This paper investigates whether larger quantities of historical data confer a competitive advantage to firms that offer Internet search. We study how the length of time that search engines retained their server logs affected the apparent accuracy of subsequent searches. Our analysis exploits...