This is written in Dec 2016, probably in response to the downturn in life expectancy that is believed to be due to the opioid epidemic. Don't know how the massaged the data, but they're graphing a life expectancy value that would result if the current infant mortality rate is held constant over the life of the infant. I think that might tend to balance out the infant mortality rate reduction component, but again, I don't know how they were able to isolate that data.

Also, thinking about "average age at death", given the raw data, you could probably come up with a value for average age at death for all ages greater than 18 or so. That would tend to isolate infant mortality changes, and would also sharpen the focus on the desired population, those who live long enough to start paying into social security. I think if that quantity were able to be isolated, and then graphed over time, you would see meaningful data about whether or not adjustments to social security eligibility ages would be needed.