Standards

An examination of the discussions related to standards suggests that they may be considered a mixed blessing. Particularly from the perspective of librarians and some IT professionals, they are emphatically good and necessary for interoperability and durability, and it is imperative that more tools and projects adopt them. However, certain scholars recoiled at the idea of standards being "imposed" on their projects, out of a concern that standards may prevent them from doing projects the way they'd like.

The vocal advocates of widespread adoption of standards pointed out a number of advantages to that approach, primarily centered around the issues of durability and interoperability. Commercial services may disappear, but open, standards-based services would enable the extraction and preservation of data by institutions after it is no longer viable to maintain it for profit. The use of standards-based repository archives for data storage, regardless of the environment that the data is consumed in (e.g. the data underlying a dictionary smartphone app, or a web interface for searching scholarly articles) was seen as one way to preserve the work that went into processing the data even after the technology behind the user environment has become obsolete. As a point of contrast, participants cited early digital humanities projects in the 1990's that produced CD-ROM versions of texts, only to have the scholarly CD-ROM become obsolete in a few years, leaving the data in a format that could not be reused. It was noted that the standards underlying the repository archives also evolves, but those standards change more slowly than trends in user interfaces.

Audiovisual materials pose a particular challenge for preservation; adherence to rapidly evolving standards requires frequent format conversions. The result is far from ideal when the material was digitized in accordance with best practices from a decade ago, with a much lower quality threshold than one expects today. Working with digital surrogates now considered to be low-quality is unavoidable, as fragile materials cannot be re-digitized every time standards improve.

Interoperability was cited as another serious challenge that could be solved by standards. Tools and data developed using the same standards should, theoretically, interoperate without a major effort to integrate them. This would allow the current digital humanities tool-building model of "let a thousand flowers bloom" to continue, while reducing the extent to which each of these tools is siloed. Standards also play a key role in an opposing approach to interoperability, the "services model", where time and effort is invested in knitting together selected tools into a single environment where the majority of scholars can do at least the majority of what they want, assuming the selection process for the included tools accurately reflects scholars' needs.

Beyond durability and interoperability, it was noted that standards could also facilitate scholars' finding and using new tools and data sets, since valuable resources sit unused if they are in the wrong format. Standards are a key piece of any attempt at federated search-- for results to be consistent across multiple databases, their metadata must be compatible and ideally adhere to the same best practices.

It was noted that not all standards should be considered equally effective at enabling interoperability. TEI was specifically called out as a standard that documents could adhere to and still fail to interoperate. John Unsworth described some examples of the diversity in TEI markup as the result of "needless and heedless" divergence from accepted ways of encoding texts, where the scholar would most likely mark up their text in accordance with best practices, if they only knew what those were. Others cited examples, such as a the development of the Encoded Archival Description (EAD) standard, where users had assessed the standard and did not feel that there was a pre-existing way to encode what they wanted, thereby necessitating divergence.

While there was near-universal agreement that standards in general were good, some participants, particularly scholars, were uncomfortable with the idea of applying standards developed by others to their own project. They were concerned that an externally developed standard may not be adequate to capture the unique nuances of the texts they work on. A perceived excess of normalization was seen as "tyranny", and there was concern that if a cyberinfrastructure project identified itself with certain standards in a public way, there would be a risk of alienating groups within the digital humanities community that dislike those standards. The concern that "standards" would be perceived negatively by scholars ran deep-- there was a discussion about whether the Project Bamboo working group titled "Standards and Specifications" should include an adjective like "recommended" to "soften the impact" of the word "standards".

Some participants concluded that people will ultimately make the decisions they want, and trying to advocate for standards-based development from the top by making it a requirement is likely to breed resentment. Instead, some suggested, a more effective path towards standards adoption is advocating for standards through telling potential users about the benefits of standards, and making training (e.g. TEI markup classes) widely available, to help scholars users make the most informed choice.

Bamboo tags: 

Bamboo see also: 

Add new comment