there is a fundamental problem here. You can guess if some longevity biomarker has a causal effect; but in the end they are only guesses. Suppose your desired endpoint is a human lifespan of 150 years. And you start experimenting on 50 year-olds, now. You cannot truly know if your intervention will work for another 100 years. At best we can start doing interventions on shorter lifespan animals (say, dogs) but there's still no guarantee that the causal factors for dog death match up with those for humans.
Not dogs but mice. Mice are members of the supraprimate superoder (the euarchonta). Dog are more distantly related to humans than are mice and they live much longer and they cannot be genetically engineered as easily. We have already established causality for aging in mice using classical genetic methods—down to the level of large chromosomal stretch that usually contain 10 to 200 candidate genes. But we do not yet know the causal single gene variants that modulate lifespan in mice but we are systematically getting closer.
Will these putative genes also modulate healthspan in humans? That is a reasonable presumption that will need to be tested, with the recognition that gene-by-environment interactions will be most important.
Mice are weird in ways that disqualify them for aging research, IMO. For example, they and rats have two insulins (no other species does). We already know the insulin/igf-1 axis is important for aging.
This feels like saying that a five-year MTBF estimate for a disk drive is meaningless until it's been tested for five years. No, we don't truly know if the estimate is correct, but nothing in life is truly known: some level of evidence will be good enough.
Yes but what evidence? We don't have any biomarkers that can be measured to determine the longevity effect of a particular intervention. For example if a hypothetical medication reduces blood glucose level (good) but also reduces skeletal muscle mass (bad) will that have an overall positive or negative effect on longevity? Who knows? It's a complex multifactorial issue.
Disc drives are much simpler. We have decades of data on failure rates.
I would say that the simplicity has to do with the linear nature of most failure modes of disk drives. The overall failure curve is nonlinear due to compounding per time probabilities. But those probabilities are only nonlinear/cross-interacting at the point where the drive is already considered to be "a loss", and our engineering/maintenance strategies take advantage of this fact. We do not generally consider a human to be "a loss", if say, they start getting chronic back pain, which can start to precipitate other problems in a compounding fashion.
Not to mention the fact that even if it works in removing some ageing factors, most of your cohort still probably ends up running into the sort of ailments that terminate most people's lives before the theoretical limit to their lifespan. And those other ailments represent challenging, but more easily testable medical problems to solve first
