On Solving Aging

On scaling for aging

Aging is the upstream cause of nearly everything that kills us. It is also a complex learning problem, one that should yield to scale.

Felix Wong · Redwood City, 2026

Why does everyone have to die? I've been asking this since I was six—the year my grandpa died—and I haven't heard a real answer. The most common non-response is "entropy", a law of closed systems.

Human mortality has long been described by Gompertz's law, which states that the risk of death rises exponentially with age in adulthood (doubling approximately every 7-8 years in humans). This statistical regularity reflects the specific evolutionary pressures that shaped our cellular repair and rejuvenation capacity. However, Gompertz's law is not a physical constraint.

In fact, multiple species have evolved sufficient repair and anti-aging mechanisms to evade the typical Gompertzian decline. For instance, certain turtles and tortoises can live for many decades with little increase in mortality risk, and the ocean quahog clam and Greenland shark can survive for centuries while maintaining function. Notably, the naked mole-rat, a rodent roughly the size of a mouse, shows no rise in mortality hazard even at 25-fold beyond reproductive maturity, effectively defying the aging curve that governs other mammals. The independent evolution of "negligible senescence" across such diverse taxa strongly suggests that aging is not an inescapable phenomenon, but rather a complex phenotype that can be modulated. Decelerating aging is an engineering problem that biological systems have solved multiple times in different ways.

Humans have evolved under limited selective pressure to extend lifespan beyond the reproductive and child-rearing years. The absence of strong evolutionary optimization for extreme longevity implies that we are not at the hard biological limit of lifespan. This suggests that there may be considerable "low-hanging fruit" for biomedical interventions to improve maintenance and repair mechanisms and ultimately increase healthy lifespan. Indeed, short-term pharmacological interventions have been observed to alter aging in mice by targeting conserved pathways (e.g. IIS/mTOR, sirtuins, AMPK). These interventions show that treating aging as a modifiable condition is scientifically tractable.

My hypothesis is that the obstacle to modifying aging has been search. Aging is complex, slow to read out, and importantly, we don't know where to look. Many failures in aging drug discovery haven't been failures of biology; rather, they've been failures of prior. We all-too-often impose a strong belief about mechanism onto a system whose mechanisms we cannot yet understand, and we all-too-often are confidently wrong.

Every discovery strategy is a choice about how strong a prior to impose. Put another way, there are three regimes, and one way to view them is as a continuum indexed by the ratio of what we know to what we don't know.

01 · Strong prior

Aim

Start from a mechanism and interrogate it deeply.

E.g., mTOR, cellular senescence, the DNA-damage response; rapamycin came out of this approach. Buys mechanistic depth and precisely targeted interventions, but pays for it in scope: you can only find what you already suspected.

02 · Weak prior

Sweep

Test many interventions in parallel, under one standard.

The NIA Interventions Testing Program is a good example. Buys freedom from tunnel vision, but pays in cost and in false positives that require investment to understand.

03 · No prior

Saturate

Let the data decide what works.

Ultra-high-content phenotypic screens across millions of perturbations in aged cells and organoids, read out by learned classifiers answering a deceptively simple question: does it look younger? Buys novel mechanisms, but pays in mechanistic interpretability.

These strategies are complementary, and any serious program should run all three. In my view, the most important approach in our time is saturate: in this regime, every additional well-designed perturbation sharpens our hypotheses, densifies the data supporting them, and improves the next round of hypothesis generation. The other two strategies are inherently limited by our priors. The best approach, then, is to shift weight rightward to saturate as fast as the data allow, and to use the leftward regimes as instruments for interpreting (in the usual, mechanistic biology sense) what the right-hand one finds.

Yet, it's not just about scale; it's also about the quality of the data. The best data to enable the saturate strategy share four properties: they are perturbational, clean, high-content, and anchored.

  • They are perturbational rather than observational: this is because aging remains a causal challenge, and correlations drawn from cross-sectional cohorts cannot address it.
  • They are clean, generated with tools such as optogenetics that switch aging-associated pathways on and off with temporal precision, rather than inferred through chemical insult.
  • They are high-content: for instance, an image of a cell carries orders of magnitude more information about its state than any single endpoint.
  • And they are anchored, or tied back to human longitudinal reality, so that the model is learning about us and not about a well plate or imaging setup.

These high-quality data should feed a flywheel (as they always do). Here, the flywheel is the closed loop of scientific understanding: models that propose hypotheses, a high-throughput platform that tests them against living systems, and a rigorous path by which experimental truth—including negative results—returns as signal. Building this closed loop is the fundamental challenge, and programs and partnerships are evidence of it working to create value.

Here's my main bet: if aging is modifiable, if it is reversible, and if it is measurable as a phenotype, then it is learnable. And if aging is learnable, then generating the right kinds of biological data at sufficient scale, inside a closed loop that refines hypothesis generation, will converge to the most fundamental insights about aging biology. Such a discovery process would enable us to intervene against aging itself, rather than only against age-associated diseases.

The current state of aging research is reminiscent of where oncology was less than half a century ago. Once a "death sentence," cancer is now increasingly manageable due to systematic scientific progress. As our scientific understanding deepens, aging may be slowed and perhaps even reversed. The most high-value question we can address now, then, is: how do we build the engine that addresses the biology of aging at scale?