The apex metric
The point at which medicine adds more than a year to your remaining life expectancy for every year that passes, so it stops declining as you age. Coined by Aubrey de Grey in 2004. A finish line defined by a rate, which means it can be measured, and measured honestly.
Medicine is already adding years, fastest at old age. Below is how far each part of the science has come, what is underway, and the next proof it needs: a therapy shown to slow aging itself in people.
One level down
The index is derived, not asserted. Three inputs, each transparent and sourced, combined with fixed weights. Click any input to drill into its own breakdown.
de Grey's field roadmap: single-agent mouse lifespan achieved, Robust Mouse Rejuvenation partway, human rejuvenation not started.
Record life expectancy at birth rises 0.24 yr/yr (1840 to 2023, measured), against the 1.0 yr/yr escape velocity needs.
Escape velocity is the last of four milestones. The first is done; the second is underway in mice; the two human milestones have not started. This is what the top input to the index tracks.
One intervention reproducibly extends lifespan in mice under rigorous, multi-site conditions. Achieved: rapamycin and several others in the NIA Interventions Testing Program.
NIA Interventions Testing ProgramA COMBINATION of therapies roughly doubles the remaining life of an already middle-aged mouse. In progress: the LEV Foundation's RMR1 (1,000 mice, 4 interventions) was a qualified win, additive but insufficient; RMR2 (8 interventions) reports around 2028.
The prediction
Escape velocity is the year the rate of life-expectancy gain crosses one year per year. There is no honest single date, only scenarios, and every one of them waits on the same event: a therapy proven to slow aging in people.
Every mark assumes a therapy that slows human aging is proven first; today’s pace (0.24 of the 1.0 a year needed, at birth) does not get there by itself. The shaded window (2050–2100) is our model if that proof arrives and progress compounds. The median is the Metaculus community forecast; the earliest mark is Aubrey de Grey’s hedged 50% bet, an opinion, not our number.
Rate at birth today
Level 1 · the decomposition
Aging is not one process. Each of its 12 hallmarks is scored by its best-proven human-relevant lever on the 9-rung readiness ladder. The weakest is the rate-limiter: the bottleneck escape velocity waits on.
The ceiling
The readiness ladder has nine rungs. The 12 hallmarks are climbing the middle of it. The top two, the ones that actually define escape velocity, are empty.
7 of 12 hallmarks have a lever proven in a human outcome trial (rung 7). Those are real, valuable, and mostly proven for a specific age-related disease.
0 of 12 have a lever proven to slow aging itself in a human trial (rung 8). And none is proven to extend human lifespan (rung 9).
That is why the index cannot rise into the middle of the scale, no matter how many promising levers exist. It is gated on the one thing that defines the goal, and that thing has not happened yet.
What would move it: the first replicated human trial to slow a validated measure of aging, or Robust Mouse Rejuvenation succeeding in combination. What would move it down: a flagship human trial failing.
Explore all four levels
Open a hallmark to see the interventions behind its score, then an intervention for its evidence.
It is a transparent model, not a measurement. There is no peer-reviewed “proximity to escape velocity” metric in the literature; we built one so the question could be tracked honestly, and every input, weight, and source is on the page. Reasonable people would weight it differently, and the methodology states every choice.
It deliberately refuses to reward optimism. Expert timelines (including de Grey’s own hedged “50% within 12 to 15 years,” explicitly funding-contingent) never add points. Only realized, replicated results move a rung. Biologists who share that standard can put it to paid use: AI labs pay them by the hour to write biology problems and check what models answer, and the biologist roles on AITraining.jobs, a jobs board we also publish, list that work with its pay.
One known limitation we are transparent about: a hallmark’s readiness reflects the interventions that act on it, graded on their best human evidence. For several top levers that best evidence is a hard outcome in an age-related disease (a heart-failure or cardiovascular trial), not a trial of the aging mechanism itself. That is a second translation gap, disease-axis to hallmark-axis, sitting inside the first. We exclude diagnostics from levers and will move to per-mechanism grades to close it.
