How the math works

The joke is the wrapper. The number underneath comes from the same kind of life table actuaries use, adjusted for a few well-studied risk factors.

Start with a life table

The Social Security Administration publishes, for every age from 0 to 119 and for each sex, the chance that a person of that age dies within the next year. We use the 2023 period life table from the 2026 Trustees Report. With no adjustments, our code reproduces SSA's published life expectancy at every age from 0 to 100 to within 0.01 years.

Adjust for where you live

Life expectancy differs a lot by state: in 2021, men in Hawaii lived 9 years longer on average than men in Mississippi. We take each state's gap from the national figure in CDC's U.S. State Life Tables, 2021 (NVSR Vol. 73, No. 7), apply that gap to the 2023 national table, and convert it to a single multiplier on yearly risk. That gap is why your result may list "Move to Hawaii" as a cheat code. The state rankings show every state's published figure, and each state page shows the curve this adjustment produces.

Adjust for health habits

Each answer maps to a hazard ratio: how much higher or lower yearly risk of death is for people with that answer, compared with a reference group, in large studies that follow people for years. We re-center each factor so the average American's answers leave the base table unchanged, multiply the factors together, and limit the combined effect to between 0.4 and 4 times average risk. Relative risks shrink with age, so the health adjustment fades to half strength between ages 75 and 95.

• Cigarettes. Jha et al., NEJM 2013 (doi:10.1056/NEJMsa1211128). Current smokers ~3x all-cause death rate vs never smokers, ages 25 to 79; quitting before 35/45/55 recovers about 10/9/6 of the ~10 years lost. HRs below are tuned so the model reproduces those year gaps (see tests).
• Exercise. Moore et al., PLOS Medicine 2012 (doi:10.1371/journal.pmed.1001335), Table 3 multivariable HRs; shares from that cohort's participant counts. MET-hours: brisk walking is about 3 MET, so 7.5 MET-h/week is about 150 minutes of brisk walking.
• Body mass index. Koster et al., NIH-AARP cohort, never-smoking men (PMC2642004, Table 2). Shares approximate, from recent NHANES adult BMI distribution.

The quiz also asks how many unread emails you have. That answer never enters the calculation. It shows up in your result marked "No effect" and may end up on the tombstone.

Turn risk into a range

Starting at your age, we multiply the yearly chances of surviving to build a survival curve. The age where it crosses 50% is the median, which is the number on your tombstone. The ages where it crosses 75% and 25% bound the middle half of outcomes.

Couples mode

Couples mode runs the same calculation for two people and puts their survival curves on one calendar. Three numbers come out of it. The chance that one person goes first is the chance, summed over every future year, that they die in that year while the other is still alive. Years together is the average time with both alive. Years solo is the average time with exactly one alive.

These treat the two lifetimes as unrelated. Real couples share habits, money, and a home, so their lifetimes tend to move together. That means the calculation probably understates the years together and overstates the years solo.

What this can't tell you

• Hazard ratios describe groups, not individuals, and they overlap: people who smoke also tend to exercise less.
• Period life tables assume today's death rates hold forever. If medicine keeps improving, real lifespans will run longer.
• We don't ask about specific illnesses, family history, income, or dozens of other things that matter.

So treat the result as a conversation starter, not a forecast. If you have questions about your own health, your doctor is the right person to ask.