A question popped up in the blogosphere recently. “Where is our Atlas of Digital Damages?” asked Barbara Sierman of the National Library of the Netherlands.
She pointed out the amazement that would greet evidence of physical books, safely stored, with spontaneous and glaring changes in their content or appearance. “Panic would be huge if this would happen in our libraries and archives.” That statement is certainly correct. Nearly everyone expects libraries and archives to have the basic resources to keep physical documentation stable, intact and fixed.
Sierman bravely points out, however, that digital items are very much at risk of loss and corruption–even when libraries and archives manage the material. Digitization sometimes yields mistakes, storage systems fail, older files rendered in newer computing environments behave oddly.
While digital preservation practitioners are well aware of this risk, Sieman called for some visual evidence to prove the point. Such evidence would, she reasoned, help make the case for a robust preservation infrastructure as well as help drive discussion about acceptable degrees of loss and the significant characteristics for digital objects.
“Because there lies a real risk for the digital collections, but making it visible with examples. it will be more convincing than all the conference papers that we have written about the digital preservation challenge.”
Her argument spurred action, and The Atlas of Digital Damages now is up and running on Flickr. This is a crowdsourced effort, and anyone can upload pictorial evidence of bits gone bad. There are currently a few dozen images available, but it is easy to imagine building quite a large collection of compelling images.
I have a few candidate images myself. Up until now, I kept them out of something like a perverse artistic appreciation, thinking perhaps they conveyed some fanciful insight into what machines see (or don’t see). Computers may not lie, but they surely can get confused.
The picture to the left, for example, suffered corruption during transfer to Flickr. A special irony is that the picture was taken of a recent exhibit of old computer hardware and media to demonstrate the tentative status of digital information. An unintended outcome, to be sure, but indicative (and evocative) evidence of maintaining fidelity for digital objects over time.
Do you have any graphical evidence of digital damage? If so, please consider sharing it so that we can help people understand what is at risk for our digital heritage.
My answer to file corruption issue is use file formats that don’t degrade(raw, nef, tiff, etc) and make multiple copies. I put my photographs on two computers, an online storage area, and a backup hard drive. The likelihood of all copies becoming corrupt would minimal.
Donald: Thanks for your comment. What you recommend makes lots of sense for personal collections. The problem gets bigger, however, for institutions that have to manage thousands (millions? many millions?) of digital files, including objects that are valuable but that don’t conform to ideal standards (older files in the papers of a notable person, for example). Quality control and fixity checks can reduce risks, but we’re still trying to figure out how best to do that sort of thing at the massive scale that is required for institutional digital collections–which will only be getting bigger!
Bill’s comment reminds me of a project I began down in Wilmington, NC. The New Hanover County Public Library’s earliest digital collections–the Louis T. Moore Collection and Robert M. Fales Collection, which date back to the early-to-mid 90s, were stored on an older system and the IT specialist warned us that the server was dying a slow, painful death, and we were in danger of losing the data stored in the collections.
There were two questions/concerns:
1) What to do with regards to the (now-dated) digital images.
2) What to do about the metadata.
There was risk of losing both (well, not so much a risk, but a question of when), and the solutions were pretty simple:
1) For the pics, rescan the originals (photographs or slides) at a higher resolution and in a better format. (we went with TIFFs, I think.)
2) Copy and paste the metadata into new files on the new DAMs (CONTENTdm).
The latter was easier to do than the former, because it involves a few clicks of the mouse and a little editorial work. The former was a bit more time-consuming, and I left the library before work was completed.
…I was hoping to locate the original sites–which I can only assume has since been shut down (a side-by-side comparison between the layouts and accessibility of the versions would have been nice), but the new sites were created in CONTENTdm.
One of the interesting issues we took into consideration involved the metadata itself. I inquired about using this opportunity to update the metadata to Dublin Core (and enter some new information in the process). My supervisor brought up an interesting and valid point: the metadata was, for the most part, written by Robert Fales, and Louis T. Moore. So it wasn’t cut-and-dry, because there was the issue of digital provenance–retaining the integrity of the information associated with the respective collections.