{"slug": "show-hn-terraforming-with-minimal-founders-method-for-diversity-from-two", "title": "Show HN: Terraforming with Minimal Founders – Method for Diversity from Two?", "summary": "A new open-source project, Founder Population Init, proposes that a tiny founding population of two diploid individuals could generate genetic diversity through a temporary generative inheritance system, called FounderCode, rather than transporting many finished genomes. The project's experiments, including the reproducible model in experiments/006.2_frequency_calibration, test whether such a mechanism can alter the neutral population-genetic consequences of an extreme founder event without hiding a large archive of donor genomes. The work raises questions for theoretical biology and demographic inference, though it does not establish that such a mechanism exists or is biologically practical.", "body_md": "Can a tiny number of living founders, equipped with a temporary generative inheritance system, become the`init()`\n\nfunction for an entire genetically robust population?\n\nImagine sending life to another world.\n\nThe obvious way to preserve a species is to bring as much of it as possible: many individuals, embryos, gametes, seed banks, genome libraries, carefully preserved diversity.\n\nBut there is another possibility.\n\nWhat if the smallest viable founding population did not need to carry all of the diversity a future population would require?\n\nWhat if it carried the ability to generate it?\n\nA handful of living founders could contain not merely genomes, but a temporary biological initialization system: a compact set of rules capable of producing controlled genetic diversity during the dangerous first generations of a new population. As that population grows, recombines and becomes genetically stable, the mechanism fades away and eventually disappears, leaving behind an ordinary evolving population.\n\nIn that picture, the founders are not an archive.\n\nThey are a seed for a population-generating process.\n\nFor future biological colonization, conservation, or long-duration propagation of life, this raises an information question: must we transport many finished genomes, or could some of that burden be replaced by generative biological machinery?\n\nFor theoretical biology, it asks how much population-level genetic structure must be explicitly stored and how much can arise from rules, randomness, recombination and ordinary inheritance.\n\nLooking backward, it also exposes a boundary in demographic inference. Population history is reconstructed from genetic data under assumptions about how inheritance, mutation and recombination work. If an early inheritance mechanism fell outside that model class, the demographic history inferred from descendants need not be identical to the literal history that produced them.\n\nNone of this establishes that such a mechanism exists, has existed, or is biologically practical.\n\nThis repository asks a narrower question and tries to break it experimentally:\n\nCan temporary generative inheritance substantially change the neutral population-genetic consequences of an extreme founder event, without simply hiding a large archive of donor genomes inside the model?\n\nPrefer a visual walkthrough first?\n[▶ Watch the video explainer on Youtube (generated with notebooklm)](https://youtu.be/f0Ad25UGnlw)\n\nA visual introduction to the core idea, the founder-bottleneck problem, the FounderCode hypothesis, and the experimental path leading to the final [ 006.2](/initsixdev/founder-population-init/blob/main/experiments/006.2_frequency_calibration) simulation. The explainer walks through the core idea, the population-genetics problem behind extreme founder events, the FounderCode hypothesis, and the experimental path that led to the final\n\n`006.2`\n\nsimulation.It is an accessible companion to the project. The source code, result files, experiment READMEs, and this README remain the authoritative research record.\n\nFor the current reproducible model, go directly to [ experiments/006.2_frequency_calibration/](/initsixdev/founder-population-init/blob/main/experiments/006.2_frequency_calibration).