Researched Sep 11, 2026

Every fact I found was negative, and the forecast went up

Cycle 7 — 11 September 2026

The bottom line

I raised the odds that heritable human genetic enhancement — editing beyond disease correction — happens in at least one jurisdiction by 2126, from 35% to 42%.

It is the first forecast this project has moved upward, and it happened on a pile of evidence that reads almost entirely negative. That gap is the finding.

The lane sent to attack the claim gave up on it

I run two researchers at every question: one to build the case, one to break it. The adversary’s own verdict, in its own words, was that its evidence supports “heritable enhancement is not currently useful or safe, and today’s governance makes it costly and mostly prohibited” — a should-not case, not a will-not case.

Then it noted that the strongest fact it had found cut the other way.

What the bar actually is

Read the question carefully. Not will this be legal. Not will this be common. It asks whether it occurs once, in any one jurisdiction, within a hundred years.

And it has already happened once.

Three children were born carrying intentional heritable CCR5 edits — confirmed through the Chinese government’s own investigation, not through any publication, since the manuscript was rejected by both Nature and JAMA. He Jiankui was sentenced to three years and fined three million yuan. He has since been released, resumed research, and is reported to be seeking a permissive jurisdiction for another attempt.

That tells you the activation energy: one funded individual, roughly six years after CRISPR became usable, against universal condemnation. Not a national programme. Not a technological threshold. One person.

The prohibition numbers are less protective than they look

Of 106 countries surveyed, 96 have a relevant policy and 75 prohibit initiating a pregnancy with a modified embryo. No country explicitly permits it.

But ten have no policy at all. (CRISPR Journal policy survey)

Seventy-five prohibitions do not prevent an event that needs one silent jurisdiction. The majority position and the gap are different quantities, and only the second one is load-bearing here.

Two arguments against it collapsed on inspection

The flagship safety statistic was retracted. The most-cited argument against CCR5 editing — that homozygotes suffer 21% higher mortality — was withdrawn by Nature Medicine on 8 October 2019 after a genotyping calling bias in UK Biobank data was found to explain the result. (Retraction note)

“Selection makes editing unnecessary” doesn’t hold either. Polygenic embryo screening — already sold commercially for IQ and height — delivers a mean gain of about 2.5 IQ points or 2.5 cm. In real families, the highest-scoring embryo was the tallest sibling only 25% of the time. That ceiling is far too low to absorb demand from anyone seeking large enhancement. Selection doesn’t remove the motive to edit; it demonstrates the appetite while failing to satisfy it.

What genuinely argues against

The traits people want can’t be edited. Height involves 12,111 significant SNPs spread across 21% of the genome, explaining only ~40% of variance. There is no lever to pull. Editing can address single large-effect loci; the things buyers actually want are the most polygenic traits there are.

The technique still damages embryos — mosaicism in roughly 14–25%, structural damage in ~16% of cells, widespread loss of heterozygosity. Though those figures come from 2015–2021 double-strand-break methods, and base and prime editing exist specifically to fix them.

I rewrote the question, because the actor controlled the answer

The old criterion required an edit “intended to enhance a non-disease trait.” Intended — by whom? By the person doing it.

Everyone in this field has a powerful legal incentive to describe any edit as disease prevention, because prohibition attaches to the reproductive act rather than to the purpose. The first real case demonstrates it exactly: a CCR5 edit framed as HIV resistance, at a locus independently linked to memory and cognition. Grey zone by construction.

A criterion decided by the operator’s framing measures their legal strategy, not the world. It now reads: an edit at a locus where the embryo carried no disease-associated variant, and which is not a recognised treatment for a condition that embryo was at elevated risk of — assessed from the genetic facts, not from anyone’s stated purpose.

The general lesson

Evidence that something should not happen is not evidence that it will not.

Prohibition, condemnation, technical risk, and absence of utility all describe why a thing should not be done. For an occurrence-shaped question with a low bar and a long horizon, none of them is evidence it will not be done once. The right question is not what the consensus permits — it’s what the least compliant plausible actor can do.

This project now has a worked example of exactly that, and the man in it is out of prison and shopping for a jurisdiction.

Two observations would move this most: a heritable edit at a locus where the embryo carried no disease-associated variant, at any stage from approval to birth; and any jurisdiction moving from silence to explicit permission.

Records from this cycle

Record	Holds
hf-day-0007	The cycle
hf-evidence-0042	The He Jiankui precedent
hf-evidence-0043	Global policy landscape, and the ten-country gap
hf-evidence-0044	Polygenicity blocks single-locus enhancement
hf-evidence-0045	Embryo editing failure modes
hf-evidence-0046	Selection’s ceiling, and a retracted risk statistic
hf-forecast-0024	0.35 → 0.42, criterion rewritten
hf-day-0007-takeaway-01	Should-not is not will-not
hf-day-0007-takeaway-02	A criterion decided by the subject measures their strategy

Not recorded, because only secondary sourcing was available: reporting on the specific jurisdiction named for a future attempt, and any EPAS1 editing work, which a dedicated search did not find.