Does Aging Stop? was published a couple of years back, and provides a good grounding in the research and viewpoints of evolutionary biologist Michael Rose, whose work is mentioned here and there in the Fight Aging! archives. He and fellow researchers have assembled a compelling set of data regarding increased longevity in flies by selective breeding, and on the late life mortality plateau in flies: if aging is defined as an increasing chance of death per unit time, there comes at point at which aging ceases. In flies at least it's quite clearly the case that their mortality rate flattens out in very late life.
Human data is much less definitive, unfortunately, with a study published last year showing no signs of mortality rates leveling out in in the oldest surviving portions of the population.
The Rose view of aging that emerges from this and related work shares some aspects in common with programmed aging theories (aging is an errant continuation of youthful genetic programs that cause damage and dysfunction) and some with damage based theories (aging is caused directly by accumulated damage at the level of cells and protein machinery). Rose is, for example, not opposed to damage repair initiatives like SENS, but argues that the late-life mortality plateau data has to indicate that damage is not the whole story. Thus by this logic, damage repair biotechnologies cannot produce a comprehensive form of rejuvenation.
(At this point, as I've argued before, I'll say that the cost of proving SENS right or wrong is probably about a billion dollars and ten years, which is far less time and money than any other approach to extending longevity will require to even get started. Proving SENS right is essentially the same as proving one class of theories of aging right and the rest wrong: most of the big ongoing debates in aging research could be solved or bypassed by creating a demonstration of SENS in old laboratory mice).
Here is a review of Does Aging Stop?
The work reported with fruit flies (Drosophila melanogaster) was long and extensive, covering 18 years and 465 generations, supplemented by a lesser amount of work with a related, also shortlived insect, the medfly (Cerititis capitata). The fruit flies averaged about 14 days per generation and lived up to a little past 100 days. Roughly, the length of generations and maximum life-span of the fruit flies in days equaled these data for humans in years, which thus are scaled several hundred times longer.
As for the results, it is consistently shown that the mortality of fruit flies, measured in terms of a probability density function giving the chance of dying in a short time interval, does not indefinitely increase. Instead it levels off or plateaus in later life, approaching a roughly constant value in which about 15-30% of the flies die per day, the variations depending on such factors as whether the flies have been specially bred for longevity. Though this is a substantial attrition rate, it is significant that it does not change much from this point on and particularly does not approach 100%, contrary to the thinking of earlier times. Instead, following the period of "aging" in which mortality rates rise, there is a period of indefinite if still finite length that the authors call "late life" when the organism does not age any further, though the aging that has already occurred is not reversed.
The whole thing is worth reading.