As academic publications age, the links in them that point to the web locations of citations or other resources begin to decay, writes Jeffrey Perkel at Nature.
A recent survey from Herbert Van de Sompel at the Los Alamos National Laboratory Research Library and his colleagues found that about a million 'web-at-large' links contained within 3.5 million articles published between 1997 and 2012 were rotten, with the portion of bad links increasing with publication age.
"You've lost a trace to the evidence that was used in the research," Van de Sompel tells Nature.
Perkel notes that there are some services that now aim to prevent such link decay. For instance, the Internet Archive and Perma.cc offer web-archiving services, and Van de Sompel and his colleagues' Memento service provides users access to all past saved versions of a web page. He tells Nature that he is "unbelievably enthusiastic" about his team's approach.
However, Stanford University's David Rosenthal notes that not all publishing companies archive their web pages — an analysis by The Keepers Registry monitoring service indicates that only about half of articles are saved. Because of this, Rosenthal says an approach like Memento's may be "excessively optimistic."