The human genome went from $2.7 billion to $100. Biology is now software.
For nearly four billion years, the central drama of life on Earth was the silent operation of a code no one could read. The instructions that distinguished a bacterium from a redwood from a human were written in DNA, but they ran the way machine language runs on a chip with no debugger — executed, never inspected. Ray Kurzweil's claim that "biology becomes information technology" marks the moment that situation inverted. Once the genome could be read, written, and copied as data, life ceased to be only a substance and became, in part, a file format. This is less a single dated event than a threshold crossing, and Kurzweil's framing — drawn out across The Singularity Is Near (sv-singularity-near) — is that the crossing obeys the same exponential logic he calls the Law of Accelerating Returns (sv-kurzweil-law).
The thesis only makes sense because life was already digital long before any human noticed. The four-letter nucleotide alphabet is a genuine information system, laid down at the Origin of Life (sv-origin-of-life) and elaborated across every later innovation in the biological story — the metabolic rewiring of the Great Oxygenation Event (sv-great-oxygenation), the modular complexity of the First Complex Cells (sv-first-complex-cells), and the recombinational gene-shuffling unleashed by the Invention of Sexual Reproduction (sv-invention-of-sex). Evolution, in Kurzweil's reading, is itself an information-processing algorithm — slow, blind, but accumulating. What changed with the Human Genome Project was not that biology became information, but that humans finally gained read-access to a repository written by Charles Darwin's mechanism of descent with modification (sv-charles-darwin).
Kurzweil's favorite illustration is the Genome Project itself. Launched in 1990 as a fifteen-year, multi-billion-dollar effort, it had sequenced only about 1 percent of the genome by 1997 — halfway through its schedule. Critics declared it a failure: at that rate, completion lay seven centuries away. Kurzweil's response was that 1 percent, with capacity doubling annually, meant the project was effectively halfway done — seven doublings from completion. It finished in 2003, on time. The cost curve that followed is among the steepest in the history of technology: roughly a billion dollars per genome in 2003 collapsing toward a thousand dollars within two decades, a millionfold drop that outpaced Moore's Law itself. This is the empirical spine of the claim — the same exponential signature that connects the deep past to the projected future.
Treating biology as a manipulable medium reframes every Kurzweilian prediction downstream. If the genome is software, then aging and disease are, in principle, debuggable — the conceptual foundation of Longevity Escape Velocity (sv-kurzweil-lev), where medical progress outruns the clock. It makes the body a target for engineered intervention, pointing toward the molecular machinery of Nanobots & Full-Dive VR (sv-kurzweil-nanobots). And it dissolves the old boundary between the born and the built: once life is editable information, the merger of biological and non-biological intelligence that defines The Singularity (sv-kurzweil-singularity) becomes a difference of degree, not of kind.
The deeper resonance is that this event reframes the entire arc behind it. The same mathematics of accelerating returns that compressed genome sequencing also drove the digital substrate — the World Wide Web (sv-www) as a planetary information layer, and the pattern-recognition leap of AlexNet & the Deep Learning Revolution (sv-alexnet-convnets) that gave machines the ability to read genomic and proteomic data at superhuman scale. In Kurzweil's grand narrative, the moment biology becomes information technology is the hinge where the story of carbon and the story of computation stop running on separate tracks. It is worth holding honestly as a documented prediction-framework rather than settled fact — but its central observation, that we can now read and edit the oldest code on Earth, is no longer speculative at all.
The reframing of biology as information technology matured at the millennium's turn. On 26 June 2000, Bill Clinton and Tony Blair jointly announced a working draft of the human genome, a public consortium (Francis Collins, NIH/DOE) racing Craig Venter's Celera; the near-complete sequence followed in April 2003, coinciding with DNA's fiftieth anniversary. This unfolded amid the dot-com bust (the NASDAQ peaked March 2000), the post-9/11 security turn, and the early Web 2.0 buildout. Moore's Law still governed expectations of exponential progress in silicon. In genomics, Sanger sequencing dominated until next-generation platforms (Solexa/Illumina, 454) arrived around 2005-2008, triggering the steep cost decline. Bioinformatics, GenBank, and the BLAST algorithm had already recast genes as searchable data. Kurzweil, fresh from The Age of Spiritual Machines (1999), published "The Law of Accelerating Returns" in 2001 and The Singularity Is Near in 2005, situating genomics within a sweeping claim that all information technologies—not just computing—advance exponentially in price-performance.
Kurzweil's contribution was conceptual rather than experimental: he generalized Moore's Law beyond semiconductors into a "Law of Accelerating Returns" and argued that once a domain becomes an information technology, it inherits exponential price-performance growth. Genomics was his proof case. Where the Human Genome Project had cost roughly $3 billion and a decade, sequencing cost-per-genome then fell from about $95 million in 2001 to under $1,000 by the late 2010s—outpacing Moore's Law itself, especially after next-generation sequencing arrived around 2008. This recast DNA not as chemistry but as readable, writable, hackable code: the "software of life." The framing licensed synthetic biology, personalized genomic medicine, and ultimately the logic that biology could be engineered and accelerated like software. It also became foundational to Singularitarian thought, supplying empirical-seeming curves to claims that biotechnology, nanotechnology, and AI would converge toward radical life extension and machine superintelligence—shifting these from science fiction into a forecast many technologists took as a planning horizon.
