Comments on LibraryTechNZ: Virtually Ideal

No worries, I enjoyed the conversation immensely. ...

2007-10-24T15:39:00.000+13:00

No worries, I enjoyed the conversation immensely. You might be interested to find out more about the project undertaken by the Nationaal Archief of the Netherlands. Their model uses a universal virtual machine and a modular emulator to recreate the host environment. I've just put up a new post on that on my blog...it be good to hear your thoughts on that :)

Heh - I guess that was a long-winded way of saying...

2007-10-18T11:23:00.000+13:00

Heh - I guess that was a long-winded way of saying that I don't disagree with you :-)

Thanks for your comment. The standards drive is pr...

2007-10-18T11:19:00.000+13:00

Thanks for your comment. The standards drive is predominantly coming from either single private vendors, or clusters of two or three technology partners looking to leverage each other's development, as is the case with VMWare and the Pacifica extensions. I'm not nay-saying the reality of this, but I am trying to point out that this standards push is not something yet occuring within the ISO space or taking place under the terms and spirit of scientific inquiry, meaning that a significant percentage of the virtualisation models now in use will end up orphaned as consolidation occurs, resulting in the risk of data loss.
I am sure that to give an underlying robustness to digital records, the great breakthru will be when the presentation layer is made arbitrary. My feeling is that for this to occur, the digital object needs to be as self-descriptive as possible, which to me suggests that it would need to contain not just the essential data (ie the text of a novel or the PCM data of a sound) but also the immediate layers below - ie a playback application, and dependencies, ideally in source code form as well as compiled form, and all relevant assembly documentation. If an object-contemporary compiler were also included it would mean that the future archivist could have the option of emulating the original operating environment to make accessible the record, so including it would only add to the chance of the record surviving.
I think that adding these components to digital objects plays well with the fact that digital storage space is one of the resources we should consider almost inexhaustible bar technological/societal breakdown, so adding extra information to the digital record carries very minimal overheads.
Creating the agnosticism I refer to would also encompass a large workflow dependency - there's a lot that can be done on the proceedure side of a project to create ease of portability - simple, obvious things like having outputs referenced against zero-points, so that "default state" in "random app" is the correct relative setting, for example, or avoiding having offsets represented in metadata when it can be represented implicitly in the object.
The key reason resources like libraries and archives exist is that they are able to provide a resource far beyond the scope of an individual to create, and if we are unable to do something either difficult or impossible for an individual to do, then we have little relevance to society. Therefore I would state that attempting to defy the underlying characteristics of a digital object is something we have to pursue very strongly to deliver information of value to the greater world.

While I agree that virtualization isn't the magic ...

2007-10-17T18:28:00.000+13:00

While I agree that virtualization isn't the magic bullet to digital preservation, a few assumptions you put forth aren't exactly true today. Virtualization standards are being developed right now, and there are virtualization tools that allow you to run applications without the operating system in a virtual infrastructure. The fact that we create digital objects in specific environments to provide a certain user experience requires that we preserve that environment. Digital objects on their own are meaningless to users without the presentation layer. Also, I'm not sure how we can create hardware and software agnostic digital objects, because that would defy the underlying characteristics of a digital object.