What is the oldest website?

If a long-dead site is restored, does it still count?

To mark the 20th anniversary of the world’s first website CERN (The European Organisation for Nuclear Research) launched a project to restore that first site. More specifically this means, in the first instance, restoring the original URL and the files that were there at its earliest iteration. More than that, they also aim to restore machine names and IP addresses to their original state and finally to have the first web address – http://info.cern.ch/ – as a destination at which the beginnings of the web are preserved for the future.

First Website URL page May 2013

CERN's page as at May 2013, announcing the project to reinstate the first website

This started me thinking, is this resurrection of the original files, machine names and IP addresses really the actual first website? What does the re-creation of a virtual artefact mean? There is evidence to suggest this site disappeared, was indeed lost, years ago. As always the Internet Archive’s Wayback Machine provides an interesting perspective. The earliest archived instance of the ‘first web address’ in Wayback dates from 10 December 1997 and the result is reasonably helpful page explaining the ‘error you have just encountered’ – a bit more than your brutal 404 Not Found message! So the address did exist at this time but not, so it appears, the content we were expecting. A year later when Wayback again collected the page at this address the information is even more informative saying ‘sorry, the hypertext and WWW information is no longer available on the info.cern.ch site. The physical machine no longer exists’. Then, from January 2002 until August 2006 the URL makes no appearance in Wayback. When it does reappear, we are greeted by Tim Berners-Lee and a message welcoming us to info.cern.ch ‘the website of the world’s first-ever web server’. Which is how it has remained, with only a little bit of a style makeover, until 30 April 2013 when the project to reconstitute the website and server was announced. So, perhaps the site just went missing for a while, the best part of a couple of decades, hijacked for a time by other content. When it returns however, is it the original site or a replica, or a recreation or a new site – well I guess, arguably, all the above in parts. The various iterations discussed here can be viewed through the Wayback Machine. The project to reconstitute the first web page at CERN certainly puts into a focus some very basic issues that web archive curators confront – particularly when put on the spot to reveal the oldest website in the archive. Most obvious is the question: how do we establish the age of the site? Copyright dates may be a guide to identify when old content was produced but can be no more than that - a guide. Dates indicating the last modification or last update of a webpage are certainly more helpful in establishing the age of the page. Other dates and content within the page may help to establish when a page was last updated. And this may work to a degree for the earliest content we have, when websites were considerably more stable and static documents. It clearly does not work so well with the dynamic delivery of content characteristic of websites today.  So, does the date of archiving (or more correctly, the date of harvesting or collecting) that defines an archived instance in the web archive – typically included in the archive URL as well as associated metadata – really more accurately define the age of the site? In other words, is it only at the time that the ‘live’ website is made a static artefact in an archival context that we can actually establish a chronological dimension of the web? Perhaps so, and it is a theme I beat a little ‘paradiddle’ on in an opinion piece the ABC’s The Drum a few years back; and discussed in some more detail in an even earlier in a conference paper. In an earlier blog post I discussed a website that was lost and then found again some thirteen years later added to the PANDORA Archive. This was not a reconstruction or reconstitution of the original site, rather it was an archaeological discovery of the digital artefact that was collected at the time the original site was (a)live – a chance survival that is evidently possible for digital material but rare and certainly unpredictable. The prospect of further such discoveries of digital artefacts however remains an exciting one.

CybErg website

Home page of the CybErg website in the PANDORA Archive

When we look into the PANDORA Archive to find those early websites – and try to answer the challenge ‘what is your oldest website? – the discoveries are moot in their ontological revelation. For example, take the website of the ‘virtual proceedings’ of CybErg 1996 (the first international Cyberspace Conference on Ergonomics) which was actually selected for preservation in PANDORA in February 1997 (yes we keep records!). This was at a time when we were still developing our harvesting systems and three years before we implemented the first version of our web archiving workflow system (PANDAS). The first archived version of this site carries the date stamp 15 August 2000. Without any other evidence we may assume that was indeed the date it was harvested – the date of the artefact. However the website itself proffers evidence of being the site as it was in late 1996. The copyright date is 1996, the conference was held in September 1996, and there are no other indications on the site of a later ‘updated’ or ‘modified’ date. There is other evidence in the text, for example the exhortation to have orders for hard copy proceedings placed before October 31 1996. This would add evidence to suggest the version, collected in August 2000 was, for all intent and purpose, the website of late September to early October 1996. But how should we date the provenance of this website? By the evidence of the content or the provenance of the artefact – 1996 or 2000?

Mindgate website

Mindgate website collected for the PANDORA Archive in early 1997

The earliest ‘archiving date stamps’ on PANDORA content date from mid-1997, though content was certainly collected before that time. The reasons are complex and relate to the ‘proof-of-concept’ approach taken to the development of web archiving at the National Library of Australia. We started collecting before we had fully formed workflow systems and infrastructure. This is complex and is perhaps the subject for a future blog post. But to illustrate the point here, certainly one of the earliest sites collected for PANDORA was the short-lived science fiction site Mindgate, which carries the archiving date of 25 June 1997. It was probably one of the sites collected earlier in 1997 or even late 1996 – we do keep records, but they are not always as good as we would hope! Again, the copyright dates are stated as 1995 and 1996 and there are no later update or modification dates to challenge the supposition that this site, as it survives in the archive, is as it was in 1996. Versions collected by the Internet Archive indicate that the site languished on the web until sometime between 18th and 25 October 2000 with no change to the site. The URL was returning a 404 Not Found message as late as May 2009. Incidentally, even if you are not a sci-fi aficionado, have a look at the homepage for the quaint – at that time state-of-the-art – animated gifs. The poor old skeleton’s jaw has been chomping away in the PANDORA Archive for sixteen years!

Australian Observer website

The May 1995 edition of The Australian Observer - the oldest webpage in the PANDORA Archive?

So, finally, what is the best claim we can make for the oldest Australian website, based on the evidence of the artefact? Probably this is The Australian Observer, the soi disant first Australian online news magazine ‘compiled by professional journalists specifically for an Internet audience’. It appears to have had a short life between 1995 and 1996. The first issue from May 1995 is in the PANDORA Archive with the archived date stamp of 31 May 1995. Astute readers will observe that this is in fact an earlier date than the establishment of the PANDORA Archive itself. How can this be? In fact there were web archivists at the State Library of New South Wales who in a rather Pirandellian way were in search of a web archive. This drama was resolved when the SLNSW began participating in PANDORA some years later we were able to add some content they had collected earlier – including some metadata about the collecting dates (yes the SLNSW kept records too!). Nevertheless this does also illustrate the need to be circumspect and cautious in regard to the provenance of archiving date stamps in isolation. Oh, and what were they reporting in the first Australian online news magazine back in May 1995? “Greenhouse gases: are they really killing the earth? …Aborigines and Australia’s hidden shame … The new cold war between East and West … Can newspapers survive on the net?” Speaks for itself, really. So, our archiving activities add a chronological dimension to the web by taking static snapshots in time. These digital artefacts are the building blocks for understanding that chronological dimension. But when it comes to the age and provenance of what we actually see on the live web or indeed in the web archive, the situation is rather more complex.