Interesting proposal, but I’m thinking it might not be necessarily so. Life expectancy is an average, and averages do not need to be exactly at 50% of the population. That’s the median, you are thinking of the median. It is quite more complicated than that, as there are hundreds of factors that alter life expectancy, thus there are many life expectancy tables depending on gender, lifestyle, cohort, race, income, etc. It will not amount to 50% of the population, specially for populations with a lot of outliers. People who die soon after birth or live exceptionally long lives will skew the average deviating it from a true central tendency. As people live longer and less people die early, then the median (the age at which at least 50% of a cohort, people born the same year, dies) will shift away from the average and occur later. If a lot of teenage and young people die, but the survivors live long lives, then the median will occur sooner. The technical terms are skewness and kurtosis of a probability distribution.