Increasing returns pose two major conceptual difficulties. First is the problem of market structure: When increasing returns are prevalent, an industry tends to have only a small number of firms, which therefore find themselves in an uneasy tension between competition (each would like to increase its market share) and cooperation (it is in the common interest of all the firms that they not engage in price wars). It is hard to predict how this tension will be resolved. Second is the problem of expectations: My decision whether or not to buy a Macintosh depends not only on how many people are currently using Macs but also on how many people I think will be using them a couple of years from now--which depends on how many people they think will be using Macs. How does one resolve the possibility for self-fulfilling prophecies?

The striking thing about Arthur's work was that he simply punted on both questions, taking the size of firms as a given and assuming that individuals make their decisions on the basis of the current situation, ignoring expectations about the future. In the story according to Arthur, people kept on telling him that his work "wasn't economics," which he interpreted to mean that increasing returns were beyond the pale. But since several hundred papers on increasing returns were in fact published in professional journals during the years people told him this, he must have misheard. Probably people were really telling him, "You haven't addressed the economic issues created by increasing returns"--a rather different complaint.