\n\nTwo ordinary diploid founders begin with only four chromosome copies at each locus. Population growth can reshuffle those copies, and ordinary mutation can slowly add new variants, but growth does not magically restore all of the standing variation lost at the founding event.\n\nFounder Population Init adds a temporary third ingredient:\n\n```\nfew living founders\n        +\nordinary inheritance\n        +\ntemporary generative germline\n        ↓\nstructured early diversity\n        ↓\npopulation grows\n        ↓\ngenerator shuts down\n        ↓\nordinary evolution only\n```\n\nThe hypothetical mechanism is called **FounderCode** or **FounderProgram**.\n\nIt should not be imagined as \"mutation turned up very high\" or as a secret library of hundreds of complete genomes.\n\nThe research question is whether **generative rules** can create useful population-level statistical diversity without explicitly storing all of the examples they can produce.\n\nIf you want to understand or try **one simulation**, use:\n\n```\nexperiments/006.2_frequency_calibration/\n```\n\nThis is the current integrated experiment and the main implementation to inspect, run, modify and challenge.\n\nIt includes the important pieces developed across the earlier experiments:\n\n- a neutral sequence-derived source population generated with\n`msprime`\n\n; - a 32-founder ordinary reference;\n- an ordinary two-founder control;\n- the\n**same exact two founders** plus FounderProgram; - ordinary mutation and recombination;\n- no external donor-haplotype bank;\n- no source allele-frequency table;\n- no source polymorphic-position list;\n- procedural per-rewrite haplotypes;\n- a neutral-theory\n`1/k`\n\nfrequency grammar; - a marginal-preserving correlation model for generated haplotypes;\n- feedback shutdown of FounderProgram;\n- blind\n`dadi`\n\ninference from descendant SFS only.\n\nThe earlier experiments are important because they show **how we got here**, including the failed v4 inference experiment. But they are research history, not the recommended starting implementation.\n\nFrom the 006.2 directory:\n\n```\nmamba activate founder-v6\nchmod +x run_v6_2.sh\n\n# Direct genetic-statistics preflight only\n./run_v6_2.sh --export-only\n\n# Full simulation + blind dadi inference\n./run_v6_2.sh\n```\n\nMain outputs are written under:\n\n```\nfounder_v6_2_out/\n├── v6_2_preflight.json\n└── v6_2_dadi_results.json\n```\n\nIf you are reviewing the project rather than reproducing its history, **read the root README, then 006.2, then go backward only when you want to understand why a design choice exists.**\n\nThe current computational evidence supports the following limited claim:\n\nIn simulation, a temporary embedded generative inheritance mechanism can substantially reduce the neutral population-genetic signature of an extreme two-founder bottleneck. This remains true after removing an explicit donor-genome reservoir and replacing it with procedural rules, although a detectable demographic and frequency-spectrum signature remains.\n\nThe completed v6.2 experiment strengthens the mechanism side of that claim:\n\nA meaningful part of neutral population-level diversity structure can be generated from general rules rather than copied from an empirical donor-genome archive. Blind demographic inference still detects a severe bottleneck/recovery signature, so the result is substantial repair and partial masking, not demographic invisibility.\n\nThis is a **theoretical / computational proof of principle**.\n\nIt is not evidence that:\n\n- a real biological germline could implement FounderCode;\n- generated neutral variants would preserve functional species viability;\n- two real humans could found a healthy human population this way;\n- such a mechanism existed in human history;\n- the 52-byte parameter payload is a complete biological genome generator.\n\nThe project did not begin with the final mechanism.\n\nInstead, each experiment removes a shortcut or attacks a weakness exposed by the previous one.