The paper The Missing Link: Assessing the Reliability of Internet Citations in History Journals presents an alarming conclusion:
The worldwide web has offered an increasingly common though ephemeral source of information. In research articles in two of the most highly respected history journals, 18 percent of web citations decayed within seven years of publication; 10 percent were inactive shortly after publication. Our findings are roughly consistent with those for science journals; we suspect that this problem extends to other humanities and social science publications. A means created to preserve internet sites — the Wayback Machine — made 57 percent of the missing articles in our sample available to scholars who knew about the archive. The other 43 percent of the missing links remained beyond the reach even of those searching the archive.
This means that around half of all the cited web pages are not accessible at all on a long term basis. This should come as no big surprise. In fact, based on my personal experience I would think that the numbers could even have been bigger. It could be argued that things change rapidly nowadays, and particularly the growing number of digital libraries at universities tend to focus on creating web links that are supposed to last “forever”. Still, it is good to think about our citing practices in our scientific writing.