Will you live to 120? · The Beautiful Wrong Idea

Where the machines come in

And the most direct test of all: researchers built a “mutator” mouse, engineered so that its mitochondrial DNA accumulates mutations far faster than normal. By the theory, this should trigger the ROS vicious cycle and produce accelerated aging — and indeed these mice do age prematurely. But when the experimenters looked for the supposed engine of that aging, the runaway ROS, it wasn’t there. The mice were not drowning in extra reactive oxygen. Their premature aging was driven by the mitochondrial mutations crippling the electron transport chain directly — by their power plants simply failing — not by a self-amplifying spiral of oxidative damage. Mitochondrial DNA damage mattered. The specific mechanism the theory proposed for how it mattered did not.

Deeper Water — what is actually true about mitochondria and aging. It would be a serious mistake to walk away from this chapter thinking “mitochondria don’t matter” or “ROS are harmless.” Neither is true, and the honest modern picture is more interesting than the theory it replaced. Mitochondrial dysfunction is a genuine hallmark of aging: as we get older, our mitochondria really do work less well — the electron transport chain becomes less efficient, mitochondrial DNA accumulates mutations and deletions, and, critically, the cell’s quality-control systems falter. That last part is where the real action turns out to be. Cells continuously police their mitochondria: they tag damaged ones for destruction and digestion (a selective cleanup process called mitophagy), and they build fresh ones (biogenesis), and they constantly fuse and split them to share components and isolate damage (dynamics). With age, this housekeeping degrades — damaged mitochondria are cleared too slowly and replaced too rarely, so dysfunctional ones accumulate. The modern view, then, is not “oxidative damage piles up and ages us” but something subtler: mitochondria decline in function, signaling, and turnover, ROS act as both damage and essential signal with an optimal intermediate level, and the failure is one of maintenance and quality control more than of simple corrosion. The mitochondrion remains central to aging. Harman was right that it matters and right that ROS are real; he was wrong about the shape of the relationship — a curve, not a line — and wrong that mopping up ROS would help. Which is exactly why the therapies that may actually work, as we’ll see, target quality control — clearing and replacing bad mitochondria — rather than blanket-suppressing the chemistry.

Where the machines come in

The collapse of the antioxidant idea reshaped the search for mitochondrial therapies, and the new direction is mechanistically smarter — and a good test of whether we’ve actually learned the chapter’s lesson. Instead of flooding the whole body with antioxidants, the targeted approaches try to be selective: antioxidants engineered to home specifically into mitochondria rather than wash through every cell (so they don’t garble signaling everywhere), and — more promisingly — drugs that boost the quality-control systems rather than the chemistry. The standout example is a compound called urolithin A, a molecule the gut produces from certain fruits, which induces mitophagy — prompting cells to clear out their damaged mitochondria — and which, in human trials, has shown modest improvements in muscle function. The whole strategy has shifted from “mop up the rust” to “take out the broken machines and build new ones,” which is the right lesson drawn from everything above.

Artificial intelligence enters here in the now-familiar role: screening for and designing molecules that enhance mitophagy or improve mitochondrial function, mining large datasets to work out which mitochondrial-DNA mutations and which mixtures of healthy-and-damaged mitochondria actually matter for a cell’s fate. These are real and useful applications. But this chapter, more than any other, should make you cautious about the style of confidence that AI can lend to a hypothesis. The free-radical theory was the most computationally and mechanistically respectable idea imaginable — it had elegant chemistry, decades of supporting data, and an obvious prediction — and it was wrong in a way that no amount of cleverness about the mechanism could have caught. Only the randomized trial caught it. AI can help find a mitophagy-inducing molecule faster than ever before; it cannot tell you whether clearing those mitochondria will make a human being healthier, and it cannot substitute for the large, randomized, placebo-controlled trial that is the only thing that ever could. In a field whose most beautiful and best-supported theory turned out to be inverted, the constraint was never the elegance of the idea or the speed of the search. It was, and remains, the experiment.

✦ ✦ ✦

Carry these three things

1. A mechanism that sounds right — even with correlations and an industry behind it — is not proof. Only the experiment is. The free-radical theory of aging had everything: elegant chemistry, observational support from vegetable-rich diets, dedicated cellular defenses confirming ROS matter, fifty years of dominance, and a clear prediction. The prediction was tested in massive randomized trials, and it failed — beta-carotene given to high-risk smokers to prevent lung cancer increased it, and pooled antioxidant trials showed no benefit and some harm. The vegetables-correlation was confounded (vegetable-eaters differ in a hundred ways); the randomized trial cut the tangle and revealed the truth. This is the single most useful habit this book can give you.

2. The dose makes the poison — and biology runs on curves, not lines. ROS are not only damage; they are essential signals the cell uses to defend and adapt itself, and the relationship between ROS and health is an inverted U with an optimum in the middle, not a downhill slide. That is why flooding the body with antioxidants can harm — it garbles the signal — and why exercise, which raises ROS, is profoundly healthy: the ROS burst is the signal that triggers the benefit. Antioxidant supplements were shown to blunt key metabolic gains from exercise, by muffling that very signal. “More is not always better” is not a platitude here; it is the literal shape of the data.

3. Mitochondria still matter for aging — just not the way the theory said. The simple “oxidative damage accumulates and ages us” vicious cycle failed every direct test: antioxidant-enzyme mice didn’t live longer, the naked mole-rat lives thirty years steeped in oxidative damage, and the mtDNA-mutator mouse aged prematurely without the predicted ROS spike. The honest modern picture is about declining mitochondrial function, signaling, and quality control — which is why the therapies worth watching aim to clear and replace damaged mitochondria rather than blanket-suppress the chemistry. That idea — that aging is partly a failure to take out the cellular trash — is exactly where the next chapter begins: with the proteins that misfold and clump together, the cleanup crews that are supposed to remove them, and what happens, in the diseases we fear most, when the crews fall behind.

All 10 chapters of Will you live to 120? · Foundations · All courses