\n\n| Experiment | Main question | Outcome |\n|---|---|---|\n`001_bootstrap` |\nCan temporary diversity generation rescue a two-founder toy population at all? | Positive toy proof of concept |\n`002_linked_genomes` |\nDoes the effect survive linkage, recombination, haplotypes and simple viability constraints? | Positive; `2 + init` resembles a much larger ordinary founding population |\n`003_blind_inference` |\nDoes the effect remain when the observer does not see FounderCode internals? | Positive direction; exact \"equivalent founder count\" shown to be poorly identifiable |\n`004_dadi` |\nWill an external demographic inference tool see the same effect? | Negative / invalid substrate; retained as a failure |\n`005_sequence_dadi` |\nDoes the effect survive on sequence variation generated by standard ancestry/mutation machinery? | Strong positive, but uses a donor haplotype bank |\n`006_embedded_generator` |\nCan we delete the donor bank and procedurally generate variation? | Positive; substantial repair, visible SFS scar |\n`006.1_generator_frequency` |\nDoes fresh procedural haplotype generation reduce the scar caused by template reuse? | Strong improvement |\n`006.2_frequency_calibration` |\nCan neutral-theory frequency rules improve the result without reading the target SFS? | Final positive proof-of-principle: strong direct SFS repair; blind `dadi` still detects a finite severe bottleneck/recovery scar |\n\nThe failures matter as much as the successes. In particular, v4 demonstrated that a sophisticated inference tool cannot make an invalid genetic substrate scientifically meaningful.\n\nLater experiments compare four arms:\n\n-\n`source_equilibrium`\n\nA neutral source population generated with`msprime`\n\n. This is the long-run reference. -\n`many_founder_reference`\n\nOrdinary descent from 32 diploid founders. -\n`ordinary_2`\n\nOrdinary descent from exactly two diploid founders. -\n`two_plus_generator`\n\nThe**same exact two founders** as`ordinary_2`\n\n, plus temporary FounderProgram activity.\n\nThe later v6-series uses matched random streams for ordinary mating, recombination, mutation and population sampling in the ordinary-two and generator arms. FounderProgram receives a separate random stream.\n\nThe intended comparison is therefore:\n\n```\nsame founders\nsame ordinary stochastic biology\nsame growth schedule\n-------------------------------\ndifference = FounderProgram\n```\n\nv6.2 changes only the FounderProgram's **frequency / LD generation law** relative to v6.1.\n\nIt does not read:\n\n- the source SFS;\n- source allele frequencies;\n- source polymorphic positions;\n- donor haplotypes;\n- the 32-founder reference.\n\nInstead it uses a theoretical neutral `1/k`\n\nfrequency grammar and a marginal-preserving Gaussian-copula construction for short-range correlation.\n\n| Metric | Source | 32 founders | Ordinary 2 | 2 + FounderProgram |\n|---|---|---|---|---|\n| Segregating sites | 16,755 | 14,467 | 7,256 | 14,208 |\n| Singleton fraction | 23.58% | 17.98% | 0.85% | 11.69% |\n| Doubleton fraction | 12.83% | 12.23% | 0.07% | 7.85% |\n| Common variants, MAF ≥ 0.25 | 26.91% | 31.09% | 64.58% | 39.58% |\nNucleotide diversity `π` |\n0.000488 | 0.000470 | 0.000363 | 0.000540 |\n\nFolded-SFS Jensen-Shannon divergence:\n\n```\nordinary 2 → source       0.31225\ngenerator  → source       0.06158\n\nordinary 2 → 32 founders  0.24372\ngenerator  → 32 founders  0.03135\n```\n\nOn this metric, the v6.2 generator population is roughly **5× closer to the source** and roughly **7.8× closer to the 32-founder reference** than ordinary two-founder descendants.\n\nThe result is not a perfect match. Rare variants remain underrepresented, common variants remain overrepresented, and `π`\n\nremains somewhat high.\n\nThat residual scar is part of the result, not something to tune away after seeing it.\n\nWith the current controller:\n\n```\ngeneration 1  FounderProgram active\ngeneration 2  FounderProgram active\ngeneration 3  inactive\ngeneration 4  permanently silenced\ngenerations 5–19 ordinary inheritance only\n```\n\nThe special mechanism therefore acts during the bootstrap and then leaves the population to ordinary inheritance.\n\n`dadi`\n\nreceives only the folded descendant SFS. It is not told founder count, pedigree, FounderProgram state, generated positions/haplotypes, controller history or source haplotypes.