This doesn't make people safer. Left Off Colfax quotes the Colorado Department of Transportation on the issue:
“Before and After” speed studies show that there are no significant changes in vehicle speeds after speed limits are changed. “Before and After” accident studies usually do not show any significant change in accident rates after speed limits are increased or decreased. National studies go further and say that “it is generally at the upper boundary of a speed range where crash involvement rates are lowest.”...
Traffic investigations have shown that most people will drive the roadway as they perceive the conditions and will ignore a speed limit that is unrealistically too low or too high. A realistic speed limit is voluntarily obeyed by the reasonable majority and more enforcement effort can be applied to the unreasonable few who drive too fast or too slow. . . .
An appropriate, “just right” speed limit will result in the maximum number of vehicles traveling at about the same speed, thus reducing conflicts caused by speed differentials.
The 85th percentile speed, that speed at or below which 85% of the traffic is moving, is widely accepted as being closest to that “just right” speed limit - a case of Majority Rule. . . .
Several studies have demonstrated that 85th percentile operating speeds typically exceed posted speeds. These studies also show that the 50th percentile operating speed either is near or exceeds the posted speed limit.
He also cited the National Cooperative Highway Research Program which notes:
[A] review of available speed studies demonstrates that the 85th percentile speed is only used as a “starting point,” with the posted speed limit being almost always set below the 85th percentile value by as much as 8 to 12 mph.
Speed limits are a classic case where empirical fact should drive the law and doesn't.