Immortality as a Physical Problem: The Strange, Unwanted Law That Says Aging Might Be Disposable

New name B I U S link H1 H2 H3 list 1. x² x₂ page image code pre HTML center hr table note notes desc ▲ ▼	Show page Syntax
{html} <article> <header> <p class="dek">In 2004, the physicist Mark Ya. Azbel’ published a short, dense paper with an impossible-sounding thesis: mortality in protected populations — humans included — is governed by an exact, universal law that holds from yeast to people, and that law implies aging is a <em>reversible, disposable evolutionary byproduct</em>. Twenty years later, the demographic data he relied on has only grown. The biology has not caught up.</p> <p class="byline">A deep read of <a href="https://arxiv.org/pdf/q-bio/0403008v2" rel="noopener">Azbel’, M. Ya. (2004), “Immortality as a physical problem,” arXiv:q-bio/0403008v2</a>.</p> </header> <section> <h2>1. An evolutionary accident we never asked for</h2> <p>Azbel’ opens with a point that is easy to miss but is the load-bearing wall of the entire paper. “Well protected human and laboratory animal populations with abundant resources,” he writes, “are evolutionary unprecedented.” For the entire history of life on Earth, animals evolved under conditions of scarce resources, fierce competition, and high extrinsic mortality. Natural selection optimized for one thing: survive long enough to reproduce. There was never any selective pressure to build a body that <em>keeps working</em> far past reproductive age, because in the wild almost nothing did.</p> <p>Then, in a geological blink — roughly the last 150 years for humans, and a few decades for laboratory animals — we built a world where predation, starvation, and infectious disease were largely removed. The thing evolution never designed for is now the default condition for billions of people and every lab mouse on Earth. Azbel’s argument is that this accidental experiment has exposed something hidden during the previous four billion years: the mortality we see in protected populations is not a fundamental wear-out process. It is a leftover. An evolutionary byproduct that, once you can see it clearly, turns out to follow a precise physical law — and laws, unlike vague biological intuitions, can be broken.</p> <p>His evidence is not a new experiment. It is a re-reading of data that demographers had been collecting since Edmond Halley (of the comet) built the first modern life table in 1693, and Leonhard Euler formalized the mathematics in 1760. By 2004, the Human Mortality Database held millions of accurately registered births and deaths from developed countries going back over a century, and biodemographers had extended the same kind of record to animals from yeast to mammals. Azbel’s contribution was to treat that pile of numbers the way a physicist treats data — not to fit a curve, but to ask what <em>exact</em> relation survives when you strip away the noise.</p> </section> <section> <h2>2. The 200-year-old equation that everyone uses, and where it fails</h2> <p>To appreciate what Azbel’ found, you have to know what the field thought it already knew. In 1825, the actuary Benjamin Gompertz read a paper to the Royal Society showing that adult human mortality rises roughly exponentially with age. The Gompertz law (later extended by Makeham) is usually written as a hazard function:</p> <p class="formula">μ(x) = α·e<sup>βx</sup> + λ</p> <p>where μ(x) is the instantaneous death rate at age <em>x</em>, α sets the baseline, β is the “actuarial ageing rate,” and λ is an age-independent background risk. Log-mortality grows in a straight line — a remarkably good approximation across adult human life and across many species. For two centuries it has been the backbone of insurance pricing and the default model in biodemography.</p> <p>But Gompertz is an <em>empirical</em> fit, not a law of nature, and by the late 20th century it was visibly leaking at the seams. James Curtsinger’s lab, studying tens of thousands of <em>Drosophila melanogaster</em> in carefully controlled cohorts (Curtsinger et al., <em>Science</em> 258:461, 1992), showed that at very advanced ages the mortality rate stops rising exponentially and <em>plateaus</em>. The same late-life deceleration shows up in human data and in medflies. Gompertz cannot produce a plateau; it just keeps climbing. Something else is going on.