In IDG's Darwin Magazine, Alan T. Schroeder Jr. writes about the growing problem of link decay, suggesting that it may be "time for the Internet Engineering Task Force to implement a solution":
"To most of us, inactive or incorrect URLs are nothing new... However, increasingly for undergraduates and professional researchers... websites are primary resources... In a recent issue of Science, [Robert] Dellavalle and colleagues determined after only three months, the journals New England Journal of Medicine, Science and Nature had 3.8 percent of their article references inactive. After 15 months, 10 percent were found inactive and after 27 months, 13 percent were inactive. These journals are print staples in most research libraries and if references in these heavyweights are susceptible to incomplete research, imagine the percentages in less reputable publications related to URL longevity. The implications of his findings should transcend libraries and be looked at through a records and information perspective in general. Imagine the expert witness with 'disappearing research' or the suddenly phantom case citation that your career hinges on?
"These disappearing acts extend into government. A representative of Electronic Publishing Services, Ltd., in London, was quoted as saying 25 percent of all British government websites change their URLs each year.... [The] Executive Office of the President (EOP) and United States Trade Representative were sued over randomly and inconsistently removing Web pages...
"The problem has technology and human components... As server hosts come and go, mergers occur, data migrates to new systems and companies go bankrupt, expect this problem to continue. Expect URLs to continue to disappear with no explanation. And expect to see a lot of conflicting and duplicative research."
[Thanks to Marcus Zillman for the pointer.]