They also report that under 17% of trips are for work, which is surprisingly low and seems like too much noise to learn much about the distribution of peak hour

They also report that under 17% of trips are for work, which is surprisingly low and seems like too much noise to learn much about the distribution of peak hour commuting distances. If I had a five mile commute, I would use surface streets. Congested highways are full of people going far enough in roughly that direction that even the congestion plus the first/last mile is faster than countless stoplights on the direct route.