</p> <p>Azbel’s claim is stronger than “Gompertz is incomplete.” He argues the entire framing — mortality as a smoothly increasing function of age — is wrong for protected populations, because it hides the actual structure. The real relation is not exponential in age. It is <em>piecewise linear in infant mortality</em>, and that change of variables turns out to expose a universal law that Gompertz never could.</p> </section> <section> <h2>3. The exact law: echelons, phases, and a strange analogy with thermodynamics</h2> <p>Here is the core of the paper, and it is worth slowing down for. Demographic life tables give you, among other things, two quantities at each age <em>x</em>: <em>q(x)</em>, the probability that someone who survived to <em>x</em> dies before <em>x</em>+1; and <em>d(x)</em>, the probability that a live newborn dies between ages <em>x</em> and <em>x</em>+1. The two are related by <em>q(x) = d(x)/l(x)</em>, where <em>l(x)</em> is survivability to age <em>x</em>. Infant mortality is <em>q(0) = d(0)</em>.</p> <p>Azbel’ makes a conjecture: in a protected population, a dominant fraction of mortality depends on a small, specific number of parameters, and the relation between mortality at any age and infant mortality is the <em>same</em> in every population subgroup. Mathematically, if <em>d(x) = f<sub>x</sub>[q(0)]</em>, then within any subgroup <em>G</em>, <em>d<sub>G</sub>(x) = f<sub>x</sub>[q<sub>G</sub>(0)]</em>. Because <em>d(x)</em> is an additive (averaging) variable — the population total is the sum over subgroups — we have &lang;f<sub>x</sub>[q<sub>G</sub>(0)]&rang; = f<sub>x</sub>(&lang;q<sub>G</sub>(0)&rang;). A standard result from probability says: if the average of an analytic function equals the function of the average, the function must be linear. If you allow singularities, it must be <em>piecewise linear</em>.</p> <p>So within each “echelon” (Azbel’s term for a linear piece), the law is:</p> <p class="formula">d(x) = d<sub>j</sub>(x) = a<sub>j</sub>(x)·q(0) + b<sub>j</sub>(x),   for q<sub>j</sub> < q<sub>G</sub>(0) < q<sub>j+1</sub>   (Eq. 1)</p> <p>For an arbitrary heterogeneous population spread across echelons with fractions <em>c<sub>j</sub></em> and <em>f<sub>j</sub></em>, this generalizes to:</p> <p class="formula">d(x) = a·q(0) + b,   where a = Σf<sub>j</sub>a<sub>j</sub>,  b = Σc<sub>j</sub>b<sub>j</sub>   (Eq. 2)</p> <p>What that says out loud: <strong>mortality at every age reduces to a linear function of infant mortality</strong>, with coefficients that are universal once you scale them with the species’ own boundary values <em>q<sub>j</sub></em>. The same relation holds for Japanese women, Swedish men, 19th-century Englishmen, and — with appropriate rescaling — laboratory flies.</p> <p>Azbel’ then notices something more. Equation 1 can be rewritten as a convex combination:</p> <p class="formula">d(x) = C·d<sup>(j)</sup>(x) + (1−C)·d<sup>(j+1)</sup>(x),   where d<sup>(j)</sup>(x) = a<sub>j</sub>(x)·q<sup>(j)</sup> + b<sub>j</sub>(x)   (Eq. 3–4)</p> <p>This is formally identical to the equation of a phase mixture in thermodynamics. Each echelon behaves like a “phase” with its own equation of state <em>d<sup>(j)</sup>(x)</em>, and <em>C</em> is the “concentration” of that phase, set by infant mortality <em>q(0) = d(0)</em>. Crossing an echelon boundary is like crossing a phase boundary: the population “homogenizes” into the next phase. That is not a metaphor Azbel’ uses lightly — he is a condensed-matter physicist, and he is suggesting that protected-population mortality is literally a phase-equilibrium phenomenon, with the “rungs” of the ladder being the phase transitions.</p> </section> <section> <h2>4. The four predictions that should not be true</h2> <p>The law, taken seriously, makes four predictions that contradict every standard evolutionary theory of aging (mutation accumulation, antagonistic pleiotropy, disposable soma) and every existing demographic model. They are:</p> <ol> <li><strong>Zero universal mortality up to a cutoff age.</strong> The empirical extrapolation of the lowest echelon to <em>q(0) = 0</em> yields <em>d(60) = 0</em> and <em>d(80) = 0</em>. In other words, the law says the universal component of human mortality is zero up to roughly age 80. Whatever kills protected people before 80 is not “aging” in any fundamental sense — it is non-universal, removable noise. (If <em>b<sub>1</sub>(x) = 0</em> held at <em>all</em> ages, the implication would be ultimate immortality.)