“Who by fire”: lifespans of websites from a web archive perspective 16:00-16:30

Russell Latham, National Library of Australia

Web archives provide a valuable resource for researchers by providing them with a contemporary snapshot of original online resources. The value of archived content increases as the original is altered, migrated or taken offline. These changes occur dynamically on the web. Sometimes change occurs quickly but sometimes it is a slow change with the website evolving through iterative updates until it eventually become unrecognisable from the archived copy. Some studies state that websites have a lifespan of only 40 to 100 days. If true, this would mean that most websites evolve very quickly after archiving. As any web archivist can tell you however, some websites have remarkable longevity and can remain live, accessible and little changed for many years. When can a website be considered gone or using an anthropomorphised term, ‘dead’? At its simplest, a website is dead when its URI (or domain) vanishes, along with all its hosted content. However, for many websites, the story is not this simple and they fall into a grey area. A URI may remain stable but the content on the website changes or, the content may remain but migrates to a new URI. In both these scenarios the website has changed substantially, but is this enough to say the website has ceased to exist and is ‘dead’?

In this presentation, I will look at the key characteristics in the lifespan of a website and its eventual ‘death’. It will also examine if a typology can be applied that will allow curators to identify websites that are most at risk of disappearing. I will do this by firstly examining a sample of the National Library of Australia’s twenty-two year old Pandora archive to find trends that lead to the end of its life. Second, I will apply this quantitatively to the NLA web archives to see if websites disappear by sudden death or slow decay.

By examining the lifecycle of a website we will better understand the critical junctures of a websites existence online and by virtue of this provide curators with greater understanding of the best timing when determining harvests schedules. Typologies assist curators to predict the likely future path of a website and allow them to make appropriate preservation actions ahead of time. As curators we are also always trying to determine what within our collection is unique material and what is no longer available and by being able to pinpoint an end point of a website we are better placed to answer this question.