The Memento Project’s archive access additions to HTTP have enabled development of new web archive access user interfaces. After experiencing this web time travel, the inevitable question that comes to mind is “How much of the Web is archived?” This question is studied by approximating the Web via sampling URIs from DMOZ, Delicious, Bitly, and search engine indexes and measuring number of archive copies available in various public web archives. The results
indicate that 35%–90% of URIs have at least one archived copy, 17%–49% have two to five copies, 1%–8% have six to ten copies, and 8%–63% at least ten copies. The number of URI copies varies as a function of time, but only 14.6–31.3% of URIs are archived more than once per month. …..
Although our research shows 35–90% of public URIs have at least one memento, coverage is inconsistent and appears dependent on several factors. Human desire for URI publicity appears to be a major factor as shown by the relatively high DMOZ and Delicious archival rates. Search engine discoverability is the next most important factor followed by explicit archiving. The best overall coverage is provided by the Internet Archive. The search engines follow, but only for very recent mementos. The specialized archives provide good coverage for the URIs they cover (but only for the URIs they cover).
Future work will include study of the relationship between the rate of change of the URI and the rate of the archiving process. This work has been done on a general sample of URIs. In future work, archived URIs will be studied based on specific languages beyond English.“
in: JCDL’11, June 13–17, 2011, Ottawa, Ontario, Canada.
Link to PDF