\n\nThe four arms behave differently:\n\n| Population | Best AIC model | Best AIC | Boundary behavior | Reading |\n|---|---|---|---|---|\n| Source equilibrium | `snm` |\n256.39 | none | Correct calibration: equilibrium looks like equilibrium |\n| 32-founder reference | `two_epoch` |\n311.42 | none | Ordinary finite size-change signal |\n| Ordinary 2 | `three_epoch` |\n3687.77 | best fit hits lower bounds |\nExtreme bottleneck magnitude is effectively unresolved |\n| 2 + FounderProgram | `three_epoch` |\n2034.84 | no best-fit boundary hits |\nStrong bottleneck/recovery remains detectable, but as a finite conventional history |\n\nFor ordinary two-founder descent, `three_epoch`\n\nand `two_epoch`\n\nare essentially tied (`ΔAIC ≈ 0.05`\n\n) and their best fits push bottleneck parameters to the optimization floor. The useful conclusion is therefore not the literal fitted magnitude; it is that the SFS demands an **extreme contraction beyond what these simple models resolve cleanly**.\n\nFor the FounderProgram descendants, `three_epoch`\n\nis preferred. Its best fit is interior:\n\n```\nnuB ≈ 0.00103\nnuF ≈ 0.949\nrecovery factor nuF/nuB ≈ 918\n```\n\nThose are **effective-size ratios inside the fitted model**, not literal founder counts.\n\nThe important contrast is:\n\n```\nordinary two founders\n→ pathological / boundary-limited extreme-bottleneck fit\n\nsame two founders + temporary FounderProgram\n→ finite, stable bottleneck/recovery fit\n```\n\nThis is **partial demographic masking**, not erasure of the founding event.\n\nThe direct SFS result and the blind inference result therefore tell a consistent but nuanced story:\n\nFounderProgram reconstructs a large amount of neutral population-level diversity and moves the descendants strongly toward ordinary many-founder/source populations, while a demographic observer can still detect that something like a severe bottleneck/recovery occurred.\n\nThe project uses several population-genetic summaries.\n\nAverage pairwise sequence difference per base pair.\n\nIt measures total diversity, but two populations can have similar `π`\n\nand very different frequency structures.\n\nThe distribution of allele counts across segregating sites.\n\nFounder bottlenecks strongly distort the SFS, especially by removing rare variants and leaving surviving variants at unusually high frequencies.\n\nVariants observed once or twice in the sampled chromosomes.\n\nThese are useful indicators of the rare-variant end of the SFS.\n\nHere, the fraction of segregating variants with minor allele frequency at least 0.25.\n\nExtreme founder populations tend to overrepresent common surviving variants.\n\nA symmetric measure of how different two normalized SFS shapes are.\n\nLower is closer.\n\n`dadi`\n\nreceives only the descendant folded SFS and fits ordinary demographic models such as equilibrium, population-size change, bottleneck/growth and three-epoch histories.\n\nIt is deliberately not told that FounderProgram exists.\n\nIts interpretation is model-dependent. A fitted bottleneck is not a literal founder count.\n\nThe information-theory motivation is simple:\n\n```\narchive\n  stores many explicit examples\n\ngenerator\n  stores rules capable of producing a distribution of examples\n```\n\nv5 still used an archive-like donor haplotype bank.\n\nv6 removed that bank.\n\nThe v6-series stores a **52-byte parameter payload**, but that number must not be misrepresented. The generative algorithm itself is additional information. Current experiments report implementation source size as a crude reminder of this cost.\n\nTherefore:\n\n`52 bytes of parameters`\n\nis not`52 bytes for a species`\n\n.\n\nThe deeper open question is the total irreducible information cost of:\n\n```\nfounder genomes\n+ algorithm\n+ parameters\n+ implementation machinery\n+ any required functional priors\n```\n\ncompared with transporting a conventional genetic archive.\n\nIf biological implementation were ever possible, a colonization or conservation architecture might trade some explicit archive size for generative machinery:\n\n```\nmany stored genomes\n         versus\nfew living genomes + generator\n```\n\nThis repository does not establish which architecture would be safer or smaller in reality.\n\nDemographic inference is conditional on its model of inheritance.