Kurzweil Scorecard: The Freezer Held the Line. The Footnote Didn’t.

On a February morning in 2005, the Australian Museum quietly issued a statement: the thylacine cloning project — the one Mike Archer had championed since 1999, the one whose 2001 milestone Ray Kurzweil would cite as evidence that science was on the verge of reviving extinct species — was over. The DNA, the museum said, was “far too degraded to even construct a DNA library.” That announcement landed the same year The Singularity Is Near hit shelves, with a footnote treating the same project as a forward marker.

A short section of the book argued that, between cryopreservation of living cells and the synthesis of patchwork DNA from inactive fragments, “those species will no longer need to become permanently extinct” (ch. “Preserving Endangered Species and Restoring Extinct Ones”). Twenty years on, that paragraph reads like two predictions stapled together. One aged well. The other had already collapsed by the time the ink dried.

What Kurzweil actually predicted

Four claims sit inside this batch, all from the same short section of the book:

1. The cryopreservation claim. Cells from endangered species, banked while the animal is alive, can later be used to re-create the species after extinction.
2. The recent-extinction claim. It will “eventually be possible to re-create animals from recently extinct species.”
3. The patchwork-DNA claim. For long-extinct species, scientists will “synthesize the necessary DNA by patching together information from multiple inactive fragments rather than finding a fully intact preserved cell” — the Jurassic Park premise, inverted.
4. The Tasmanian tiger evidentiary footnote. Kurzweil reported that in 2001, scientists had synthesized DNA for the thylacine, sixty-five years after extinction, “with the goal of bringing the species back to life.”

The interesting thing is that all four are now testable in 2026 in a way they were not testable in 2005. The freezers have been running long enough that the cells in them are older than the predictions about them.

The freezer worked

The most quietly remarkable result of the past five years is that a population biologist’s pre-internet daydream — bank cells now, revive the species later — has actually produced living mammals.

In 1980, a Przewalski’s horse stallion known as SB615 had a skin biopsy taken and frozen in liquid nitrogen at what would become the Frozen Zoo at the San Diego Zoo Wildlife Alliance. The stallion died in 1998. In August 2020, a foal named Kurt was born to a domestic surrogate mare, cloned by somatic cell nuclear transfer from those 1980 cells. In February 2023, a second clone, Ollie, followed. The 2025 paper documenting the work in Animals notes that nearly all surviving Przewalski’s horses descend from twelve founders; Kurt and Ollie carry genetic variation no living individual still possesses.

The black-footed ferret tells the same story on a tighter timeline. A wild-caught female named Willa died in the late 1980s; her cells were banked in 1988. The U.S. captive breeding program had been rebuilt from just seven founders. In December 2020, a clone of Willa — Elizabeth Ann — was born. In April 2024 the U.S. Fish and Wildlife Service announced two more, Noreen and Antonia, from the same cell line.

The Frozen Zoo passed eleven thousand cell lines in 2024 and now holds more than 11,500, representing 1,337 species. Papers on de-extinction crossed eighty published works in 2025 alone — five times the rate of a decade earlier. The ones doing actual lab work tend to converge on the same workflow: frozen fibroblasts, interspecies SCNT, surrogate dam from a close relative. The freezer is doing what Kurzweil said it would do. Verdict: ahead of schedule on the cryopreservation claim — the proof exists, with names and birth dates.

The footnote that didn’t survive its own footnote year

The Tasmanian tiger claim has aged worse than almost anything else in this scorecard series.

Mike Archer announced the Australian Museum’s thylacine cloning project in May 1999, using a pup that had been preserved in alcohol since 1866. By 2002–2003, the project had managed PCR amplification of small thylacine gene fragments — the basis for the “synthesized DNA” phrase that traveled through the press and ended up in Kurzweil’s footnote. In February 2005, the same year The Singularity Is Near published, the museum formally ended the project, stating that the recoverable DNA was too degraded for even a library, let alone an organism.

The 2009 Genome Research paper on Thylacinus cynocephalus — the first complete thylacine mitochondrial sequence — is when the field actually had its first defensible genome-scale dataset on the species. That came four years after the cloning project’s collapse. Everything before it, including the 2001 work Kurzweil cited, was fragmentary at best.

The Tasmanian tiger claim as Kurzweil reported it doesn’t score “wrong.” It scores overstated at the time he wrote it. Verdict: behind schedule as cited; the actual project collapsed the same year the prediction was published.

Restored, or merely re-skinned?

The recent-extinction claim is where the story gets uncomfortable.

