I did not read all the papers

The other piece here makes the argument that independence is not something we accumulate. We lose it, slowly, starting from now. Your expected years of independence are at their highest today and they only go down.

I will not make that argument twice. This is about what I did next.

Observation: there is a lot of noise about living longer, and very little about the thing that actually decides whether those years are any good.

Once I accepted that this is a losing game, the question became what slows the loss. So I went looking.

What I found first was the loud part. Biomarkers. Continuous glucose, ApoB, epigenetic clocks, VO2 max as a longevity score. All of it aimed at preventative health and living longer, all of it heavily discussed, some of it good.

What I did not find in that conversation was the other body of work. There are decades of research on the aging years themselves. Not how long people live, but whether they can still cross a road, get up off the floor, and manage their own house. Thousands of people, followed for years, measured with a stopwatch and a chair.

Nobody talks about it. It does not sell a subscription.

That gap is the reason this calculator exists.

How I actually did the research

Observation: I am an engineer and a builder. I have a PhD. I know how to do a literature survey. So I sat down to read the papers.

I lasted about an hour.

My patience for sitting with a full paper has dropped a lot in fifteen years. That is the honest version. I can still read a conclusion, follow a data table and tell whether a chart supports what the abstract claims. What I no longer do is work through forty papers front to back.

So I used AI to do the part I was going to do badly. Search, sift, pull out which papers actually matter for this question, put the numbers side by side. Then I read the conclusions and followed the tables and charts myself.

This is not an exhaustive review of the literature, and I am not going to pretend it is. Someone who did this properly would have read more, and would have read it more carefully than I did. What I have is a serious attempt by somebody who cared about getting it right, built with a tool that is good at finding things and bad at being trusted.

Observation: a tool that is bad at being trusted needs checking, and that check found something.

Every reference behind this calculator was looked up against its record. Thirty had never been checked. Twenty-eight were fine. Two could not be found at all and are now labelled as such.

Three were wrong. The one worth telling you about was the push-up study. The finding was right, the sample size was right, and the journal, volume and page numbers were completely invented. A real paper, pointed at an address that does not exist. It was live on this site for a week.

That is exactly the mistake this method makes, and it is why the bibliography now says which entries have been verified and which have not.

What the research actually says

Two measurements have serious weight behind them.

How fast you walk. 51,248 people across 79 studies, and the result holds up across countries. About 1.2 metres per second is what it takes to cross a road on the signal. Below that, the built world stops cooperating with you.

Aerobic capacity. 16,278 people measured on a treadmill in 34 laboratories. Below roughly 18 for men and 15 for women, living on your own stops being sustainable.

Then a set of tests with thinner evidence but a clear pass mark: getting off the floor without a hand, standing on one leg with your eyes closed, getting out of a chair five times. Useful as pass or fail. Not useful for telling you where you rank.

There is no published way to combine any of this into one score. I looked. No weights exist that would justify a formula, so this calculator gives you one target per measurement and no total.

And the thing that made me finish building it: run the numbers on an average 47-year-old man and he is already below what he would need to still be managing at 85. Not on one measurement. On both of the strong ones.

Not ill. Not diagnosable. Just behind, with nobody telling him.

The details

That is the summary. The method, the formulas, every reference and the two that could not be verified are all written out.

For any interested soul, or their LLM: the method and the evidence.