Shortly after my last post, a thread  surfaced on the Archives Listserv  asking the best way to crawl and record the top few layers of a website. This led to many posts suggesting all sorts of software geared toward this purpose. This post shares some of my thinking on the subject.
Adobe Acrobat can capture a website and convert it into a PDF. As pointed out in the thread above, that would loose the original source HTML – yet there are more issues than that alone. It would also loose any interaction other than links to other pages. It is not clear to me what would happen to a video or flash interface on a site being ‘captured’ by Acrobat. Quoting a lesson for Acrobat7 titled Working with the Web  : “Acrobat can download HTML pages, JPEG, PNG, SWF, and GIF graphics (including the last frame of animated GIFs), text files, image maps and form fields. HTML pages can include tables, linkes, frames, background colors, text colors, and forms. Cascading Stylesheets are supported. HTML links are turned into Web links, and HTML forms are turned into PDF forms.”
I looked at a few website HTML capture programs such as Heritrix , Teleport Pro , HTTrack Web  and the related ProxyTrack . I hope to take the time to compare each of these options and discover what it does when confronted with something more complicated than HTML, images or cascading style sheets. It also got me thinking about HTML and versions of browsers. It think it safe to say that most people who browse the internet with any regularity have had the experience of viewing a page that just didn’t look right. Not looking right might be anything from strange alignment or odd fonts all the way to a page that is completely illegible. If you are a bit of a geek (like me) you might have gotten clever and tried another browser to see if it looked any better. Sometimes it does – sometimes it doesn’t. Some sites make you install something special (flash or some other type of plugin or local program).
Where does this leave us when archiving websites? A website is much more than just it’s text. If the text were all we worried about I am sure you could crawl and record (or screen scrape) just the text and links and call it a day being fairly confident that text stored as a plain ASCII file (with some special notation for links) would continue to be readable even if browsers disappeared from the world. While keeping the words is useful, it also looses a lot of the intended meaning. Have you read full text journal articles online that don’t have the images? I have – and I hate it. I am a very visually oriented person. It doesn’t help me to know there WAS a diagram after the 3rd paragraph if I can’t actually see it. Keeping all the information on a webpage is clearly important. The full range of content (all the audio, video, images and text on a page) is important to viewing the information in its original context.
Archivists who work with non-print media records that require equipment for access are already in the practice of saving old machines hoping to ensure access to their film, video and audio records. I know there are recommendations for retaining older computers and software to ensure access to data ‘trapped’ in ‘dead’ programs (I will define a dead program here as one which is no longer sold, supported or upgraded – often one that is only guaranteed to run on a dead operating system). My fear is for the websites that ran beautifully on specific old browsers. Are we keeping copies of old browsers? Will the old browsers even run on newer operating systems? The internet and its content is constantly changing – even just keeping the HTML may not be enough. What about those plugins – what about the streaming video or audio. Do the crawlers pull and store that data as well?
One of the most interesting things about reading old newspapers can be the ads. What was being advertised at the time? How much was the sale price for laundry detergent in 1948? With the internet customizing itself to individuals or simply generating random ads how would that sort of snapshot of products and prices be captured? I wonder if there is a place for advertising statistics as archival records. What google ads were most popular on a specific day? Google already has interesting graphs  to show the correspondence between specific keyword searches and news stories that google perceives as related to the event. The Internet Archive  (IA) could be another interesting source for statistical analysis of advertising for those sites that permit crawling.
What about customization? Only I (or someone looking over my shoulder) can see my MyYahoo page. And it changes each time I view it. It is a conglomeration of the latest travel discounts, my favorite comics, what is on my favorite TV and cable channels tonight, the headlines of the newspapers/blogs I follow and a snapshot of my stock portfolio. Take even a corporate portal inside an intranet. Often a slightly less moving target – but still customizable to the individual. Is there a practical way to archive these customized pages – even if only for a specific user of interest? Would it be worthwhile to be archiving the personalized portal pages of an ‘important’ or ‘interesting’ person on a daily basis – such that their ‘view’ of the world via a customized portal could be examined by researchers later?
A wealth of information can be found on the website for the Joint Workshop on Future-proofing Institutional Websites  from January 2006. The one thing most of these presentations agree upon is that ‘future-proofing’ is something that institutions should think about at the time of website design and creation. Standards for creating future-proof websites  directs website creators to use and validate against open standards. Preservation Strategies for institutional website content  shows insight into NARA ‘s approach for archiving US government sites, the results of which can be viewed at http://www.webharvest.gov/ . A summary of the issues they found can be read in the tidy 11 page web harvesting survey .
I definitely have more work ahead of me to read through all the information available from the International Internet Preservation Consortium  and the National Library of Australia’s Preserving Access to Digital Information (PADI) . More posts on this topic as I have time to read through their rich resources.
All around, a lot to think about. Interesting challenges for researchers in the future. The choices archivists face today often will depend on the type of site they are archiving. Best practices are evolving both for ‘future-proofing’ sites and for harvesting sites for archiving. Unfortunately, not everyone building a website that may be worth archiving is particularly concerned with validating their sites against open standards. Institutions that KNOW that they want to archive their sites are definitely a step ahead. They can make choices in their design and development to ensure success in archiving at a later date. It is the wild west fringe of the internet that are likely to present the greatest challenge for archivists and researchers.
- Leveraging Google Reader’s Page Change Tracking for Web Page Preservation 
- Celebrating NASA’s 50th with NASA Images 
- SEO Evaluation of an Archival Website: Looking at UMBC’s Digital Collections 
- SAA 2006: Research Library Group Roundtable – Internet Archiving