A bundle looks like a discount, but its real job is to smooth over the fact that customers disagree about what each piece is worth. George Stigler noticed this in 1963 while studying why film distributors made theaters book movies in blocks. If one theater values film A highly and film B barely, and a second theater feels the opposite, a single package price can capture most of what both will pay, while separate prices force the seller to choose between high prices that lose a buyer and low prices that leave money on the table. The economists William James Adams and Janet Yellen formalized that logic in 1976.

Unbundling runs the machine in reverse, and Netflix learned how fast. In July 2011 it split its $9.99 plan, which covered both streaming and DVDs by mail, into two $7.99 plans. Customers who wanted only one service got a small cut, but everyone who wanted both saw their bill rise 60 percent, to $15.98. In the following quarter Netflix’s US subscriber count fell by roughly 800,000, its share price lost most of its summer value, and the plan to spin the DVD business off as “Qwikster” was abandoned within weeks of being announced.

The subscribers who left were not people who disliked streaming. They were households whose uneven valuations had been averaged into one price they accepted, and the split forced them to price each piece on its own. When you remove a bundle, the real question is not whether each component is fairly priced, but how many customers were only there because of the average.