</li> <li><strong>Mortality is reversible at any age.</strong> Because the law ties all-cause mortality at age <em>x</em> directly to infant mortality <em>q(0)</em>, anything that lowers <em>q(0)</em> also lowers mortality at every other age. An 80-year-old’s death rate can, in principle, be reset to the value it had decades earlier.</li> <li><strong>Mortality has short memory.</strong> The relaxation time is on the order of one year for humans, one day for flies — a few percent of the lifespan. Your mortality at 70 is dictated mostly by conditions <em>now</em>, not by the conditions you lived under at 30.</li> <li><strong>Adaptation is stepwise, not gradual.</strong> Populations do not slide smoothly toward lower mortality. They hop between echelons — phase transitions — when infant mortality crosses a boundary. The singularities of the law are the rungs of a ladder.</li> </ol> <p>None of these predictions is friendly to the orthodoxy. Standard theories say aging is the cumulative buildup of damage, that mortality should be history-dependent, and that lifespan has a hard ceiling set by the slowest repairable failure mode. Azbel’ says the data do not support any of that — and then spends the rest of the paper showing the data agree with him.</p> </section> <section> <h2>5. The human evidence: Switzerland, Japan, Norway, and one natural experiment called Germany</h2> <p>Start with the most counterintuitive prediction: zero mortality at young ages. In 2001 Switzerland, in age cohorts of roughly 60,000 girls, exactly <strong>1 girl</strong> died at each of ages 5, 9, and 10; 5 died across the 4–7 and 9–13 ranges; 10 or fewer died at any single age from 2 to 17; no more than 16 died at any age from 2 to 26. The same pattern held across all Western developed countries from 1999 to 2002. When a stochastic death count over tens of thousands of person-years is in the single digits, the honest inference is that the underlying rate is statistically indistinguishable from zero — at least in the lowest-mortality subgroups.</p> <p>At advanced ages, the gains are no less dramatic. In 1999 Japan, female survivability was 97% to age 51, 95% to 58, 85% to 73, and 73% to 80. In 1950, only 6% of Japanese women survived to 80. That is not a doubling; it is a twelve-fold increase in the probability of reaching your ninth decade, accomplished in a single generation by improvements in nutrition, sanitation, and medicine — none of which targeted “aging” as a mechanism.</p> <p>The reversibility prediction gets its cleanest human test from Norway. Azbel’ computes cohort mortality for Norwegian women born in 1900 — neutral Norway, no major war disruption. Beyond age 17, their mortality fell monotonically until age 40, at which point it had reversed to the value it had at age <strong>12</strong>. It then barely changed until age 50. Only at age 59 did it climb back to the value it had at age 17 — meaning a 59-year-old, by this measure, had the mortality of a 17-year-old, a 42-year reset. This is the law’s signature: mortality at one age can drop to the value it had at a much younger age, and stay there for decades.</p> <p>The short-memory prediction gets its cleanest test from German reunification. After 1990, mortality in the East — especially among the elderly, who had lived 45 years under very different conditions — fell rapidly toward Western levels. (Vaupel, Carey, and Christiansen reported this in <em>Science</em> 301:1679, 2003; subsequent work by Vogt and Missov in <em>Population Health Metrics</em> 2017 quantified it in detail.) If mortality were the cumulative record of a lifetime’s insults, you would not expect 70-year-old East Germans to converge on West German death rates within a few years of reunification. They did. The history-dependence that standard aging models take for granted is simply not there in the data once you look at how fast old-age mortality responds to changed conditions.</p> <p>Azbel’ verifies the universality of Eqs. (1) and (2) quantitatively on data from 18 developed countries — Austria (1948–1999), Canada (1921–1996), Denmark (1921–2000), England (1841–1898), Finland (1941–2000), France (1899–1897), West Germany (1956–1999), Italy (1906–1998), Japan (1950–1999), Netherlands (1950–1999), Norway (1896–2000), Sweden (1861–2000), Switzerland (1876–2001), and others — producing over 3,000 male and female <em>d(x)</em>-vs-<em>q(0)</em> curves. Piecewise-linear fits converge to <strong>five universal echelons</strong>, the same five for every country. Extrapolating the lowest echelon to <em>q(0) = 0</em> gives <em>d(60) = 0</em> and <em>d(80) = 0</em> — the zero-mortality prediction, recovered independently from Swiss, Japanese, Norwegian, and Swedish data.