by Derek Morrison, 4 September 2008
Following on from yesterday’s posting Third Life – Auricle: the next generation (Auricle, 3 September 2008) here’s a few more reflections on blogs as information object repositories. I believe the issues addressed yesterday and in today’s posting are relevant to any context where unique and persistent object identities matter, not just blogs.
As suggested yesterday, I was fairly happy with the concept and benefits of “discrete addressability” afforded by each blog posting automatically being allocated a unique identifier. The problem being of course that the discrete addressability was applicable only in the context of both the host domain and the specific environment being used to host the quasi-information objects, i.e. in this case blog postings. A change of host domain or a change of host environment would break the discrete addressability. In my case a change of domain from bath.ac.uk to auricle.org and a change of underlying blog engine from pMachine to WordPress.
Before considering a specific example to illustrate the point we should remember that the process of migrating data from the pMachine blog successfully imported the content of the postings, but at the cost of WordPress applying its own object id system. As a result, any existing internal references between imported postings would be broken. pMachine also used an eccentric tagging system for URLs and so, consequently, some imported content contained spurious markup. Because Auricle has extensive content, fixing such broken links and spurious markup manually would become a very time-consuming process.
Turning now to the example.
In my new WordPress implementation of Auricle I have a posting titled The online ‘filling station’ model of e-learning revisited. This posting has a WordPress object id of “314”. However, “314” contained a hyperlink to the now obsolete pMachine version of Auricle hosted at Bath with an object id of “347_0_4_0_C”. So despite a discrete address the house had been demolished when we called 🙂 I have since manually associated this link with the correct internal WordPress posting.
But …
The pMachine version of Auricle has been archived at Auricle.org and this present both an opportunity and a dilemma. Only the domain part of the pMachine discrete address is inaccessible. The pMachine object ids are still preserved in the archive. Consequently, WordPress hyperlinks could be edited manually (or automatically at database level) and so be made to connect to pMachine objects now hosted in the new domain.
But …
Users would end up jumping between two blogs which apparently contain some of the same content, i.e the WordPress blog would contain the old postings plus whatever had been added since. Also the pMachine archived version of Auricle itself contains internal references which would need updating to reflect the new domain. In this situation it would have been better not to have migrated the pMachine content to WordPress at all but simply updated its domain information and referred users to the pMachine blog for, say, pre 2006 content.
But …
I don’t want a fragmented information base or to have to switch between environments and I suspect readers wouldn’t want that either. Also, the purpose of upwards migration was to preserve what I had written and keep in relevant to what I was writing today. Nevertheless, it will probably be useful to update the pMachine internal references to reflect the new domain and so improve the integrity of the archive copy. The pMachine variant, however, now also contains many dead external links but that will not be fixed in the archived version.
But …
This experience causes me to question the concept of discrete addressability when it is in reality so dependent upon a specific domain, or a specific technical environment. But what are the alternatives? There is always going to be the need for some sort of environment but that can be at the micro level of the blog or the more macro level of the network.
I’m now moving out of my comfort zone but since this is supposed to be a reflection opportunity so here goes
My next thought was to consider how easy it would be to find a specific example Auricle posting using Google. So I searched for a posting including “filling station model of e-learning”. Reassuringly, Google returned the results at the top of its first page … but … these were pointing to the discrete address of the house that had burned down, not the new house. There was better luck with “filling station model of e-learning + auricle.org”. Incidentally the Google discrete address was a really user friendly “http://www.google.co.uk/search?hl=en&q=filling+station+model+of+e-learning+auricle.org&btnG=Search&meta=” so perhaps we need the discrete addressability of a bit.ly or TinyURL? 🙂 The alternative Google blogsearch for “filling station model of e-learning” gave similar results to the plain Google search.
My next thought was automatic redirection by the old domain to the new but that would require the permanent co-operation of a domain for whom such co-operation may simply be construed as representing extra work, setting a precedent for all former staff, and potentially conflicting with institution IT policies.
I’ll finish this with the thought that what is required is something similar to the hardware address of a network card, i.e. a MAC Address equivalent. Theoretically, each hardware network device (ethernet and wireless) has an absolutely unique address and so if I move one such device from one machine in, say, the UK and put in another in, say, the US, other parameters may change but the MAC address remains constant. Anyone concerned with configuring home or larger networks will be familiar with this concept since the MAC address is required when configuring access to a wireless network for instance. If we had the equivalent of such a MAC address for information objects (or perhaps their hosting environment?) then at that point redirection services could come to the fore. After all that is how “URL-shortener” services such as bit.ly, or TinyUrl are able to render unwieldy urls to something more manageable. It is also how the mighty Domain Name System which acts as the “phone book” of the internet functions. So in my thought experiment when an object is moved it would automatically notify its redirector which would then update the references. A pipe dream maybe? At this point millions will rush to provide me with examples of where it is already happening 🙂
Postscript – 5 September 2008
- The obsolete pMachine blog internal references have been updated to point to the Auricle.org domain and not bath.ac.uk. Consequently, hyperlinks to other postings in the pMachine archive copy should now work. The external hyperlinks will not be updated in this earlier iteration of Auricle.
- The internal references in postings imported from the pMachine blog to the new WordPress blog now link to the archive copy of the pMachine blog. There was so many such links that I thought it better to link to something useful rather than just hyperspace. That means, however, that Auricle readers following older links will get to enjoy the benefit of two different blog styles. Over the next weeks I will start updating the internal links in the imported postings so that they point to the equivalent postings in WordPress.
