There have been some notable exceptions to this pattern. In 1964, Barry Goldwater won the Republican nomination standing well to the right of the average voter, and was unable or unwilling to reposition himself in the center. In 1972, George McGovern won the Democratic nomination standing well to the left of the average voter, and was unable or unwilling to reposition himself. Both lost in landslides.
A problem with the Hotelling model when applied to commerce is that the results are very sensitive to the cost assumption. There must be some cost to traveling because customers prefer the closest vendor. But these costs must be small, because the people at the end of the beach continue to buy the same amount no matter how far they are from the nearest vendor. If traveling costs are less, then people might not care whether they go to the nearest vendor. If they are greater--so that when the vendor gets far away--people do not bother to go, the vendors will no longer cluster at the middle.
Suppose that the beach is a long beach, and people more than 1000 feet away from any seller buy nothing. Also assume that the beach is 4000 feet long, and the two vendors start at the middle. Originally George sells to customers located from the 1000-foot mark to the middle at 2000 feet, and Henry sells from 2000 feet to 3000 feet. If George moves to the 1000-foot mark, he will gain 1000 feet of new territory, and he will lose only 500 feet to Henry. At the 1000-foot mark, he will sell to all people from 0 to 1000 feet. He will also sell to those people between him and Henry who are closer to him. Because Henry did not move, but stayed at the 2000-foot mark, George will get all the customers up to the 1500-foot mark. Equilibrium in this case will occur only when Henry moves to the 3000-foot mark.
In Hotelling's original model with small traveling costs, location decisions were not economically efficient. By increasing traveling costs, it seems that we can have location decisions that are economically efficient. However, the next section shows that adding transport costs results in new efficiency problems.