</p> </section> <section> <h2>6. The animal evidence: yeast, flies, and a worm that lived six times too long</h2> <p>If the law is universal across species, the animal data should show the same signatures — and they do, more starkly than the human data.</p> <p><strong>Yeast.</strong> S. Michael Jazwinski’s group (reported in <em>Experimental Gerontology</em> 33:571, 1998) found yeast mortality was zero for the first half of its mean lifespan. Half a lifetime with no deaths is not slow aging; it is no aging, in a single-celled eukaryote separated from humans by roughly a billion years of evolution.</p> <p><strong>Flies.</strong> Curtsinger’s large <em>Drosophila</em> cohorts (the same Science 258:461, 1992 data that exposed the late-life mortality plateau) included a striking datum: of 7,500 dietary-restricted flies, only 2 died at 8 days. At the population level that is, again, indistinguishable from zero over a meaningful window of life.</p> <p><strong>Flies on demand.</strong> The short-memory and reversibility predictions get perhaps their most controlled test from Linda Partridge’s lab. Mair, Goymer, Pletcher, and Partridge (<em>Science</em> 301:1731, 2003) showed that switching dietary restriction on at day 14 of a fly’s life restored its mortality, within 3 days, to the value it had at day 7 — a 2.5× age reversal, accomplished in three days. Crucially, the same longevity benefit appeared whenever the restriction was switched on, independent of the fly’s prior dietary history. The mortality was listening to <em>current</em> conditions, not the past.</p> <p><strong>The worm that breaks the ceiling.</strong> The single most striking number in the paper comes from <em>C. elegans</em>. Arantes-Oliveira, Berman, and Kenyon (<em>Science</em> 302:611, 2003) — Cynthia Kenyon’s lab, building on her 1993 discovery that <em>daf-2</em> insulin/IGF-1 receptor mutations double worm lifespan — perturbed a small number of genes and tissues and produced animals with a mean lifespan of <strong>124 days</strong>, roughly <strong>six times</strong> the wild-type value, with, in Azbel’s words, “no apparent loss in health and vitality.” Of 1,368 such animals, none died before day 25 (about 90 in human-equivalent years); only 5% died by day 40; only 15% died in the first three months (over 200 human-equivalent years). Scaled to humans, that is a 430-year lifespan.</p> <p>This is the experiment Azbel’ built the paper around, and it is worth pausing on why it matters. A six-fold lifespan extension is not a refinement of Gompertz. It is a demolition. Standard theories predict that lifespan is bounded by the slowest-accumulating damage — DNA mutations, protein aggregates, mitochondrial decline. None of those mechanisms can be cleared six-fold by perturbing “a small number of genes and tissues.” Either those mechanisms are not what is actually setting lifespan, or they are being actively and reversibly regulated. Azbel’ argues the latter: the universality of the law means the same regulation exists in yeast, flies, worms, and humans, and it is disposable.</p> <p><strong>Where memory does persist — and why it matters.</strong> The paper is honest about a wrinkle. When dietary restriction in flies was switched <em>back</em> to full feeding, longevity stayed higher than in the always-full-fed control. When fly temperature was shifted from 27°C to 18°C (or vice versa), the mortality change driven by the previous temperature persisted. Azbel’s interpretation is not that the law is wrong but that the relaxation time matters: temperature and diet were changed abruptly, and the population may not have had time to re-equilibrate. This is a testable prediction, not a fudge — the law says adaptation is fast (<em>~1 day for flies</em>), and if you change conditions faster than that, you will see transient history-dependence that fades as the new equilibrium is reached.