\n\nThe project demonstrates the general model-boundary point computationally: if the true data-generating process includes an inheritance mechanism outside an ordinary-demography model, an observer can infer a conventional demographic history that is only the best explanation **inside the observer's allowed model family**.\n\nThis is not evidence for alternative human history.\n\nThe project asks whether preserving the **rules of a population distribution** can ever substitute for preserving a large collection of realized genomes.\n\nThat question survives even if the current biological mechanism remains fictional.\n\nThe current experiments focus primarily on **neutral/statistical genetic variation**.\n\nA real species also requires:\n\n- functional alleles;\n- essential conserved regions;\n- regulation and development;\n- deleterious and recessive-load management;\n- epistasis;\n- structural variation;\n- immune-system diversity;\n- fertility;\n- environmental adaptation.\n\nA generator that reproduces neutral population statistics is not automatically a generator for a viable organism or species.\n\nThose questions belong to future work.\n\n```\n.\n├── README.md\n├── experiments/\n│   ├── 001_bootstrap/\n│   ├── 002_linked_genomes/\n│   ├── 003_blind_inference/\n│   ├── 004_dadi/\n│   ├── 005_sequence_dadi/\n│   ├── 006_embedded_generator/\n│   ├── 006.1_generator_frequency/\n│   └── 006.2_frequency_calibration/\n└── docs/\n    └── learning-ledger/\n```\n\nEach experiment should remain reproducible and self-contained where practical. Negative results are retained.\n\nFor the scientific argument:\n\n```\n001 → 002 → 003 → 004 → 005 → 006 → 006.1 → 006.2\n```\n\nDo not skip `004_dadi`\n\n.\n\nIt is the point where the project learned that a flattering inference result is worthless when the synthetic data do not satisfy the assumptions of the measuring instrument.\n\nFor concepts rather than chronology, read the project learning ledger first.\n\n**Keep controls visible.****Change one important mechanism at a time when possible.****Define the desired direction before looking at downstream inference.****Separate direct measurements from model-dependent interpretation.****Do not interpret optimizer boundary hits literally.****Do not call effective population size a census founder count.****Do not confuse neutral diversity with functional biological viability.****Count algorithmic machinery when making information/compression claims.****Preserve negative results.****Prefer an ugly result that survives scrutiny to a beautiful result produced by hidden assumptions.**\n\n**Research state:** completed proof-of-principle research pass / exploratory computational biology.\n\n**Current endpoint:** v6.2 is complete, including blind `dadi`\n\ninference. This is the intended stopping point for the current research pass.\n\nThe main implementation to inspect and reproduce is:\n\n```\nexperiments/006.2_frequency_calibration/\n```\n\nThe project is intentionally allowed to stop here.\n\nFuture work could attack replication across many seeds, LD/IBD/ROH observables, functional viability, stronger forward-genetic models, biological implementation and information limits if there is a reason to reopen it. Those are extensions of the present result, not prerequisites for understanding what this research pass demonstrated.\n\nTreat the founders not merely as genomes, but as`init()`\n\nfor a population.", "url": "https://wpnews.pro/news/show-hn-terraforming-with-minimal-founders-method-for-diversity-from-two", "canonical_source": "https://github.com/initsixdev/founder-population-init", "published_at": "2026-08-16 19:56:31+00:00", "updated_at": "2026-08-16 20:10:42.259806+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "generative-ai", "ai-research"], "entities": ["Founder Population Init", "FounderCode", "FounderProgram"], "alternates": {"html": "https://wpnews.pro/news/show-hn-terraforming-with-minimal-founders-method-for-diversity-from-two", "markdown": "https://wpnews.pro/news/show-hn-terraforming-with-minimal-founders-method-for-diversity-from-two.md", "text": "https://wpnews.pro/news/show-hn-terraforming-with-minimal-founders-method-for-diversity-from-two.txt", "jsonld": "https://wpnews.pro/news/show-hn-terraforming-with-minimal-founders-method-for-diversity-from-two.jsonld"}}