Some years ago, New Zealand poet and scholar Michele Leggott was able to date a series of poems by Robin Hyde after noticing that one of the pages of the manuscript bore faint traces of another piece of writing. It was a fragment of a short story that Leggott knew well, and whose date of composition had already been established. What must have happened – she realised – is that Hyde used the page as a backing sheet when working on the story, before rolling it into her machine again to type out the poems. Under ultraviolet light, the slight indentations left on the backing sheet became visible alongside the new words typed in ink.
I often think of this story in relation to the status of digital documents. Objects tell us a great many things, but electronic files aren’t ‘objects’. Had computers existed when Hyde was alive, her work might have been saved on a hard drive instead of being printed on paper. But hard drives don’t last very long. The data they contain gets copied and distributed. File dates change in the process. The digital chain of production leaves no evidence concerning the age of a document, not like the kind found by Leggott: no external, independent physical clues to corroborate the claim to truth of the metadata.
This observation applies by extension to the internet, which is nothing but a very large collection of digital documents. Researchers are very familiar with this set of problems. You enter a string of text into Google. The search engine returns a list of results in order of relevance. Attempts to restrict the search by date are almost always futile. Most often, the actual date of publication of each document has to be assessed individually. Many of them aren’t dated at all. When they are, there is often no clear indication of when they were last modified, and in what ways.
It’s not just researchers or archivists who are affected by this uncertainty: the entire experience of the internet is marked by temporal disorientation, a fact so ordinary that it requires extraordinary events to make it noticeable. Take maps, one of the most common reasons to go online. Maps tell us with a great degree of precision where we are, but seldom when we are. For several months after the Christchurch earthquakes of 2010 and 2011, as Google rolled out periodical updates to its street-level photos of the city, a user moving through virtual Christchurch would stand on a pre-quake corner one moment, then step into the rubble the next. Another step would take them back in time, and so forth, without apparent logic other than the inscrutable schedule for database maintenance at the Googleplex, Silicon Valley. This patchwork of different time points is common to all of Google’s photographic maps, but it only becomes evident after a disaster or major construction project.
I used to think that internet time was flat, and that the internet was an engine for stripping texts of their temporal context. I can see now that internet time is infinitely looped, labyrinthine.
There is the permanent ‘now’ of the social media exchange, undercut by myriad acts of repetition, deletion and revision. But even there, even on what goes by the absurd name of ‘my Twitter timeline’, there is a user who sends day-by-day dispatches from the Second World War and another who types out the diaries of a long-dead writer.
Then there is the great ocean of texts that we navigate spatially, in search of words, and where time has little or no meaning. But time is meaning. Just like dating those poems by Hyde enabled scholars to form a better understanding of her work, so too being able to place all kinds of information in time is a key to assessing its usefulness and truth-value.
There have been attempts to archive the internet, driven by awareness that the mass of documents produced in the early years of this new medium are of immense socio-cultural and historical value, and it’s therefore imperative that they be properly preserved and curated. The most well-known institutional effort is the Internet Archive, a non-profit organisation that began taking regular ‘snapshots’ of the web in 1996, which users can access through its Wayback Machine. However, this is nothing like the actual web, but rather a random collection of pages taken at random moments in time. It’s virtually unsearchable, due to the fact that one needs to know the address of the page to be recalled in the first place, and it’s very difficult to navigate from any of the pages because most of the links are dead.
The most immediate effect of attempting to restore a temporal dimension to the internet, then, is to break the internet. And if it’s true that time is meaning, as I’m suggesting here, this loss is going to get compounded as more and more documents are born digital and published directly and solely on the network, without leaving physical traces for us to interrogate.
If the current juncture teaches us anything, it’s that the reliability of our systems of knowledge production is in crisis. Perhaps the way out of this crisis is to find new ways of telling time.
If you enjoyed this column, buy the issue