</p> </section> <section> <h2>7. What “disposable evolutionary byproduct” actually means</h2> <p>Azbel’s central conceptual claim is that universal mortality — the part of death risk that follows the exact law — is not a designed-in death program and not the inevitable accumulation of damage. It is a side effect of how protected populations expose a phase-equilibrium mechanism that evolved for something else entirely (likely stress response, reproduction timing, or metabolic regulation). Side effects, unlike designed-in programs, can be removed without redesigning the organism.</p> <p>This reframes the entire aging-research program. For decades the field has hunted for the <em>causes</em> of aging: telomere shortening, oxidative damage, senescent-cell accumulation, mitochondrial dysfunction, genomic instability. Azbel’s law says those mechanisms are real but they are <em>not what sets the curve</em> — the curve is set by a small number of phase parameters that translate environmental inputs into mortality. If you want to bend the curve, you do not necessarily need to fix every form of damage. You need to identify the phase parameters and shift them. The Kenyon lab’s six-fold worm lifespan is the existence proof: a small genetic perturbation moved a phase boundary, and the worm’s mortality followed.</p> <p>The thermodynamic analogy is doing real work here. In a phase equilibrium, macroscopic properties (density, magnetization) are set by a few state variables (pressure, temperature, field), and the microscopic details wash out. Azbel’ is suggesting that mortality in protected populations has the same structure: the bewildering heterogeneity of individual causes of death — cancer, heart disease, infection, accident — reduces, in aggregate, to a small number of scaling parameters tied to cellular processes. That is why the same five-echelon law fits 18 countries over 150 years. The mechanism is common to yeast and humans because it sits at the cellular level, not the organ or organism level.</p> </section> <section> <h2>8. What the law does <em>not</em> say, and the friction in translating it</h2> <p>It is worth being precise about the limits, because a paper titled “Immortality as a physical problem” invites overreach.</p> <p><strong>The law covers protected populations only.</strong> The derivation assumes “regular” conditions — infant mortality changing relatively little year-to-year — and “well-protected” populations, where demographic fluctuations are within generic statistical accuracy. Azbel’ explicitly excludes years during and immediately after major wars, epidemics, and food or water contamination. The law is not a theory of mortality in famine, war, or the wild. Extrapolating it to evolutionary time scales where extrinsic mortality dominates is not warranted.</p> <p><strong>“Zero mortality” is a limit, not an observation.</strong> The lowest echelon extrapolates to <em>d(60) = d(80) = 0</em> at <em>q(0) = 0</em>, but no real human population has reached <em>q(0) = 0</em>. What we see are very small numbers — single-digit deaths over tens of thousands — that are statistically consistent with zero but not proof of it. Azbel’ himself notes that “higher accuracy may change these ‘empirical’ zeroes into small universal values.” The strong claim (“ultimate immortality” if <em>b<sub>1</sub>(x) = 0</em> at all ages) is a mathematical possibility, not a demonstrated fact.</p> <p><strong>Human maximal lifespan has not moved.</strong> Azbel’ notes, accurately, that human maximal lifespan has stayed around 120 years from ancient Rome (where tombstones recorded birth and death) to the present. The law explains why <em>mean</em> lifespan has doubled — protected populations have moved through echelons toward the lowest one — but it does not, by itself, predict that the maximal lifespan will keep rising. The six-fold worm extension shows the ceiling is movable in <em>some</em> species; whether it is movable in humans, and how, is exactly the open question.</p> <p><strong>The 6-fold worm result has caveats.</strong> A later study (Bansal et al., 2015; revisited by Podshivalova, Kerr, and Kenyon, <em>Cell Reports</em> 2017) found that <em>daf-2</em> mutants can spend a disproportionately long final period in a decrepit state — though that turns out to be largely an artifact of bacterial colonization of the gut, and when colonization is prevented, the mutants are healthier, proportionally, than wild type. The point is that “healthspan” and “lifespan” do not always scale together, and any human application has to track both.</p> <p><strong>The mechanism is unspecified.