The field of epigenetic clocks and "biological age" has built an impressive edifice on correlational noise that we're mistaking for mechanism.

We measure DNA methylation at a few hundred sites, feed them into a machine learning model trained on chronological age, and then claim we've discovered the true rate of aging. But the correlation between these clocks and actual health outcomes is modest at best. The DunedinPACE paper reported a hazard ratio of 1.5 for mortality. That's meaningful, but it's not a clock—it's a rough weather vane.

What if we've been confusing a biomarker for a mechanism for decades?

The epigenetic changes we observe with age are largely downstream epiphenomena—the visible rust on the ship, not the runaway engine. Methylation shifts reflect transcriptional chaos, cellular identity drift, and accumulated dysregulation. But they're not driving aging. Treating them as the "cause" is like treating a fever as the disease itself.

These measures aren't useless. They're correlated with mortality risk, and that has value for epidemiology. However, we're making fundamental errors:

1. Assuming correlation across populations means mechanism within individuals
2. Ignoring that methylation is tissue-specific while these clocks rely on blood-based averages
3. Pretending a few hundred CpG sites capture the complexity of systemic aging

The real issue is how much funding has flowed into clock refinement rather than mechanism. We've optimized our measurements of a phenomenon we don't understand. It's the biomarker equivalent of drawing increasingly precise maps of a city we're not allowed to enter.

What we need isn't better clocks. We need to return to first principles and ask what actually drives the aging process at a causal level. The clock literature has given us a false sense of progress.