Both are good points. It would probably be more accurate to use life expectancy at 25... so it starts around the time people start their careers, and it exclude

Both are good points. It would probably be more accurate to use life expectancy at 25... so it starts around the time people start their careers, and it excludes all children deaths (who never contributed). If we did that, it's an increase from 71.5 to 80 (more than doubling retirement years, from 6.5 years to 15). Source is here, starting on page 53, there's a nice table of life expectancy by age and date: https://www.cdc.gov/nchs/data/nvsr/nvsr65/nvsr65_08.pdf https://www.cdc.gov/nchs/data/nvsr/nvsr65/nvsr65_08.pdf