</strong> The law tells you mortality behaves like a phase equilibrium. It does not tell you what the phases <em>are</em> biologically — which genes, which cell processes, which molecular switches set the parameters <em>q<sub>j</sub></em>, <em>a<sub>j</sub>(x)</em>, and <em>b<sub>j</sub>(x)</em>. Azbel’ is explicit: identifying those is “a challenge and calls for extensive physical and biological study,” and he suggests yeast as the simplest system in which to do it. Twenty years on, that program is still incomplete.</p> </section> <section> <h2>9. The two-decade verdict: the law has aged better than its critics</h2> <p>It is now over twenty years since Azbel’ posted this paper (and since its peer-reviewed publication as “Universal mortality law and immortality” in <em>Physica A</em>, 2004). The intervening period has been generous to its empirical core and unkind to its dismissive critics.</p> <p>The mortality plateaus Curtsinger saw in flies have been confirmed and re-confirmed across species. The dietary-restriction effect has been extensively characterized in rodents: meta-analyses (Swindell, <em>Ageing Research Reviews</em> 2011) find median lifespan increases of 14–45% in rats and 4–27% in mice, with the caveat that effects are genotype-dependent and absent in some wild-derived strains — exactly the kind of echelon-boundary heterogeneity Azbel’s framework predicts. The insulin/IGF-1 pathway that Kenyon’s lab identified in worms has been shown to regulate lifespan in flies, mice, and (through human genetic association studies) people, with low IGF-1 levels and IGF-1-pathway polymorphisms over-represented in human centenarians. The German reunification convergence has been re-analyzed in finer detail and holds up.</p> <p>What has <em>not</em> happened is the molecular identification of the phase parameters. The geroscience field has produced an enormous catalog of aging-related pathways — mTOR, AMPK, sirtuins, NAD+, senescence clearance, partial reprogramming — and many of them demonstrably shift mortality curves in model organisms. But no one has, to my knowledge, mapped them cleanly onto Azbel’s <em>q<sub>j</sub></em> and <em>b<sub>j</sub>(x)</em>. That is the unfinished work. It is also, ironically, the work that would let the law become predictive rather than descriptive: tell me which cellular process sets the phase boundary, and I can tell you which intervention will move it.</p> <p>The strongest reason to take Azbel’ seriously is not any single data point. It is the universality. A relation that holds from <em>S. cerevisiae</em> to <em>H. sapiens</em> — from a single cell to a trillion-cell organism, from a 2-day lifespan to a 100-year one — is not a coincidence of biochemistry. It is a conservation law. Conservation laws in physics are the deepest things we know; they survive revolutions in the underlying theory because they reflect symmetry, not mechanism. Azbel’ is arguing that mortality has one. If he is right, aging is not a problem to be engineered around. It is a parameter to be set.</p> </section> <section class="bottom-line"> <h2>Bottom line: if you wanted to act on this today, what can we do</h2> <p>The full paper, with the derivations, the demographic figures, and the country-by-country fits, is here: <a href="https://arxiv.org/pdf/q-bio/0403008v2" rel="noopener"><strong>Azbel’, M. Ya. (2004), “Immortality as a physical problem,” arXiv:q-bio/0403008v2</strong></a>. Read it before acting on anything below; the recommendations are mine, distilled from the law’s empirically supported predictions plus two decades of follow-on biology, and they are probabilistic, not guaranteed.</p> <ol> <li><strong>Treat your current conditions as the dominant input, not your history.</strong> The law’s short-memory prediction is the single most actionable thing in the paper. Your mortality at 60 is set mostly by your conditions <em>now</em>, not by what you did at 30. This is not a license to be reckless in your youth, but it is a strong argument against fatalism in middle age. Quitting smoking, fixing sleep, improving diet, and starting exercise in your 50s and 60s move the current-conditions dials that the law says matter most — the Norway cohort and the dietary-restriction-in-flies data both show mortality reversing to younger-age values within months, not decades.</li> <li><strong>Lower your “infant-mortality-equivalent” inputs.