Circadian rhythm is 40-70% genetic. And there's an evolutionary advantage to not having everyone asleep at the same time. Early humans would have been safer at night with some people in the group awake.
But how big is the inter-person variation? "Out of some 200 hours for the entire study, for only 18 minutes were they actually all sleeping synchronously" is a pretty useless quote -- how much of that was due to a handful of extreme outliers? People's sleep times were obviously not evenly distributed over the other 1422 minutes per day. And obviously environment is going to play an outsize role in causing outliers, since people who work the graveyard shift aren't going to spontaneously revert to a normal sleep schedule when they enter a sleep lab.