It comes down to risk. Using the
Georgia Tech risk model, plugging in either Carroll or NH numbers, there is a 1% chance that someone that you spend time with will be contagious. If reported numbers are low (likely), say 10X low, then there is a 10% chance that someone that you meet is contagious. Using statistical methods, a party of five has a 2% chance that one person will be infectious (20% if 10X underreported). In some states, the risk is much higher.
The new state guidelines seem reasonable to me. Several weeks after the holidays, as winter hibernation sets in, the spreading rate should settle down.