</strong> In the law, all-cause mortality at every age scales with <em>q(0)</em>, the infant-mortality parameter. In an adult, the closest proxies are the things that set baseline inflammation, immune competence, and metabolic stress — because those are the systems that, when dysregulated, push you up the echelon ladder. Concretely: chronic infections (treat them), dental disease (the oral-systemic link is real), uncontrolled blood glucose, chronic sleep debt, and sustained psychological stress all act to push <em>q(0)</em>-equivalent upward. Fixing them pushes it down, and the law says every age benefits.</li> <li><strong>Don’t chase single-mechanism longevity drugs yet — but watch the ones that move the curve.</strong> The law predicts that interventions which genuinely shift a phase parameter will produce stepwise, not incremental, mortality reductions. Rapamycin, the most robust pharmacological lifespan extender in mice across multiple labs, looks more like a phase shifter than a damage-fixer. Metformin data is murkier. Partial reprogramming (Yamanaka-factor cyclic expression) is too early to call in humans. The actionable stance: stay current with the geroscience literature, but be skeptical of any “anti-aging” intervention that does not produce a clear, multi-species mortality-curve shift, because the law says the real ones will.</li> <li><strong>Use the law as a filter on health information.</strong> Anything that promises lifespan extension without a plausible mechanism for shifting cellular state parameters — a supplement, a diet fad, a peptide — is unlikely to move a phase boundary and therefore unlikely to do what the law would recognize. Conversely, interventions that operate on conserved cellular pathways (mTOR, AMPK, insulin/IGF-1, autophagy) are the natural candidates, because the universality of the law says the relevant machinery sits at the cellular level and is shared across species.</li> <li><strong>For researchers and funders: the unfinished job is molecular identification of the phase parameters.</strong> Azbel’s law is descriptive until someone maps <em>q<sub>j</sub></em>, <em>a<sub>j</sub>(x)</em>, and <em>b<sub>j</sub>(x)</em> onto specific cellular processes — ideally in yeast, as he suggested, where the genetics is tractable and the lifespan is short. That mapping would convert the law from a retrospective fit into a prospective tool: given a candidate intervention, predict which echelon boundary it moves and by how much. This is, in my view, the single highest-leverage unfunded question in biogerontology.</li> <li><strong>For policy: protect the conditions that move populations down the echelon ladder.</strong> The twelve-fold increase in Japanese female survival to age 80 between 1950 and 1999 was not achieved by anti-aging medicine. It was achieved by sanitation, nutrition, vaccination, antibiotics, and basic chronic-disease management. The law says those gains were phase transitions, not incremental improvements — and that the next transitions are available to the populations that still sit in higher echelons. The most cost-effective “longevity intervention” on Earth remains bringing everyone down to at least the conditions of 1999 Switzerland.</li> </ol> <p>None of this is immortality. But the law — read seriously, two decades on — says the ceiling is not where the orthodox theories put it, and the dials that move it are more accessible than we thought. The gap between what the data supports and what the biology explains is, at this point, a funding question.</p> </section> <section class="sources"> <h2>Sources</h2> <ul> <li>Azbel’, M. Ya. (2004). “Immortality as a physical problem.” <em>arXiv:q-bio/0403008v2</em>. <a href="https://arxiv.org/pdf/q-bio/0403008v2" rel="noopener">https://arxiv.org/pdf/q-bio/0403008v2</a> — <em>the main study; all equations, predictions, demographic figures, and the country list are from this paper.</em></li> <li>Azbel’, M. Ya. (2004). “Universal mortality law and immortality.” <em>Physica A</em> (the peer-reviewed publication of the same work). <a href="https://www.sciencedirect.com/science/article/abs/pii/S0378437104007812" rel="noopener">https://www.sciencedirect.com/science/article/abs/pii/S0378437104007812</a></li> <li>Azbel’, M. Ya. (2002). “Law of universal mortality.” <em>Physical Review E</em> 66, 016107. <a href="https://link.aps.org/doi/10.1103/PhysRevE.66.016107" rel="noopener">https://link.aps.org/doi/10.1103/PhysRevE.66.016107</a></li> <li>Gompertz, B. (1825). “On the nature of the function expressive of the law of human mortality.” <em>Phil. Trans. R. Soc.