Had the genomic cost curve not collapsed so steeply—had sequencing tracked only Moore's Law rather than vastly outpacing it after 2008's next-generation platforms—Kurzweil's strongest empirical exhibit would have weakened, and "biology as information technology" might have remained metaphor rather than apparent law. The technological driver here was contingent: Solexa/Illumina's massively parallel reversible-terminator chemistry, not Kurzweil's prediction, produced the inflection. Absent it, costs could have plateaued, vindicating critics like Paul Allen who argued a "complexity brake" governs the life sciences. Yet the deeper reframing was overdetermined: Watson and Crick's 1953 "code," GenBank (1982), and bioinformatics had already digitized biology. So even without Kurzweil, biology-as-information would have advanced—but the specific Singularitarian narrative, with its confident exponential extrapolation toward radical life extension and AGI timelines, depended heavily on the dramatic genomics curve. Without that curve, transhumanist forecasting would likely have been more hedged, and the cultural authority Kurzweil acquired—later joining Google in 2012—correspondingly diminished.
The central dispute concerns whether the Law of Accelerating Returns is a genuine law or selective extrapolation. Paul Allen and Mark Greaves, in "The Singularity Isn't Near" (MIT Technology Review, 2011), argued that a "complexity brake" governs the life sciences and neuroscience: deeper biological understanding does not yield Moore's-Law-style acceleration, and Kurzweil cherry-picks favorable curves. Kurzweil's rebuttal ("Don't Underestimate the Singularity," 2011) cited genomics—recalling that critics declared the genome project failing at 1% completion, not grasping that exponential doubling meant it was nearly done. Skeptics including Theodore Modis and Steven Pinker question whether disparate technologies can be aggregated onto a single exponential, while economists (Robert Gordon) note stagnant aggregate productivity despite digital advance. Historians of science such as Hallam Stevens (Life Out of Sequence, 2013) and Evelyn Fox Keller interrogate the "genome as information/code" metaphor itself, arguing it obscures biological materiality and overstates programmability—a critique that cuts against the very premise that biology straightforwardly "becomes" information technology.
Myth: Kurzweil's claim that 'biology is now information technology' means we have understood the genome and can already reprogram our biology like software.
Reality: Sequencing a genome is not the same as understanding it. As critics including geneticist PZ Myers and biologists writing in Science have stressed, reading the DNA letters quickly does not grant an equally speedy understanding of what they mean. Traits are emergent properties of networks of interacting proteins, not one-to-one outputs of individual genes, and the 'missing heritability' problem (the gap between heritability estimated from pedigrees and that explained by discovered variants) has frustrated geneticists for over a decade. Kurzweil's framing of DNA as 'software' is a documented prediction/analogy about where biology is headed, not an accomplished fact.
Myth: The Human Genome Project (completed in 2003) sequenced the entire human genome.
Reality: The HGP and its follow-ups left roughly 8% of the genome unresolved, mostly in highly repetitive regions like centromeres and the short arms of acrocentric chromosomes. The first truly gapless, telomere-to-telomere human genome (T2T-CHM13) was not published until April 2022 by the T2T Consortium, adding nearly 200 million base pairs and ~2,000 gene predictions. Even that assembly initially lacked a complete Y chromosome. So 'the genome' people cite as finished in 2003 was a high-quality draft, not the complete sequence.
Myth: We achieved the '$1,000 genome,' so a fully sequenced and interpreted personal genome now costs about $1,000.
Reality: The widely cited $1,000 figure covers only sequencing consumables (reagents, chips) and excludes the sequencer, the technician, and—most importantly—analysis and clinical interpretation. Researchers dubbed this 'the $100,000 analysis' (Elaine Mardis) or even 'the $1-million interpretation' (Bruce Korf). Real-world clinical whole-genome sequencing, including counseling, bioinformatics, interpretation, and follow-up, has been estimated at roughly $3,000-$10,000 per patient. The headline number is, in the words of one critique, a 'bait and switch.'
Myth: DNA sequencing costs fell exponentially forever, vindicating Kurzweil's Law of Accelerating Returns at a steady pace.
Reality: Sequencing costs did fall dramatically—and from about 2008 to 2012 they dropped roughly 2,000 times faster than a Moore's Law baseline after next-generation sequencing arrived. But that super-exponential plunge was not permanent. NHGRI cost data show the curve flattened markedly after about 2015, with the rate of decline slowing well below its earlier pace. The trajectory is one of bursts driven by specific technology shifts, not a smooth perpetual exponential.
Myth: Kurzweil's track record proves these biotech predictions are about 86% accurate, so they should be treated as near-certainties.
Reality: The ~86% figure comes largely from Kurzweil grading his own predictions from 'The Age of Spiritual Machines,' a method critics consider too subjective because he selects which parts of vague statements to score as 'correct.' A LessWrong analysis ('good accuracy, poor self-calibration') found his self-ranking sits in roughly the 99th percentile of plausible scorings—i.e., generously self-favorable. His genome and biotechnology claims should be framed as documented projections from a self-described optimist, not as validated forecasts.
"When the genome project was first announced, skeptics said, 'You'll never get this done.' Halfway through the fifteen-year project, only 1 percent had been collected, so the skeptics were going strong. But the project had been doubling every year, and 1 percent is only seven doublings from 100 percent. It was indeed completed seven years later." — Ray Kurzweil, recounting the Human Genome Project as an illustration of the Law of Accelerating Returns (a point he makes in The Singularity Is Near, 2005, and repeats in interviews and his MIT Technology Review reply "Don't Underestimate the Singularity," 2011)