[sowbug]

It can't. The point you're replying to is that being alive at age X gives less information about longevity than at X+1. For the younger person we don't know whether they'll make it to age X+1. For the older person we have the additional data that they did make it. That data is what makes the older person's life expectancy higher.

[colanderman]

Not sure what you mean, the table [1] is expressly in terms of current age. The better way to approach such a "death clock" I think would be to slow the countdown rate to match the curve implied by that data.

[1] http://www.health.state.ny.us/health_care/medicaid/publicati...

[sowbug]

Pretty sure this is a Bayes' Theorem thing, but happy to be wrong and learn something.

I think we both agree the table talks in terms of current age. But making it to age 50 doesn't mean you'll make it to age 51, so you can't add that extra bit of life (the one you think of in terms of a day lived counting a little less than a day against your life expectancy) to the age 50 row in the table. You can't add that bit until you actually are age 51. Otherwise you're counting both a known thing (you did make it to 51) and an unknown thing (whether you will make it to 51 if you're only 50 today) to a predicted thing (when you'll die if you're 50 today). Not exactly double counting, but taking for granted something that's not yet proven.

I believe that extra bit is what the other commenter was wondering about.