</em> A 115, 513. Context in Kirkwood, T. B. L. (2015), <em>Phil. Trans. R. Soc. B</em> 370, 20140379. <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC4360127" rel="noopener">https://pmc.ncbi.nlm.nih.gov/articles/PMC4360127</a></li> <li>“Gompertz–Makeham law of mortality” (background on the formula). <a href="https://en.wikipedia.org/wiki/Gompertz%E2%80%93Makeham_law_of_mortality" rel="noopener">https://en.wikipedia.org/wiki/Gompertz%E2%80%93Makeham_law_of_mortality</a></li> <li>Curtsinger, J. W., et al. (1992). “Demography of genotypes: failure of the limited life-span paradigm in Drosophila melanogaster.” <em>Science</em> 258, 461. <a href="https://www.science.org/doi/10.1126/science.258.5081.461" rel="noopener">https://www.science.org/doi/10.1126/science.258.5081.461</a></li> <li>Mair, W., Goymer, P., Pletcher, S. D., & Partridge, L. (2003). “Demography of dietary restriction and death in Drosophila.” <em>Science</em> 301, 1731. <a href="https://www.science.org/doi/10.1126/science.1083819" rel="noopener">https://www.science.org/doi/10.1126/science.1083819</a></li> <li>Arantes-Oliveira, N., Berman, J. R., & Kenyon, C. (2003). “Healthy animals with extreme longevity.” <em>Science</em> 302, 611. <a href="https://www.science.org/doi/10.1126/science.1089269" rel="noopener">https://www.science.org/doi/10.1126/science.1089269</a></li> <li>Kenyon, C., et al. (1993). “A C. elegans mutant that lives twice as long as wild type.” <em>Nature</em> 366, 461. <a href="https://www.nature.com/articles/366461a0" rel="noopener">https://www.nature.com/articles/366461a0</a></li> <li>Podshivalova, K., Kerr, R. A., & Kenyon, C. (2017). “How a mutation that slows aging can also disproportionately extend end-of-life decrepitude.” <em>Cell Reports</em> 19(3), 441. <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC5526670" rel="noopener">https://pmc.ncbi.nlm.nih.gov/articles/PMC5526670</a></li> <li>Vaupel, J. W., Carey, J. R., & Christensen, K. (2003). “It’s never too late.” <em>Science</em> 301, 1679.</li> <li>Vogt, T. C., & Missov, T. I. (2017). “Estimating the contribution of mortality selection to the East–West German mortality convergence.” <em>Population Health Metrics</em> 15, 33. <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC5606051" rel="noopener">https://pmc.ncbi.nlm.nih.gov/articles/PMC5606051</a></li> <li>Swindell, W. R. (2011). “Dietary restriction in rats and mice: a meta-analysis and review of the evidence for genotype-dependent effects on lifespan.” <em>Ageing Research Reviews</em> 11(2), 254. <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC3299887" rel="noopener">https://pmc.ncbi.nlm.nih.gov/articles/PMC3299887</a></li> <li>Human Mortality Database. University of California, Berkeley (USA) and Max Planck Institute for Demographic Research (Germany). <a href="https://www.mortality.org" rel="noopener">https://www.mortality.org</a></li> </ul> </section> </article> <style> article { max-width: 100%; margin: 0 auto; font-family: Georgia, 'Times New Roman', serif; line-height: 1.65; color: #1a1a1a; padding: 0 1rem; } article header h1 { font-size: 2rem; line-height: 1.25; margin-bottom: 0.5rem; } article .dek { font-size: 1.05rem; color: #444; font-style: italic; margin-bottom: 0.5rem; } article .byline { font-size: 0.9rem; color: #666; border-bottom: 1px solid #ddd; padding-bottom: 1rem; } article h2 { font-size: 1.35rem; margin-top: 2.2rem; margin-bottom: 0.6rem; } article p { margin: 0 0 1rem; } article .formula { font-family: 'Courier New', monospace; background: #f4f4f4; padding: 0.6rem 1rem; border-left: 3px solid #888; margin: 1rem 0; font-size: 0.95rem; } article ol { margin: 0 0 1rem 1.5rem; } article ol li { margin-bottom: 0.6rem; } article strong { color: #111; } article em { color: #333; } article a { color: #1a5276; text-decoration: underline; } article .bottom-line { background: #f9f9f5; padding: 1rem 1.2rem; border: 1px solid #e0e0d6; border-radius: 4px; margin-top: 2rem; } article .bottom-line h2 { margin-top: 0.3rem; } article .sources { font-size: 0.88rem; color: #444; border-top: 1px solid #ddd; padding-top: 1rem; margin-top: 2rem; } article .sources h2 { font-size: 1.1rem; } article .sources ul { margin-left: 1.2rem; } article .sources li { margin-bottom: 0.4rem; } </style> {/html}
Password Summary of changes

📜 ⏱️ ⬆️