Global Village Construction Set	50 open-source machines	Open designs and built prototypes	Testing is what makes a list real
Memory of Mankind	Portrait of the present era	Ceramic tablets in a salt mine	Archives need people who remember them
Internet Archive	The web, books, media	Large-scale digital collection	Breadth complements curation

Set C is in preparation

A third inventory, Set C, is being prepared with 20 more categories. Its contents have not been described here and should not be assumed. Based on how the first two sets describe themselves, any further set is expected to follow the same discipline: stable identifiers, explicit dependencies, safety and rights fields, and a clear statement that proposals are not acquisitions.

Frequently asked questions

Is Noah's Ark an actual archive I can download?

No. Both inventories describe themselves as proposals. Set B says that none of the fields "means the package has been acquired, licensed, tested or preserved." The acquisition leads point to real collections, such as the World Health Organization, OpenStax, Project Gutenberg and the Smithsonian's Folkways label, where an editor could begin lawful acquisition.

What exactly is in a single package?

A title and stable ID, a reason for inclusion, proposed contents (diagrams, worked examples, glossaries, reading selections and so on), dependencies, safety and rights conditions, and an acquisition lead. In the inventory's words, it is "larger than a title and smaller than an entire library."

Why exactly 100 packages per category?

The categories were allocated before the entries were written. The introduction says this makes their "balance inspectable," while admitting that a quota "expresses a design decision, not a discovery that every subject deserves precisely the same space."

Is Set B a check on Set A by a different AI?

No. Both were produced in one coordinated, AI-assisted editorial process commissioned by Austin Little, using a common taxonomy. Set B "must not be attributed to another model or presented as an independent replication of Set A." Its categories were chosen deliberately to complement Set A's.

Why does a survival-minded list include games, humor and philosophy?

Because the editors argue that a recovering community needs more than food and shelter. Set B says a community also needs ways to "teach, repair relationships, navigate, care for people, preserve plural traditions and enjoy being alive." Packages such as B-W4-001 — Play agreements before competition and B-W4-100 — Game ending and repair ritual are about keeping a group together.

Can I use these entries as instructions?

No. Set A says a package about water, health, electricity or workshop practice "is not itself a safe instruction manual," and real implementation "may require trained people, local standards, suitable materials, validated measurements and professional review." The entries deliberately leave out hazardous procedures.

Why so much attention to consent and refusal?

Because cultural knowledge is not freely available just because it is visible. "A cultural package is not permission to detach a story, image, ritual or recording from its community." Some knowledge "may properly remain restricted," and an archive "can record that a resource exists and how permission should be sought without taking possession of its contents."

Why are so many leads American institutions?

Many of the practical and civic leads are US agencies, such as the EPA, OSHA, the CDC and the National Park Service, while the cultural categories draw on institutions worldwide. The introductions acknowledge that needs depend on local environment, tools, languages and institutions, so readers elsewhere would need to substitute their own standards and authorities.

What happened to the earlier 200-item list?

The historical "Vesper Quill" candidate list "could not be recovered through the accessible routes used for this task." Sets A and B are new and "explicitly unreconciled" with it. If it is recovered, the plan is to keep it intact and build a crosswalk rather than overwrite it.

How would anyone know whether a package works?

By testing it. The proposed pilot gives a small number of packages to readers who did not help choose them and checks whether they can find the prerequisites, understand the diagrams, recognize the limits and spot a missing step. Until that happens, the inventory "supplies no such success claim."

How is this different from the Svalbard Global Seed Vault?

The Seed Vault physically stores more than a million seed duplicates and has actually returned samples to a depositor, ICARDA, after the Syrian war. Noah's Ark is still a plan for what knowledge to gather. Both are built on the idea of backing up rather than replacing, and both treat ownership and provenance seriously.

Closing: an ark that admits what it has not yet loaded

The most useful thing about Noah's Ark is not any single entry, though many are thoughtful: a water-test notebook that keeps detection limits, a fire package that ties heat literacy to containment, a storytelling package about the right not to tell. The useful thing is the shape of the whole. It is an inventory that shows its reasoning, writes down its dependencies, puts safety and rights in every entry, and ends with B-W4-500 — Preservation limits openly declared.

Real preservation projects show what it takes to move from a list to an archive: seeds frozen in a mountain, languages etched in nickel, encyclopedias compressed for offline use, machines built and tested on a farm, tablets stored in a salt mine, a trillion web pages saved. Noah's Ark has not done any of those things yet. It describes, in 4,000 specific and checkable entries, what such work could carry and what conditions it would have to meet first.