In October 2024 and January 2025, Colossal Biosciences announced the birth of three canids — Romulus, Remus, and Khaleesi — that the company calls dire wolves. The work, led by Beth Shapiro, took a gray wolf cell line and made twenty edits across fourteen genes, targeting traits like coat color, texture, and skull morphology inferred from a 13,000-year-old tooth and a 72,000-year-old skull. No dire wolf DNA was inserted; the edits were chosen to recapitulate visible traits associated with the extinct lineage.

The scientific reception was sharp. Science ran a piece titled “Is the dire wolf back from the dead? Not exactly.” The Bulletin of the Atomic Scientists called it a “colossal misrepresentation.” Shapiro herself, afterward, conceded the framing: “I don’t care if you call it a dire wolf. De-extinction is never going to be about creating perfect copies of something.”

Read literally, twenty gene edits in a gray wolf satisfies Kurzweil’s claim about re-creating animals from recently extinct species. Read in the spirit of “the Tasmanian tiger walks again,” it does not. The same applies to Colossal’s woolly mouse from March 2025: seven simultaneous gene edits in Mus musculus gave mice a hair phenotype reminiscent of woolly mammoths. That is a multiplexed gene-editing achievement. It is not an extinct mammal. Verdict: wrong mechanism. The destination — functional ecological proxies — is being reached, but through engineered hybrids rather than reconstructed lineages.

Patchwork DNA: the part Kurzweil got most right, eventually

The patchwork-DNA claim is the strangest to score, because the technique Kurzweil described — assembling a usable genome from many inactive fragments rather than waiting for one intact cell — is now standard practice. The University of Melbourne’s TIGRR lab under Andrew Pask, working with Colossal since 2022, announced in October 2024 that long RNA molecules from a 110-year-old pickled thylacine head, combined with short DNA reads from other specimens, had yielded a reconstructed thylacine genome the team claims is more than 99.9% complete and contiguous. The fat-tailed dunnart cell line they edit as a downstream surrogate now carries more than 300 unique genetic changes. The 2025 Proceedings of the Royal Society B paper “Illuminating the mystery of thylacine extinction” is built directly on that assembly.

The patent landscape reflects the bet. Colossal’s WO2024211655 application — “Woolly mammoth specific gene variants and compositions comprising same” — claims forty-two mammoth gene variants and the recombinant elephant cells and transgenic animals that carry them, with sequence-identity thresholds reaching down to 80%. Specific genes named include FGF5 (hair length) alongside elephant homologs being edited for cold tolerance, lipid metabolism, and hemoglobin oxygen affinity. Patent experts quoted in MIT Technology Review in April 2025 expect examiners to narrow the claims sharply, but the filing is a marker: the patchwork-genome era is being staked out as intellectual property, not just published as science. Verdict: ahead of schedule on the technique, too early to call on the outcome.

The scorecard

What Kurzweil missed (and what he nailed)

Two patterns emerge.

First, the prediction Kurzweil grounded in infrastructure aged well; the one he grounded in current events aged poorly. The Frozen Zoo was, in 2005, already a quarter-century-old institution with thousands of cell lines; it kept doing what it does, and twenty years later it has produced living animals. The Tasmanian tiger announcement was, in 2005, a press release on its way to a quiet retraction. The book treated them as parallel evidence. They weren’t.

Second, the field converged on something Kurzweil didn’t quite anticipate: the most viable path isn’t literal resurrection but engineered functional proxies. Colossal’s official line is that a cold-tolerant Asian elephant filling the woolly mammoth’s ecological role counts as success, even if it carries 0% of the mammoth genome by base count. The Pask lab’s thylacine work is closer to literal reconstruction, but even there the surrogate is a dunnart and the output will be a thylacine-like marsupial. Whether that counts as de-extinction is now a definitional fight rather than a technical one. Kurzweil’s framing assumed the technical question was the hard part. In 2026 the technical question is mostly answered, and the definitional question is where the controversy lives.

A final detail: The Singularity Is Nearer, Kurzweil’s 2024 update, doesn’t return to this topic. Extinction appears only in the context of asteroid impacts, nuclear war, and existential risk to humans. The de-extinction paragraph from 2005 was not extended, not retracted, not defended. Sometimes silence is its own verdict.

Method note

Verdicts were assembled by reading the actual claims and figures from the cloning papers (the 2025 Animals paper on Przewalski’s; the 2024 black-footed ferret preprint; the 2025 Proceedings B relaxed-selection paper), the published patent claims of Colossal’s mammoth filing (WO2024211655), and the press and scientific responses to the dire wolf announcement. Trend counts come from full-text searches across roughly 9.3 million U.S. patent documents and roughly 357 million scientific works. The Singularity Is Nearer was searched for any restatement of the original predictions; none was found.