It is educational, not medical advice. Nothing here is a recommendation to take, stop, or change any drug, supplement, or protocol.
Share of hallmarks with a lever proven to slow aging in humans. The next step: the first one moves the index most.
The first controlled trial proves a rejuvenation therapy meaningfully reverses aging in humans. Not started: no intervention has been proven to extend human lifespan.
Aging-US, off-label geroprotector evidence review (2025)Remaining life expectancy rises by more than one year per calendar year, and keeps rising. Not reached: in the longest-lived country, life expectancy at birth rises about 0.24 years per year, and at 80 the rise has sped up about 13-fold since before 1950 but is still 0.09 a year.
HMD + UN WPP 2024Escape velocity is a rate: remaining life expectancy rising a year for every year that passes. This is the rate today, computed by us from the life tables of the longest-lived country each year, through 2023.
Years of life gained per calendar year
In the longest-lived country each year, at birth and for people who have already reached 65 or 80.
At birth the record has risen about 2.9 months a year since 1840 (0.24 yr/yr), a quarter of the 1.0 a year escape velocity needs, and a little slower since 2000. The gains have moved to old age, where aging acts: remaining life at 65 now rises 4× and at 80 13× faster than before 1950. Data through 2023. HMD + UN WPP 2024; at-birth line first drawn by Oeppen & Vaupel.
Where it is speeding up
faster gains in remaining life at 80 since 2000 than before 1950. At birth the pace eased to 0.18 yr/yr since 2000 (Olshansky 2024): the gains moved to the ages where aging acts.
What has to change
The curve
A therapy proven to slow human aging (rung 8-9, currently 0 of 12) is what lifts the rate toward one year per year. Every forecast below waits on that proof, which the mouse studies now underway lead to.
Today's pace alone
Not on its own
At birth the record rises ~0.24 years per year (~0.18 since 2000). At 80 the rise is ~13 times faster than before 1950, but still ~0.09. Extended in a straight line, today's pace does not reach 1.0: getting there takes a therapy that slows aging itself.
HMD + UN WPP 2024Science-accelerated
IF the first human rejuvenation is proven around 2042 (after Robust Mouse Rejuvenation, ~2028) AND the rate then compounds at 5% a year, it crosses 1.0 yr/yr around 2071. Plausible window 2050-2100.
LEV Foundation, RMR studyExpert opinion
Aubrey de Grey puts a hedged 50% chance within 12-15 years (about 2039), explicitly contingent on funding. This is an opinion, not our computation, and the model never lets a projected timeline add points to the index.
de Grey, 50% within 12-15 years (2025)Move the assumptions yourself
If the science accelerates, escape velocity by
about 45 years from now · 29 years after human rejuvenation is proven
Baseline: demographic trend
Not reached this century
(today's pace 0.24 at birth, target 1.0)
t* = onset + ln(1.0 / 0.24) / ln(1 + g) = 2042 + 29 = 2071
A model under deep uncertainty, not a measurement. Move the knobs: the year is a function of assumptions. The starting rate is measured; the onset and the compounding are what the science now underway will decide.
How much the multiplier moves the answer (onset held at 2042)
| Annual multiplier g | Years from onset | Escape velocity |
|---|---|---|
| 3% | 48 | 2090 |
| 5% | 29 | 2071 |
| 7% | 21 | 2063 |
| 10% | 15 | 2057 |
The multiplier is a knob, not a fact, which is the point: the honest answer is a wide window (roughly 2050-2100 in the accelerated case) with a hard floor of never on the trend. Community forecasts are aggregated on Metaculus.
Record in 2023: Hong Kong
The record is the highest period life expectancy among Human Mortality Database countries of at least a million people (shaded: the two eras compared above). At birth the line has been close to straight since 1840, which Vaupel and colleagues found continued to 2019, while Olshansky and colleagues argue gains at birth have slowed since 1990. Both readings agree on the shift this chart shows: the gains now come at old age.